Hydrological extremes and flood-related hazards across Central Asia: a systematic review and evidence-based synthesis of drivers, regional patterns, and research gaps

Climate change is causing extreme hydrological events in Central Asia (CA) and influencing the hydrological characteristics of river basins, flood risks, and water resource management. The goal of this systematic review is to synthesize the current knowledge on hydrological extremes and flood-related hazards in CA under changing hydro-climatic conditions, including an understanding of the dominant hydro-climatic processes, regional variability, methodological approaches, quality of the evidence, and implications for climate adaptation in CA. A PRISMA-guided systematic review and structured evidence appraisal framework were used to identify studies for this review from January 2021 to December 2025, published in peer-reviewed journals. A total of approximately 600 studies were found, and 42 studies were selected for inclusion in the final synthesis. Rather than a single factor causing flood-related hazards throughout CA, the review shows that hydro-climatic processes interact and contribute to these hazards. Extreme precipitation, alterations in the timing and amount of snowmelt, transitions between snow and rain, rain-on-snow processes, seasonal redistribution of runoff, and upstream–downstream human-climatic interactions are the most prominent drivers. The evidence also suggests significant regional variations in hydrological response, depending on the physiographic setting, hydro-climatic conditions, and availability of evidence. Despite recent advances in our understanding of climate-induced hydrological change, there are significant knowledge gaps in integrated multi-driver evaluations, mountain observations, transboundary synthesis, and the process of moving from science to decision-ready flood risk information. This review not only summarizes the current state of the science but also offers a structured evidence-based approach to assessing the scientific strength and regional representativeness of existing literature, as well as directions for future research. These findings provide comprehensive evidence to guide future research, flood risk assessment, and climate adaptation in CA, especially for national water authorities, river basin managers, infrastructure planners, and policymakers working on disaster risk reduction and climate adaptation.

Climate changes in temperature and precipitation in arid, semi-arid, and mountainous regions can strongly affect hydro-climatic extremes, including runoff formation, seasonal water availability, and flood risk (Fang et al., 2022; Boulange et al., 2021). CA is particularly sensitive to such changes because of its sharply continental climate, limited water resources, and strong dependence of river runoff on seasonal snow and glacier melt (Yapiyev et al., 2025). Because runoff is strongly seasonal, changes in snow accumulation, melt timing, and precipitation intensity can directly influence water regimes and flood risk conditions. According to current regional estimates, CA has already warmed by approximately 1.2 °C over the past 70 years, with snow depth decreasing by approximately 20% (Fallah et al., 2024). By the end of the 21st century, under the SSP5–8.5 scenario, warming could reach approximately 6.1 °C, indicating substantial potential changes in snow storage, runoff formation, river flow seasonality, and flood-generating conditions in the region’s major river systems (Fallah et al., 2024). Recent attribution-based analyses further indicate that anthropogenic forcing has increased the likelihood and intensity of extreme temperature and precipitation events across CA, particularly in mountain regions already vulnerable to landslides, floods, and snow and glacier-related hydrological changes (Fallah et al., 2023, 2024).

Flood hazards in CA are not controlled by a single mechanism. They arise from interacting processes, including heavy precipitation, snowmelt timing, glacier and alpine runoff, seasonal redistribution of river flow, and, in some settings, anthropogenic pressure on water systems (Nurbatsina et al., 2025b). Their relative importance varies across physiographic settings: Lowland and steppe river systems are more strongly associated with snowmelt-dominated spring hydrology, whereas mountain regions of the wider Tian Shan involve more complex interactions among rainfall, snowmelt, glacier processes, and high-elevation runoff generation (Shahgedanova et al., 2024). Recent studies also suggest that warming-driven shifts in snow accumulation, earlier seasonal snow peaks, increasing rainfall contributions, and intensifying rain-on-snow (ROS) processes may alter the timing and mechanisms of hydrological extremes in mountain-fed systems (Sadyrov et al., 2025; Yang et al., 2025). In the Tianshan–Pamir system, ROS frequency and intensity are projected to increase by about 49–67% by mid-century, indicating that cold-season precipitation phase shifts may become increasingly important for flood generation under warming (Yang et al., 2025).

The complexity of these interacting processes means that understanding flood hazards must not only incorporate the dominant hydrological processes but also use a framework for flood attribution to assess the relative contribution of the various flood-generating drivers. These frameworks take into account the interplay between meteorological forcing, snow and glacier processes, antecedent catchment conditions, basin characteristics, and human regulation to understand flood generation and evolution (Liu et al., 2022). Meanwhile, the development of new models for forecasting floods and inundations has emphasized the need to explicitly represent uncertainty in addition to deterministic forecasts. Uncertainty-aware flood-inundation mapping and Bayesian forecasting can be used to provide both flood occurrence and associated uncertainty/confidence, which can support a more robust flood risk assessment, emergency response, and adaptation planning under changing hydro-climatic conditions (Al-Wardy et al., 2026; Garshasbi et al., 2025).

Precipitation extremes are evident in CA and are expected to increase further with continued climate warming. During the 1936–2005 observational period, PRCPTOT (annual total precipitation from wet days) increased by 9.78 mm decade −1, and Rx1day (annual maximum one-day precipitation) increased by 0.45 mm decade −1, as indicated by observational analyses by Yao et al. (2021). The same study noted a positive trend in R95pTOT (precipitation from very wet days), suggesting a rise in the proportion of extreme rainfall to annual precipitation. Moreover, future projections of precipitation extremes under both RCP4.5 and RCP8.5 scenarios are consistently higher in the Central Asian region, reflecting the increased risk of extreme precipitation and flooding events in the predominantly arid regions of CA (Yao et al., 2021; Fallah et al., 2024). However, there is uncertainty in interpreting such changes in precipitation data. These uncertainties arise from the poor distribution and irregularity of meteorological stations in CA, especially in mountainous areas, where in situ observations are limited to short time series (Wei et al., 2023). Additionally, satellite- and gridded precipitation products exhibit significant differences over complex terrain due to biases and difficulties in capturing orographic precipitation, which is influenced by elevation (Duisebek et al., 2025). Furthermore, uncertainties related to the differences between reanalysis datasets and climate-model forcing add to the ambiguity of precipitation variability and hydrological processes in high-mountain regions (Wei et al., 2023; Sadyrov et al., 2025). In recent years, machine learning methods, alongside traditional statistical approaches, have also been applied to identify and explain hydro-climatic variability and long-term climate change in mountain catchments (Ali et al., 2025a,b).

Additionally, mountainous basins and transboundary basins present specific challenges, especially those associated with seasonality, topography, elevation, and the choice of precipitation dataset (Duisebek et al., 2025; Wei et al., 2023). To enhance the understanding of the contributions of snowmelt, glacier-fed runoff, rain-on-snow processes, and other flood-generating mechanisms operating in comparable environments, evidence from adjacent mountain regions, such as the Caucasus and the Tibetan Plateau, is used where applicable. This evidence is included solely to provide process-based context and interpretative support and is not used in the regional synthesis or as a substitute for evidence specific to CA.

These hydro-climatic changes are crucial for water security, agriculture, infrastructure, and long-term climate adaptation (Wang et al., 2024). CA hydrological systems are further complicated by transboundary water dependence, uneven seasonal water demand, and significant topographic contrasts between plains, foothills, and high mountains (Yapiyev et al., 2025). Recent analyses of Central Asian catchments indicate that observed changes in annual and seasonal discharge, including shifts in spring streamflow peaks, can, in several cases, be attributed to climate change (Didovets et al., 2024).

Recent watershed studies have reinforced the importance of morphometric features and basin parameters, along with combining physiographic data and hydro-climatic analysis, for environmental risk assessment (Ali et al., 2023) and for understanding hydrological responses and prioritization of watersheds. At the same time, flood-related risk is shaped not only by hydrological processes but also by exposure, settlement patterns, adaptive capacity, and water governance (Broomandi, Bagheri et al., 2025; Broomandi, Satyanaga et al., 2025; Fard et al., 2025; Sidle et al., 2025a). Accordingly, this review primarily focuses on the hydro-climatic and hydrological drivers of flood hazards while considering exposure, governance, and adaptation-related factors only where they directly inform flood risk interpretation, hazard assessment, or adaptation planning. However, the literature remains stronger in diagnosing hydro-climatic change than in translating it into operational risk thresholds, infrastructure criteria, or decision-ready adaptation frameworks.

Although relevant studies have increased substantially, the literature on hydrological extremes and flood-related hazards in CA remains fragmented across themes, methods, and spatial scales. Very few studies connect the various hydrologic aspects of extreme precipitation, snow and glacier processes, redistribution of runoff, impacts of flooding, debris flows, water stress, and adaptation within a coherent regional hydrologic framework that spans interconnected river basins, mountain-lowland systems, and transboundary catchments (Yao et al., 2021). A recent systematic review of the literature on climate change floods also highlighted the need for a more integrated approach to flood assessment and adaptation to climate change, with greater emphasis on hydrological processes, flood susceptibility, flood vulnerability, flood risk assessment, and adaptation strategies (Ali et al., 2025a). Uneven observational coverage, different modeling strategies, and variable hydro-climatic input data quality across mountain, lowland, and transboundary environments reinforce this fragmentation. As a result, it remains difficult to determine where evidence is robust, where uncertainty persists, and how current knowledge can support hydrologically grounded, basin-scale, and adaptation-oriented flood risk assessment in CA (Duisebek et al., 2025; Ceresa et al., 2025).

A systematic review dedicated to this topic is therefore needed to synthesize this rapidly expanding but varied evidence base. Recent flood attribution studies indicate that flood behavior is often controlled by multiple processes and is influenced by heavy rainfall, antecedent wetness, snowmelt, basin storage, catchment characteristics, and flow regulation (Liu et al., 2022). There is a conceptual benchmark to assess the extent of research on these flood-generating mechanisms in CA. Many of these interacting controls are not yet unified or synthesized across studies and are only loosely connected in the regional literature (Duisebek et al., 2025; Song et al., 2022).

Data availability is still limited throughout the region, partly due to reduced hydrometeorological station coverage, limited post-flood surveys, and uneven access to risk-related information, all of which constrain flood hazard assessment and decision-making across the region (Ceresa et al., 2025). The present review follows a predefined systematic review methodology based on the PRISMA framework, which includes structured literature searches, explicit eligibility criteria, and a transparent evidence appraisal process to systematically synthesize this fragmented evidence. The complete search strategy, databases, eligibility criteria, and study -selection procedure are described in Section 2. Thus, a systematic review is required not only to summarize existing studies but also to provide a logical structure for the existing knowledge, its scientific robustness, important limitations, and implications for adaptation measures.

The present review builds on the existing literature and integrates climate drivers, hydrological processes, flood-generating mechanisms, regional patterns, evidence quality, and adaptation relevance within a structured framework of evidence synthesis. It also systematically assesses the scientific robustness, methodological consistency, and regional representativeness of the available evidence and outlines priority research directions and gaps relevant to adaptation in CA.

The evidence synthesis and research priorities identified will help shape future research and evidence-based decision-making by national water authorities, river basin and reservoir managers, infrastructure and urban planners, and policymakers involved in climate adaptation and disaster risk reduction in CA.

The rest of the article is laid out as follows. The objectives of the review and research questions are outlined in the following subsection, with the systematic review methodology described in Section 2. In Section 3, evidence on hydrological extremes and flood-related hazards available for CA, including evidence robustness, uncertainties, and research gaps, is synthesized and analyzed. The paper ends with a summary of the main conclusions and recommendations for further studies.

Aim and review questions

The review focuses on hydrological extremes and flood-related hazards in the CA region and is particularly interested in studies related to understanding hydro-climatic processes, flood-generating mechanisms, and regional variability of responses to changing climatic conditions (Figure 1). Especially, the high-elevation areas where the effects of snowmelt, glacier dynamics, and runoff generation significantly shape downstream hydrological hazards are analyzed (Ceresa et al., 2025; Sadyrov et al., 2025). Common transboundary river basins, reliance on mountain-derived hydrologic resources, and the spatial and temporal limitations of observations across CA result in transboundary challenges and flood risk vulnerabilities at a regional scale.

This review aims to systematically synthesize the peer-reviewed literature published from January 2021 to December 2025 to provide an extensive overview of hydrological extremes and flood-related hazards in CA in a changing hydro-climatic environment. The review covers the current state of knowledge regarding the dominant drivers of the hydroclimate, flood-generating mechanisms, regional patterns, methodological approaches, evidence quality, and implications for flood risk assessment and climate change adaptation. Findings are interpreted based on a structured framework for evidence appraisal, which distinguishes conclusions supported by strong and consistent evidence from those that are more tentatively interpreted based on the quality, consistency, geographic representativeness, and methodological strength of the available evidence. A comprehensive description of the literature search, databases, eligibility criteria, study selection process, and evidence appraisal process is provided in Section 2.

To achieve these objectives, the review follows an integrated approach based on systematic evidence synthesis, which consolidates evidence related to climate drivers and hydro-climatic processes, flood-generating mechanisms, patterns, evidence quality, and adaptation-relevant outcomes across CA. It constitutes a thematic synthesis and systematic assessment of geographic relevance, hazard relevance, data quality, methodological rigor, future relevance, and decision relevance. These dimensions are evaluated against predetermined criteria and an ordinal scale for evidence appraisal, as outlined in the Methodology (Section 2), to ensure a transparent, consistent, and reproducible scientific evaluation of the robustness, consistency, spatial representativeness, and practical relevance of the available evidence.

Accordingly, the following research questions are addressed in this review: Under changing climatic conditions, what are the most common studies conducted on hydrological extremes and hazards linked with flooding in CA?

What are the major hydro-climatic and, if applicable, anthropogenic processes that contribute to flood hazards in the region?

What is the current evidence base for this topic—how robust, consistent, and representative is it, considering differences in countries, datasets, analytical approaches, and future climate scenarios?

What are the key gaps and future research needs in CA to better understand, assess, and adapt to hydrological extremes and flood-related hazards in the context of climate change?

The objective of this study was to conduct a systematic review to synthesize recently available evidence on climate-induced hydrological extremes and flood-related hazards in CA. The review was carried out through the identification, screening, and critical evaluation of relevant literature published from 1 January 2021 to 31 December 2025. This time window was chosen to represent the most recent stage of research development in the region, including the growing adoption of new climate projections, regional assessments of water resources, and more recent analyses of flood-related hazards. Thematic synthesis was applied together with a structured evidence appraisal to evaluate the strength, consistency, and relevance of the included studies. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement (Sarkis-Onofre et al., 2021) was followed in reporting the review. PRISMA 2020 provided a framework for conducting the reporting of the study identification, screening, eligibility assessment, and final inclusion in a transparent manner. Methods for the study selection process are summarized in the PRISMA 2020 flow diagram (Figure 2), and the methods used for the review encompassed the following: preestablished search strategy, eligibility criteria, screening process, thematic synthesis, and structured evidence appraisal. The present review approach to CA is a region-specific, process-oriented evidence synthesis in contrast to previous studies that have focused on specific aspects of flood research, such as quantitative flood hazard assessment methods (Maranzoni et al., 2023), systems approaches to flood risk management (Awah et al., 2024), bibliometric and methodological trends in flood mitigation research (Rotimi et al., 2025), compound flooding in coastal environments (Green et al., 2024), or basin-scale flood risk modeling using remote sensing and machine learning methods (Ali et al., 2025a,c). The methodology does not focus on the assessment of flood modeling techniques, hazard assessment frameworks, or a single type of flood, but integrates a PRISMA-guided systematic review with structured qualitative evidence appraisal to synthesize hydro-climatic drivers, hydrological processes, regional patterns, evidence quality, and adaptation relevance from the available literature. This method can be used for the comparative evaluation of the strength of scientific evidence, spatial representativeness, methodological consistency, and usefulness for decision-making, as well as for identifying gaps in scientific knowledge and priorities for further research in the region.

This environmental and hydrological evidence synthesis did not meet the need for prospective registration in PROSPERO or the Open Science Framework (OSF), which are typically used for health-related systematic reviews. However, a predetermined review protocol was designed, encompassing the search strategy, eligibility criteria, and evidence appraisal structure to ensure methodological transparency, consistency, and reproducibility. This environmental and hydrological review had not been registered a priori. The present review was not subject to a discipline-specific prospective registration requirement, as PROSPERO mainly focuses on systematic reviews in health and related fields. Despite this, a predefined review protocol was established prior to the screening process and included the search strategy, eligibility criteria, screening process, thematic classification, and the structure for evidence appraisal to facilitate methodological transparency, consistency, and reproducibility. In addition to thematic classification, the included studies were examined using a structured quality appraisal framework based on six criteria: geographic relevance, hazard relevance, data quality, methodological rigor, future relevance, and decision relevance. This enabled a systematic evaluation of the evidence base in terms of scientific strength, regional relevance, and adaptation usefulness (Figure 3).

The same evidence appraisal logic was also used to support the semi-quantitative matrices, conceptual synthesis figures, and regional interpretation frameworks presented in the Results and Discussion section. These visual syntheses were derived from the reviewed literature and the supplementary evidence tables. Figure-specific input matrices, evidence classifications, temporal tags, process linkage logic, and explanatory notes are provided in the Supplementary material.

The main emphasis of the review was on CA, and studies had to directly contribute to the interpretation of hydrological extremes, hazards associated with snowmelt or precipitation, flood-related impacts, transboundary hydrological processes, or adaptation-relevant regional patterns in the area. Within this scope, the review integrated country-specific and transboundary basin and mountain system evidence relevant to hydro-climatic processes and flood-related hazards in the CA region. A few more general comparative and inland arid studies were included when they offered process-level or conceptual evidence directly applicable to the interpretation of CA hydrological extremes. Selected studies in adjacent or similar areas were used selectively when evidence of a particular process pathway within CA was lacking. These studies were not considered valid alternatives to CA-based studies and were treated as background information if they did not fall into the categories of formally reviewed studies. A geographical breakdown of the formal review sample is presented in Supplementary Table S1, and the characteristics of all 42 studies included are presented in Supplementary Table S2 (study-level characteristics).

A literature search was conducted in the Scopus and Google Scholar databases, covering articles published from 1 January 2021 to 31 December 2025. The final database search was completed on 31 December 2025, and all records published up to that date were screened. Articles published online or assigned to journal issues on or before 31 December 2025 were included, provided a publication date of 2025 had been assigned. The search targeted studies on climate-driven hydrological change, flood hazards, snowmelt processes, runoff change, and extreme precipitation in CA.

The search using various combinations of regional identifiers, country names, hydro-climatic process terms, names of major basins and mountain systems, and hazard terms. Core regional terms included “Central Asia,” “Kazakhstan,” “Kyrgyzstan,” “Tajikistan,” “Uzbekistan,” and “Turkmenistan.” Additional search terms included “Tien Shan”/“Tian Shan,” “Pamir,” “Amu Darya,” “Syr Darya,” “Ili River,” “Naryn River,” and “transboundary basin.” These were combined with hazard- and process-related terms such as “flood,” “flood hazard,” “flood risk,” “hydrological extremes,” “hydro-climatic extremes,” “extreme precipitation,” “heavy precipitation,” “snowmelt,” “rain-on-snow,” “runoff,” “runoff redistribution,” “streamflow,” “peak discharge,” “flood timing,” “flood seasonality,” “glacier melt,” “cryosphere,” “mountain hazards,” “debris flows,” “water stress,” “hydrological regime,” and “climate change.” An example Scopus query combined regional and process terms in the following format: (“Central Asia” OR Kazakhstan OR Kyrgyzstan OR Tajikistan OR Uzbekistan OR Turkmenistan OR “Tien Shan” OR “Tian Shan” OR Pamir OR “Amu Darya” OR “Syr Darya” OR “Ili River” OR “Naryn River”) AND (“flood” OR “flood hazard” OR “hydrological extremes” OR “extreme precipitation” OR snowmelt OR runoff OR streamflow OR “rain-on-snow” OR “glacier melt” OR cryosphere OR “debris flow”) AND (“climate change”). To ensure uniformity in screening, interpretation, and comparison of the studies, only English-language publications were searched. It is acknowledged that this may have led to language bias, as potentially relevant studies published in Russian and the various national languages of Central Asian countries were not included. However, it was determined that the best approach to guarantee methodological consistency, transparency, and comparability of the analyzed studies would be to include only those published in English-language peer-reviewed literature.

Retrieved records from both databases were merged, and duplicate entries were checked and removed before screening. Google Scholar results were screened in the order of relevance provided by the database. To maintain transparency and avoid uncontrolled expansion into low-relevance results, screening was limited to the first 600 records for each major search combination. This threshold was established prior to the screening process to help maintain consistency and manageability and to ensure that the records most relevant to the review goals were screened. At the same time, the literature search was conducted in Scopus, and the removal of duplicate records and eligibility screening were performed for all the retrieved records. It is important to note that the sequence of search results provided by Google Scholar can vary over time and from user to user, as can the ranking of those results. The search strategy, search terms, search period, and screening process are described in detail to enhance the reproducibility of the search, and the final search was performed on 31 December 2025. Duplicate records were eliminated, and then two successive screening operations were applied to the retrieved records. The initial screening for relevance to the review purpose and objectives excluded 430 records, leaving 170 records. A further screening of abstracts excluded another 107 records that did not meet the thematic, geographical, and methodological inclusion criteria. Thus, 63 full-text articles were evaluated for eligibility. Of those, 21 did not meet all the eligibility criteria after full-text review, and 42 were included in the final qualitative evidence synthesis (Figure 2).

The key study characteristics were systematically extracted and tabulated for each of the 42 studies in the formal sample for the systematic review, including geographic setting, basin or subregion, study period, study design and analytical approach, main data sources, main hydro-climatic processes or hazards, and historical or future climate scenario context. These study-level characteristics supported the geographic, thematic, methodological, and process-based synthesis presented in the Results and Discussion. A full summary of the included evidence is provided at the study level in Supplementary Tables S2–S6 include more detailed process-specific and quantitative evidence for selected studies.

Eligibility criteria and study selection

Studies were included if they met all of the following criteria. First, they focused on the impact of climate change on hydrological extremes, snowmelt, runoff, flood hazards, or extreme precipitation. Second, they provided empirical, statistical, modeling, observational, or remote sensing-based analyses that offered interpretable evidence relevant to the review objectives. Third, their geographical focus included at least one country within CA, a transboundary basin relevant to the region, or a broader Central Asian domain.

Publications that focused solely on engineering, such as the hydraulic design of flood control structures, structural design of dams, reservoir engineering, channel engineering, and dam safety assessments, without explicit consideration of hydro-climatic drivers, hydrological processes, or climate change impacts were not retained. Publications that lacked direct relevance to flooding, hydrological extremes, snowmelt, runoff-related processes, or closely related hazard drivers were also not retained. Publications that did not fall within the geographic scope of CA or were not clearly related to the hydro-climatic systems, transboundary basins, or mountain source environments of the region were excluded as well. Additionally, publications that lacked methodological clarity necessary for meaningful synthesis, including at least a basic description of the study domain, data inputs, and analytical methods, were not retained. Peer-reviewed journal articles formed the main evidence base of the review. Reports, review papers, and preprints were considered selectively and primarily used for contextual framing or supplementary interpretation when they provided relevant regional insight. However, such sources were not allowed to dominate the principal evidence synthesis.

Screening and evidence appraisal

The study selection process was carried out in several steps to ensure consistency and relevance during the review. Records were screened sequentially after duplicate removal based on a set of predefined eligibility criteria, first by title and abstract, and then by full-text assessment. The titles and abstracts were initially screened for thematic and geographic relevance, followed by full-text evaluations of potentially relevant studies. Studies were included in the final synthesis only if they were within the scope of the review and provided interpretable methodological, analytical, or process-level information. Records that appeared potentially relevant were then subjected to full-text assessment using predefined inclusion and exclusion criteria. Only studies clearly aligned with the review scope and that provided interpretable methodological, analytical, or process-level information were retained for final synthesis. The first author conducted the initial screening, and studies that were unclear were reviewed with the co-authors before a final decision was made regarding their inclusion. The inter-rater reliability statistic, such as Cohen’s kappa, was not calculated since the review was consensus-based rather than based on independent duplicate screening. When the eligibility to include or exclude a study was uncertain, the authors discussed the decision, considering the agreed eligibility criteria to achieve consistency and transparency in the screening process. Likewise, the first author completed the initial evidence appraisal and score assignment based on the predefined evidence appraisal framework and scoring criteria described below. If there was any doubt regarding the classification or score for a case, it was re-examined and discussed with the co-authors until an agreement was reached. No formal inter-rater reliability statistic was calculated for the scoring process, as it was not conducted independently in duplicate by two or more assessors.

The last studies included were evaluated using a structured system of evidence appraisal, which emphasized the scientific strength of the evidence, geographic applicability, and practical utility. First, each of the studies included in the synthesis was assessed in terms of six appraisal dimensions: geographic relevance, hazard relevance, data quality, methodological rigor, future relevance, and decision relevance. These dimensions were also used for the semi-quantitative evidence synthesis of the predefined geographic study groups and thematic study groups. Geographic relevance refers to how well the evidence reflects CA hydro-climatic systems and environments. Hazard relevance indicates the direct applicability of the evidence to hydrological extremes, flooding, snowmelt processes, changing runoff, rain-on-snow processes, and related hazards. Data quality is based on the robustness, consistency, validation, and reliability of observational, reanalysis, remote sensing, or modeling data. Methodological rigor is assessed based on the transparency and robustness of the analytical methodology applied, such as process-based analysis methods, hydrological modeling, attribution analysis, multi-model ensembles, and other advanced analytical techniques. The extent to which the evidence includes future climate projections, scenario analysis, or transferable insights for assessing future hydro-climatic risk indicates future relevance. Decision relevance refers to the applicability of the evidence for flood risk management, adaptation planning, monitoring, policy support, or water management decision-making.

Each study group received an ordinal evidence score of 1 (Limited), 2 (Moderate), or 3 (Strong) for the six dimensions of evidence as part of the semi-quantitative evidence synthesis. A Strong score was assigned where there was high direct relevance, robust or consistent supporting evidence, strong methodological support, or a clear connection to CA flood risk assessment and planning. In contrast, a Limited score signified weak relevance to the region, limited methodological depth, sparse supporting evidence, weak validation, or low applicability to flood risk assessment. Intermediate levels of evidence or partial fulfillment of the assessment criteria received Moderate scores. The score assigned was based on an interpretive synthesis of the literature reviewed and considered factors such as study frequency, spatial representativeness, hazard specificity, robustness of the datasets, methodological sophistication, integration of future climate, consistency of reported findings, and decision applicability. Details of scoring definitions and matrices, definitions of the six appraisal dimensions, ordinal evidence classes, and interpretive criteria are available in Supplementary Tables S7–S10.

The evidence appraisal framework is not a formal quality filter but is used comparatively. Quality scores of the studies were not used as an exclusion criterion, and no weighted aggregate quality score was computed. Rather, the results of the appraisal were presented as part of the narrative synthesis to support the understanding of the strength of evidence, identify patterns in the methodology and geography of the studies, highlight areas of agreement and disagreement among studies, and distinguish findings that are well supported from those that require a more cautious interpretation. The framework also served to highlight the relative strengths, methodological weaknesses, geographical scope, and evidence gaps in the reviewed literature. The findings were analyzed based on study frequency, consistency of reported findings, spatial representativeness, methodological robustness, hazard specificity, and decision relevance, rather than relying on only one of these criteria. Conclusions described as well supported are those reported in several studies, have a solid methodological basis, and are relevant to a wider region, while conclusions to be interpreted with caution are usually based on limited regional data, inconsistent findings, lower methodological support, or limited literature. The appraisal framework was therefore used to guide the comparison of literature, not to weight studies or serve as a selection mechanism. There was no formal statistical sensitivity analysis of the evidence scores. The scoring system was not intended as a statistical weighting system, a study exclusion system, a meta-analytic model, or a pooled-effect estimator, but rather as an evidence synthesis tool to provide a qualitative/semi-quantitative approach. The robustness of the synthesis was assessed by the consistency of findings reported in studies, study frequency, geographic representativeness, methodological robustness, hazard specificity, and decision relevance, rather than through statistical perturbation of an aggregate score. A further explanation of the scoring logic, evidence classifications, interpretive criteria, and reproducibility framework is included in the Supplementary material.

Geographical and temporal distribution of studies

The literature reviewed indicates that in recent years, climate-driven hydrological extremes and associated flood-related hazards in CA have received increasing research attention. A total of about 600 records were initially identified through the database search, and 42 studies were eligible and included for final synthesis after screening and eligibility assessment. The key characteristics of these 42 studies, including geographic setting, basin or subregion, study period, methodological approach, principal data sources, hydro-climatic focus, and climate scenario context, are systematically summarized in Supplementary Table S2. This growth reflects increasing attention to regional climate change, runoff-regime shifts, extreme precipitation, mountain hazards, and adaptation-relevant flood risk, although the evidence base remains uneven across topics and locations. Unless otherwise indicated, all descriptive statistics, evidence syntheses, and thematic summaries reported in this section are from the 42 studies identified in the CA systematic review. References to neighboring or similar areas are made clear in the text and are only used to provide process-based background information or to help interpret hydro-climatic mechanisms when CA-specific evidence is sparse.

The greatest number of studies was published in 2025, with 45.2% (19/42) of the literature published that year. This is due to the increased volume of research published in the last few years included in the review, rather than a bias in the search strategy, since all studies published up to the predetermined end date of the search, 31 December 2025, were eligible for inclusion. This concentration indicates that the current evidence base is strongly shaped by recent studies, particularly those involving regional flood risk assessment, water stress interpretation in CA, runoff redistribution, event-based flood analysis, and ROS intensification (Ceresa et al., 2025; Wu et al., 2025; Guo et al., 2025; Yang et al., 2025; Issakov et al., 2025). The geographic distribution of the reviewed studies is uneven. The largest single geographic category in the reviewed sample is region-wide or multi-country CA, comprising 12 of the 42 included studies, which focus on regional climate trends, extreme precipitation, hydrological stress, flood risk, attribution, and adaptation (Fallah et al., 2024; Ceresa et al., 2025; Wu et al., 2025; Yao et al., 2021; Guo et al., 2025; Issakov et al., 2025; Fallah et al., 2023; Wei et al., 2023). The second-largest category is Kazakhstan-specific evidence, with 11 studies primarily focused on basin hydrology, runoff redistribution, low-flow change, water stress, and selected flood impacts (Nurbatsina et al., 2025a, 2025b; Laiskhanov et al., 2025; Tursunova et al., 2025; Birimbayeva et al., 2025; Makhmudova et al., 2023; Issaldayeva et al., 2023; Tursunova et al., 2025b; Makhmudova et al., 2025; Yapiyev et al., 2025; Aitzhanova et al., 2025). Together, these two categories account for more than half of the reviewed sample, indicating that the evidence base is structurally strongest at the regional CA scale and in Kazakhstan-specific basin studies. The mutually exclusive geographic classification used for these descriptive statistics is provided in Supplementary Table S1, and the corresponding study-level geographic, temporal, methodological, and hydro-climatic characteristics are listed in Supplementary Table S2.

Similarly, the relative absence of several countries and subregions from the reviewed literature should not be interpreted as an indication of reduced flood hazard or hydro-climatic risk. Instead, it is more likely due to unequal research efforts, variations in the availability of hydrometeorological data, the lack of inclusion of locally published research in international databases, and the limitation of published research to English. These factors contribute to key geographical evidence gaps that are relevant for interpreting the geographical distribution of the literature and identifying priorities for future research.

In addition, transboundary basin studies and comparative mountain system investigations provide important process-level insights into snowmelt, glacier influence, runoff redistribution, extreme precipitation, and associated flood-generating mechanisms. This is evident in transboundary or shared-basin studies on the Ili, Naryn, and Panj-related systems (Didovets et al., 2024; Duisebek et al., 2025; Gulakhmadov et al., 2025; Sadyrov et al., 2025), as well as mountain system studies from the Tian Shan and Pamir that address moisture sources, flash floods, ROS processes, and debris-flow hazards (Guan et al., 2022; Ren et al., 2023; Shahgedanova et al., 2024; Yang et al., 2025; Zhang et al., 2025). Although fewer in number, these transboundary and mountain system studies are disproportionately important because they elucidate mechanisms that cannot be adequately captured through country-level lowland studies alone. A smaller group of global or broader comparative papers was used primarily to support conceptual or methodological interpretation, especially on flood risk, flood timing, seasonality of heavy precipitation, and rainfall–snowfall phase change (Tabari et al., 2021; Fang et al., 2022; Gründemann et al., 2023; World Meteorological Organization, 2024; Li Y.-P. et al., 2025). The inclusion of these broader comparative studies in the formal sample reflects their predefined process-level relevance to the CA synthesis rather than equivalent geographic relevance to CA-specific studies.

A notable feature of the reviewed literature is that many studies are organized around river basins, mountain systems, and transboundary hydrological domains rather than national boundaries alone, reflecting the physically interconnected nature of climate-driven water hazards in CA. The country-level evidence is also uneven. Beyond the region-wide CA literature, Kazakhstan contributes the largest number of stand-alone case studies, particularly on basin hydrology, hydrological regime shifts, low-flow change, and selected flood impacts (Issaldayeva et al., 2023; Makhmudova et al., 2023, 2025; Aitzhanova et al., 2025; Birimbayeva et al., 2025; Laiskhanov et al., 2025; Nurbatsina et al., 2025a, 2025b; Tursunova et al., 2025, 2025b; Yapiyev et al., 2025). By contrast, evidence from other individual Central Asian countries remains more limited and is more commonly represented through basin-scale, transboundary, or comparative studies than through broader national syntheses, as illustrated by the smaller number of country-specific studies from Kyrgyzstan, Tajikistan, and Uzbekistan (Gulahmadov et al., 2021; Uwamahoro et al., 2021; Kalashnikova et al., 2024; Zhuo et al., 2025). Turkmenistan is even less represented, so conclusions for that setting should be regarded as especially cautious and provisional.

Overall, the reviewed literature shows that the strongest evidence base is provided by region-wide Central Asian studies, Kazakhstan-based basin analyses, and transboundary or mountain system investigations (Didovets et al., 2024; Fallah et al., 2024; Ceresa et al., 2025; Duisebek et al., 2025; Nurbatsina et al., 2025a, 2025b; Sadyrov et al., 2025; Yang et al., 2025). Although the formal descriptive statistics and core synthesis presented in this section are based on the 42 studies retained in the reviewed sample, an additional 22 papers were consulted selectively during interpretation to address process-level gaps in the core evidence base. These supporting papers were not included in the formal review counts or descriptive analyses but were used to clarify mechanisms that were only partially resolved in the reviewed sample, particularly for snowmelt and glacier runoff, extreme precipitation processes, runoff seasonality, and flood dynamics in Xinjiang/Tarim and adjacent high-mountain environments (Fang et al., 2023; Wang et al., 2023; Li Z. et al., 2024; Yao et al., 2024; Li Y.-P. et al., 2025). These supporting studies improve process interpretation, but they do not eliminate the underlying geographic imbalance of the core CA evidence base.

Figure 4 and Supplementary Table S1 summarize the reviewed studies by geographic category. Of the 42 included studies, 12 were assigned to the region-wide or multi-country CA category, 11 were Kazakhstan-specific, 4 represented other single-country Central Asian settings, 4 were assigned to transboundary basin studies, 5 to comparative mountain system studies, 1 to the Tarim/arid inland subregional system category, and 5 to global or broader comparative support. Thus, 54.8% of the reviewed sample falls into either the region-wide CA or Kazakhstan-specific categories, whereas only 9.5% consists of other single-country Central Asian studies. This imbalance highlights the need to interpret several process domains, especially mountain hydrology and transboundary runoff change, through broader regional and shared-basin evidence rather than through evenly distributed national literatures. The individual studies underlying these categories, together with their basin or subregional setting, study period, analytical approach, main data sources, hydro-climatic focus, and climate scenario context, are reported in Supplementary Table S2.

Thematic and methodological profile of the literature

Figure 5 provides a statistical summary of the thematic concepts (Figure 5A) and methodological approaches (Figure 5B) represented in the reviewed literature. The study-level characteristics summarized in Supplementary Table S2 show that the thematic diversity of the evidence base is accompanied by substantial variation in geographic setting, temporal coverage, datasets, analytical approaches, hydro-climatic processes, and climate scenario use. The reviewed studies show a broad but uneven thematic distribution, indicating that research on hydrological extremes and flood-related hazards in CA has developed around a combination of direct hazard manifestations, hydro-climatic drivers, and changing runoff regimes, rather than solely fully integrated flood risk assessments. As shown in Figure 5A, which is based on the 39 reviewed studies that were coded for substantive thematic focus, the most frequently represented themes are flood hazards and impacts and runoff and low-flow changes, each accounting for 35.9% of the coded studies. Flood-related hazard and impact studies include work on transboundary flood risk, basin-scale flood modeling, flash floods, inundation, debris-flow/flood susceptibility, and flood consequences (Makhmudova et al., 2023; Shahgedanova et al., 2024; Wang et al., 2024; Yao et al., 2024; Aitzhanova et al., 2025; Ceresa et al., 2025; Laiskhanov et al., 2025; Nurbatsina et al., 2025a; Sidle et al., 2025a; Zhang et al., 2025). Studies on runoff and low-flow change are similarly prominent and focus on hydrological drought, streamflow attribution, runoff redistribution, low-flow sensitivity, and seasonal water availability (Gulahmadov et al., 2021; Issaldayeva et al., 2023; Didovets et al., 2024; Fallah et al., 2024; Birimbayeva et al., 2025; Gulakhmadov et al., 2025; Makhmudova et al., 2025; Tursunova et al., 2025; Wu et al., 2025; Yapiyev et al., 2025; Zhuo et al., 2025; Nurbatsina et al., 2025b). Extreme precipitation is the next most frequent theme, accounting for 23.1% of the coded studies, and includes work on changing precipitation extremes, moisture sources, event types, and anthropogenic influence on extreme rainfall (Yao et al., 2021; Guan et al., 2022; Song et al., 2022; Fallah et al., 2023; Gründemann et al., 2023; Wei et al., 2023; Guo et al., 2025). Snow- and glacier-related processes account for 17.9% of the coded studies and include work on snowmelt representation, glacier contributions, ROS processes, and shifting snow–rain partitioning (Uwamahoro et al., 2021; Li Z. et al., 2024; Li Y.-P. et al., 2025; Sadyrov et al., 2025; Yang et al., 2025). Water stress and water resources are the least represented primary theme, with 10.3% of coded studies, and are discussed primarily in studies on water insecurity, adaptation, and surface water stress (Issakov et al., 2025; Tursunova et al., 2025b). Note that the thematic frequencies in Figure 5A reflect the relative amount of research attention dedicated to each topic and not the size, severity, or spatial extent of the hazards associated with each topic. Thus, water stress-related risks should not be considered of lower priority because of the relatively limited number of studies focused on water stress. Rather, this highlights an important research gap, especially in light of projections of significant drought and water stress risks in CA in the future. Note that some studies focused on more than one thematic area; thus, the thematic categories are not mutually exclusive, and percentages do not sum up to 100%.

Overall, Figure 5A shows that the literature is strongest in diagnosing hydrological instability and flood-generating conditions but comparatively weaker in explicitly connecting those conditions to integrated risk, vulnerability, or decision support frameworks. The relatively limited representation of water stress studies reflects a gap in the current evidence base rather than a lower regional importance of water security challenges.

Taken together, the thematic profile indicates that direct hazards and broader hydro-climatic instability are well represented, whereas snow storage, glacier melt, ROS interactions, exposure, vulnerability, and operational decision-making remain less integrated.

Figure 5B provides a quantitative summary of the methodological approaches adopted across the reviewed studies and indicates a diverse but uneven evidence base. These methodological classifications are consistent with the study-level information summarized in Supplementary Table S2, which documents the analytical approaches and principal data sources for each of the 42 included studies. Using the full set of 42 reviewed studies, Figure 5B shows that the most frequently used methodological approach is observational trend or attribution analysis (54.8%), followed by climate scenario modeling based on CMIP/RCP/SSP frameworks (33.3%). This is followed by hydrological modeling (16.7%), event-based impact analysis, and review-type approaches (16.7%), remote sensing/GIS/DEM/UAV-based methods (14.3%), and dataset evaluation or bias correction studies (7.1%). Because several studies employed more than one analytical approach, the methodological categories are not mutually exclusive, and their percentages do not sum to 100%. Methodologically, the literature is strongest in diagnosing past changes and projecting future conditions but is weaker in translating hydro-climatic changes into operational flood risk tools or management thresholds. Observational and attribution-based studies are prominent in research on runoff trends, hydrological drought, precipitation extremes, seasonality changes, and climate-driven discharge shifts (Gulahmadov et al., 2021; Fang et al., 2022; Fallah et al., 2023, 2024; Issaldayeva et al., 2023; Wei et al., 2023; Didovets et al., 2024; Wu et al., 2025). Climate scenario modeling is strongly represented in studies projecting future runoff regimes, flood risk, ROS changes, and water stress under CMIP5/CMIP6, RCP, or SSP scenarios (Ceresa et al., 2025; Gulakhmadov et al., 2025; Makhmudova et al., 2025; Wu et al., 2025; Yang et al., 2025; Nurbatsina et al., 2025a, 2025b; Tursunova et al., 2025b). Hydrological modeling remains important in basin-scale flood and runoff studies, especially where snowmelt, glacier influence, or data scarcity affect simulation quality (Uwamahoro et al., 2021; Makhmudova et al., 2023, 2025; Ren et al., 2023; Kalashnikova et al., 2024; Zhang et al., 2025; Nurbatsina, et al., 2025a).

Event-based impact studies and review-type approaches mainly contribute to the interpretation of recent extreme floods, flood consequences, and adaptation-oriented understanding (Tabari et al., 2021; World Meteorological Organization, 2024; Yao et al., 2024; Ceresa et al., 2025; Issakov et al., 2025; Laiskhanov et al., 2025; Sidle et al., 2025a). Remote sensing, GIS, DEM, and terrain-based methods are used in spatially explicit hazard mapping, flood-extent analysis, debris-flow assessments, and floodplain consequence studies (Li Z. et al., 2024; Shahgedanova et al., 2024; Wang et al., 2024; Aitzhanova et al., 2025; Laiskhanov et al., 2025; Zhang et al., 2025). Dataset evaluation and bias correction studies are less frequent but remain important where the reliability of precipitation or hydro-climatic inputs strongly affects interpretation, particularly in mountain and transboundary environments (Song et al., 2022; Duisebek et al., 2025; Zhuo et al., 2025).

Overall, the reviewed literature is expanding in scope and process representation but remains uneven: observational/attribution analysis and climate scenario modeling dominate, while remote sensing, terrain-based hazard mapping, dataset evaluation, predictive techniques, snow/glacier-related hazards, water stress interactions, and spatially detailed risk translation are less developed. This imbalance highlights the need to connect hydro-climatic drivers, basin processes, hazard impacts, and adaptation-relevant interpretations more explicitly.

Structured appraisal of evidence quality and relevance

The quality and relevance of the evidence were evaluated and presented in Figure 6 for the six criteria examined under the large categories of studies identified in Supplementary Table S1 to clarify the reviewed literature. Supplementary Table S2 provides characteristics of the individual studies included, while Figure 6 shows the relative strength and relevance of the predefined study groups and should not be interpreted as a ranking of individual papers. The assessment was designed as a qualitative semi-quantitative appraisal framework intended to compare the relative strength, relevance, and applicability of the reviewed evidence groups. The classifications shown in Figure 6 were derived through the semi-quantitative evidence appraisal framework described in Section 2, with supporting evidence summarized in Supplementary Tables S1–S5 and detailed scoring definitions and interpretive criteria provided in Supplementary Tables S6–S9. The “strong” rating indicates several directly relevant and consistent studies on hydrological extremes or flood hazards in CA. “Moderate” means evidence that is useful but partial, in that it may be geographically or temporally narrow or have limited applicability in practice. The term “limited” is used for data that are sparse, indirect, ancillary, poorly covered at the country level, have limited observational or validation support, or have low direct relevance for adaptation and decision-making.

The appraisal shows that region-wide CA and Kazakhstan-specific studies provide the strongest overall evidence base, particularly for geographic and hazard relevance. Mountain system and transboundary studies are essential for process interpretation, especially for snowmelt, ROS, glacier influence, and shared-basin dynamics, but they remain more constrained by data availability and decision relevance. Arid inland analog and global comparative studies provide useful process-level or conceptual support, but their geographic and decision relevance to CA is more limited. Thus, strong process signals do not necessarily imply equally strong data quality, methodological consistency, or decision-ready applicability.

Hydro-climatic drivers and process changes

Changes in extreme precipitation

Following the evidence-quality appraisal presented in Figures 6, 7, synthesizes hydro-climatic baseline changes and process signals across CA, including warming, intensification of precipitation extremes, declining snow depth, and changing seasonal concentrations of hydro-climatic conditions, while Figure 8 compiles regional evidence strength for major hydro-climatic drivers and flood-related processes. Together, these figures demonstrate that extreme precipitation is one of the most consistently reported and quantitatively supported hydro-climatic drivers of flood-related hazards across CA.

Extreme-precipitation-related research accounted for 23.1% of the 39 studies coded for substantive thematic focus, with evidence ranging from region-wide CA assessments to the mountain headwaters of Kyrgyzstan and Tajikistan, as well as country-specific studies from Uzbekistan. Although the studies used different precipitation indices, time periods, and analytical methods, the reported quantitative changes generally showed a similar trend. At the regional scale, R95p increased from 36.64 to 47.84 mm after 1986, with its trend accelerating from 0.8 to 5.06 mm/10a (Wei et al., 2023). Under RCP4.5, PRCPTOT and Rx1day were projected to increase by 4.95%/°C and 5.79%/°C, respectively (Yao et al., 2021). Attribution evidence from vulnerable upstream regions of Kyrgyzstan and Tajikistan indicated an increase of around 20% in the frequency of extreme precipitation events and more than 100% in their intensity due to anthropogenic forcing (Fallah et al., 2023). Country-specific evidence from Uzbekistan also projected an increase in the 100-year 24-h rainfall intensity in Jizzakh from 1.88 to 2.36 mm/h (Opitz-Stapleton et al., 2022). These estimates cannot be directly compared because they are based on different precipitation metrics and study designs. However, their generally consistent direction provides quantitative support for the continued significance of extreme precipitation in the reviewed evidence base.

Across the reviewed literature, regional, transboundary, and mountain-basin evidence consistently identifies changes in heavy rainfall and precipitation extremes as a major factor influencing flood-generating conditions, particularly in basins where rapid runoff response, limited infiltration, snowmelt interaction, or strong topographic concentration increase sensitivity to short-duration or high-intensity events. This is especially important in CA, where both mountain-fed and arid to semi-arid river systems can experience severe hydrological impacts from intense precipitation despite generally limited annual water availability (Song et al., 2022). The reviewed evidence further indicates that precipitation change should be interpreted not only as a direct flood trigger but also as a broader indicator of hydro-climatic change, particularly where warming, circulation variability, topography, and catchment response interact (Yao et al., 2021; Song et al., 2022; Fallah et al., 2024). At the regional scale, this intensification is reflected in the increase of R95p from 36.64 to 47.84 mm after 1986, with its trend accelerating from 0.8 to 5.06 mm/10a, indicating a marked shift toward heavier precipitation extremes across CA (Wei et al., 2023). This broader destabilization signal is reinforced by projections showing that, under RCP4.5, PRCPTOT and Rx1day increase by 4.95 and 5.79% per 1 °C of warming, respectively, implying thermodynamically strengthened rainfall extremes under continued warming (Yao et al., 2021).

Observed and historical studies indicate that precipitation extremes have intensified in many parts of the region, although the signal is spatially uneven. A regional assessment by Yao et al. (2021) showed that, during 1936–2005, nearly all major extreme precipitation indices across CA, except for consecutive dry days, exhibited increasing trends, with stronger statistically significant changes in wetter subregions, especially in northern and southeastern CA (Yao et al., 2021). This spatial differentiation is important because the regional increase in precipitation extremes does not imply a uniform increase in flood-generating potential; the hydrological consequences depend on local precipitation regimes, basin characteristics, elevation, and interactions with snow- and glacier-influenced processes.

The intensification of precipitation extremes is also supported by attribution evidence, which showed that anthropogenic forcing increased the frequency of extreme precipitation by more than 20% and its intensity by more than 100% in vulnerable upstream regions of Kyrgyzstan and Tajikistan (Fallah et al., 2023). This interpretation is further supported by recent regional studies showing that the spatial structure, elevation dependence, and types of precipitation events contributing to flooding vary considerably across the region. As a result, precipitation-related flood sensitivity appears to be becoming more differentiated rather than uniform across CA (Fallah et al., 2024; Yao et al., 2021). The strong concentration of extreme precipitation in mountainous areas is also evident in event-based analyses, where high-altitude parts of CA experienced up to 698 extreme precipitation events during 1979–2023 (Guo et al., 2025). Comparative evidence from adjacent mountain and arid regions supports this interpretation: for example, Chen et al. (2024) documented a marked interdecadal increase in extreme precipitation in southwestern Xinjiang and linked this to changes in moisture transport, cyclone anomalies, and snow-related circulation feedback. Further supporting studies indicate that extreme precipitation across the broader arid–mountain belt is shaped by interacting circulation controls and moisture-source variability, including South Asian aerosol-related mechanisms and Tianshan moisture transport pathways (Guan et al., 2022; Chen P. et al., 2024; Ji and Tian, 2024). Such process sensitivity is also observed in the moisture-source structure of the Tianshan system: continental moisture sources account for approximately 93.2% of precipitation, while direct oceanic moisture accounts for only 6.8%, with substantial seasonal differences between the dominance of Central Asian and Western Asian moisture sources (Guan et al., 2022). Taken together, these findings show that extreme precipitation in CA is not only intensifying but also becoming more spatially structured and process-dependent, which may lead to greater variability in flood response from one basin to another.

As can also be seen in Figure 7, the literature varies across the region regarding the hydro-climatic drivers that are represented. Warming and extreme precipitation are well documented, while an increase in ROS events is primarily presented as a projected process in Kyrgyzstan and the wider Tian Shan–Pamir system. Major runoff changes are observed in Kazakhstan and at the regional scale of CA, while the pattern of runoff redistribution is better documented in these contexts. Therefore, precipitation intensification alone should not be interpreted as a sufficient predictor of flood response; precipitation timing, antecedent conditions, catchment characteristics, and the resulting river basin response are also important factors in determining whether increasing precipitation translates into greater flood hazard.

However, the reliability of such conclusions remains limited due to significant uncertainty associated with the datasets and climate forcing factors used (Fallah et al., 2024; Yao et al., 2021). Studies evaluating precipitation products in CA and the transboundary Ili Basin show that widely used gridded datasets differ considerably in their accuracy, with some products such as GPCC and CHIRPS generally outperforming others, while ERA5-Land tends to overestimate precipitation, particularly in mountainous terrain (Duisebek et al., 2025; Song et al., 2022). GPCC was best able to represent long-term precipitation when evaluated over 328 meteorological stations, including 272 stations in Central Asian countries and 56 stations in Xinjiang, China, while ERA5-Land, GPM IMERG, and PERSIANN-CDR showed overestimation biases ranging from around 20–60% in some configurations (Song et al., 2022). In the Ili Basin, bias-corrected ERA5 precipitation improved runoff simulation substantially, with NSE reaching 0.83 in calibration and 0.85 in validation, compared with much poorer performance using uncorrected data (Li Z. et al., 2024). In general, the thematic frequency, extensive geographical scope, and various quantitative assessments support the conclusion that extreme precipitation is one of the most frequent and quantitatively well-supported hydro-climatic drivers in the analyzed CA literature. However, the size of the reported changes cannot be directly compared from one study to another due to differences in precipitation indices, time periods, climate scenarios, and analytical techniques. The quantitative interpretation is also sensitive to the choice of forcing data, elevation-dependent bias, and the sparseness of observations within each basin. Snow/ROS-related processes show stronger evidence in Kyrgyzstan, Tajikistan, and the Tianshan–Pamir domain, while compound mechanisms are best represented in Tianshan–Pamir and Tarim-related settings. This confirms that flood hazards in different regions of CA are influenced by spatially variable combinations of precipitation, snow/ROS processes, runoff regime change, and compound interactions.

Snowmelt shifts and snow-to-rain transition

Figures 9, 10 and Supplementary Tables S4, S5 provide the main evidence base for this subsection by summarizing changes in runoff regime, seasonal water availability, snow-related hydro-climatic conditions, and broader baseline shifts across CA. Figure 9 links warming, precipitation intensification, snow decline, snow-to-rain transition, ROS intensification, and anthropogenic pressures to hydrological responses and downstream hazards, while Figure 10 summarizes the regional distribution and strength of related flood evidence. Snowmelt timing is a critical control on hydrological extremes across CA because a large proportion of regional runoff is seasonally concentrated and influenced by snow accumulation, melt timing, and high-elevation hydro-climatic processes. In these environments, climate change affects not only the magnitude of snow resources but also the timing of meltwater release, the seasonal structure of river flow, and the balance between snowfall- and rainfall-driven hydrological responses. This importance is reinforced by regional evidence showing that ~1.2 °C warming over the last ~70 years has already been associated with an approximately 20% decline in snow depth across CA, indicating that snow storage is highly sensitive to ongoing climate change (Fallah et al., 2024).

The reviewed studies provide increasing evidence that climate change is affecting the timing of snowmelt and the seasonal structure of runoff across CA. These shifts are especially important in snow-dominated and mountain-fed basins, where river flows strongly depend on the timing and amount of meltwater. Regional and country-level studies project earlier flood peaks, stronger seasonality, and reduced summer discharge in several key basins under CMIP6 scenarios, including Kazakhstan (Nurbatsina et al., 2025b). Related studies from western Kazakhstan similarly show increasing winter low flow and declining summer–autumn low flow, suggesting that warming is already altering the seasonal structure of runoff beyond peak-flow changes alone (Makhmudova et al., 2025). Comparative mountain-basin evidence further suggests that warming can reduce the relative contribution of snowmelt-driven flooding while increasing the role of rainfall-driven flood processes in some catchments (Niu et al., 2025). More broadly, studies of precipitation phase change indicate that extreme rainfall is increasing faster than extreme snowfall under warming, supporting the interpretation that some cold region and mountain basins may gradually shift toward more rainfall-dominated flood-generating conditions (Li Y.-P. et al., 2025; You et al., 2020). In particular, Sadyrov et al. (2025) showed earlier SWE peaks, reduced snow storage, and an increasing rain contribution in the Naryn basin, while Meng et al. (2022) and Naqi et al. (2025) further highlighted the sensitivity of snow cover and snowmelt representation in Tianshan- and Pamir-related high-mountain environments. This shift has been most quantitatively substantiated in the wider Tianshan–Pamir system by a recent modeling study by Yang et al. (2025), which suggests that by the mid-21st century, the frequency of ROS events will rise by ~49%, the maximum daily intensity of ROS events will rise by ~43–45%, and the annual total ROS intensity will rise by ~66%, while the contribution of snowmelt to runoff will reduce by ~11.97–11.39 percentage points under SSP2-4.5 and SSP5-8.5 conditions. In the Naryn basin, future warming is associated with a ~ 17% decline in maximum SWE, an advance of peak SWE timing by about one month, and runoff declines exceeding 40% in some upper-basin areas under high-emission scenarios (Sadyrov et al., 2025). Concurrently, supporting evidence from other mountain environments suggests that warming does not necessarily lead to a uniform reduction in the significance of snowmelt as a flood-generating factor. In some basins, snowmelt still plays a major role in flooding while rainfall contribution grows. Niu et al. (2025), for instance, estimated the contribution of snowmelt to snowmelt-driven floods and rainfall-driven floods in the Xiying River basin as 85.78 and 10.70%, respectively. The comparative evidence presented here does not replace CA-specific findings but is used to support their interpretation. Combined, these studies suggest that, in addition to impacting melt timing, climate change is also altering the prevailing hydrological drivers of hydrological extremes in snow-influenced basins.

Other processes can also be more abrupt than seasonal snowmelt and, constituting hazards at high altitudes. GLOFs can cause rapid downstream flooding following the sudden drainage of glacial lakes, while permafrost degradation can affect slope stability and hydrological pathways, contributing to cascading mountain hazards. Recent evidence from the Central Asian Tianshan shows a substantial increase in the number and area of glacial lakes and suggests that parts of the western Tianshan are at elevated risk of GLOFs (Chen M. et al., 2024). Permafrost-related controls on the formation and drainage of short-lived glacial lakes have also been recorded in the northern Teskey Range (Daiyrov and Narama, 2021). These hazards are relatively underrepresented in the reviewed CA evidence base and thus represent an important high-altitude research gap rather than processes of limited regional relevance.

These changes have important implications for both flood hazard and seasonal water management. Earlier snowmelt and earlier spring flood peaks can increase flood sensitivity during transitional seasons, while reduced warm-season discharge can intensify water stress later in the year (Fang et al., 2022). At the same time, snow-to-rain transitions may alter the dominant mechanisms of extreme hydrological response, particularly in mountain and transboundary basins. The same applies to basins where spring runoff has historically been regulated by seasonal snow accumulation, as increased rainfall fractions can shorten runoff response times and enhance runoff concentrations during periods of heightened hazard (You et al., 2020; Niu et al., 2025). Enhanced ROS processes further complicate this relationship by combining precipitation phase changes, snowpack conditions, and warm-season transition processes into a single hazard-generating mechanism.

These differences are also important for adaptation in high-altitude catchments. Along with better monitoring of snowpack and runoff, areas affected by GLOFs and permafrost need more focused monitoring of potentially hazardous glacial lakes and unstable terrain. Early-warning systems and improved upstream–downstream information sharing are also needed to respond to rapidly developing mountain hazards (Daiyrov and Narama, 2021; Chen M. et al., 2024). Figure 10 further indicates that ROS-driven flood potential and spatial heterogeneity are more widely represented across regional and mountain systems than direct evidence for snowmelt-only floods. However, the existing evidence base remains spatially heterogeneous, and the number of studies that quantitatively examine these changes as a single, complex process encompassing snowmelt timing, precipitation phase, runoff concentration, and downstream flood risk is still limited (Tursunova et al., 2025b). The limited CA-specific coverage of GLOF and permafrost-related processes adds another important dimension to this evidence gap at high altitudes. This represents a significant gap in the regional literature, particularly regarding the downstream propagation of flood processes, cascading high-altitude hazards, and their interpretation at the river basin level.

Runoff seasonality and redistribution

The focus of this subsection is on Supplementary Table S5 and Figure 11, which present a summary of the reviewed studies related to changes in the runoff regime and seasonal water availability. Together, the evidence indicates that seasonal runoff redistribution is one of the most consistent hydro-climatic features reported for CA. The warming experienced in CA is associated not only with changes in overall water availability but also with changes in water availability at different times of the year, in concentration, and in the timing of flow. The timing of runoff formation has greater hydrological importance than the total amount of runoff in river basins with significant snow cover and snowmelt cycles, particularly where meltwater constitutes a considerable part of the flow (Yapiyev et al., 2025; Tursunova et al., 2025). Hence, seasonal redistribution of runoff has become a major focus in the literature on hydrological extremes in the region.

For CA, the general conclusion of the reviewed studies is a trend toward earlier and more concentrated dates of runoff formation. Regional and country-level data confirm the earlier timing of flood peaks in some important basins, especially in Kazakhstan, along with stronger seasonality and reduced discharge during summer in Kazakhstan and other basins (Nurbatsina et al., 2025b). Studies from western Kazakhstan similarly show increasing winter low flow and declining summer–autumn low flow, indicating that warming and changing precipitation conditions are altering seasonal minimum-flow regimes as well as flood-relevant runoff timing (Makhmudova et al., 2025). Related evidence from broader CA also shows that runoff changes are influenced by both climatic and anthropogenic factors, with upstream sectors often more strongly climate-driven and downstream systems more affected by land-use and water withdrawals (Gulahmadov et al., 2021). Supporting studies from adjacent high-mountain and inland-basin settings also suggest that warming is reducing runoff concentration in some source regions while increasing cold-season runoff and modifying glacier- and snow-fed contributions, which is consistent with a broader reorganization of annual flow structure (Shan et al., 2023; Li Z. et al., 2025). This reorganization is quantitatively evident in several settings. In the Sarykan River, summer runoff declined by 5.3%, while runoff in the remaining months increased by 6.4%, indicating clear intra-annual redistribution (Issaldayeva et al., 2023). In western Kazakhstan, winter low flow increased by up to 5.2 times in parts of the Zhaiyk–Caspian system (Birimbayeva et al., 2025). At the regional scale, Didovets et al. (2024) showed that earlier spring streamflow peaks were widespread across CA catchments, while Gulahmadov et al. (2021) quantified strong upstream–downstream asymmetry in Tajikistan. Their attribution analysis indicated that runoff changes in upstream catchments were predominantly controlled by climate variability (87.96%), whereas runoff changes in downstream catchments were predominantly associated with anthropogenic activities (92.47%). Related evidence from glacier-fed inland basins also shows that climate-driven runoff change can be redistributed unevenly among hydrological components; in the Yarkant River Basin, glacier runoff accounted for 60.5% of the total runoff increase during 1975–2020, highlighting the importance of distinguishing total runoff change from its glacier- and snowmelt-driven components (Li et al., 2025).

High evidence scores across multiple dimensions, not just a single process, are found in the literature for runoff-regime change in Kazakhstan and at the CA scale, as illustrated in Figure 11. The scores do not reflect the magnitude of runoff-regime change and should not be interpreted as implying greater change than in other regions, but rather the strength, consistency, and breadth of the available evidence. Kazakhstan shows strong evidence for runoff spring peaks, runoff redistribution/intensification, winter-flow increase, and water stress pressure, consistent with basin-level evidence of intra-annual redistribution and increasing cold-season flow. The regional CA profile similarly indicates strong evidence for spring peak change and runoff redistribution, along with moderate evidence for reduced summer discharge, winter-flow increase, snowmelt/ROS sensitivity, and water stress pressure. In contrast, the evidence base is somewhat less developed for Kyrgyzstan, Tajikistan, Uzbekistan, and the wider Tian Shan–Pamir system. Existing studies are more likely to address individual processes related to flow-regime modifications rather than considering all these processes together.

The seasonal changes have important implications for adaptation and water resource management. Earlier and more intense spring runoff can exacerbate flooding in transitional seasons. Meanwhile, a decrease in summer flow can lead to increased irrigation needs, greater environmental pressure, and stronger competition for water resources during the latter half of the year (Makhmudova et al., 2025). This makes CA an especially challenging region, where hydrological systems are already situated in a highly spatially heterogeneous environment and are strongly influenced by basin type, elevation, and transboundary water resource management conditions. The reviewed literature thus indicates that seasonal redistribution also represents a management challenge for reservoir operation, irrigation reliability, downstream allocation, and the coincidence of flood pressures with water scarcity in the late season (Gulahmadov et al., 2021; Tursunova et al., 2025, 2025b; Yapiyev et al., 2025; Nurbatsina et al., 2025b). A notable example is western Kazakhstan, where over 70% of the area could face severe water stress by 2050, and per-capita water availability could drop to 900–1,000 m 3 /person/year (Tursunova et al., 2025b). While several studies have shown seasonal-scale redistribution of runoff, very few translate these changes into basin management, infrastructure, or flood risk assessment. This remains a major gap, highlighting the need to connect seasonal flow redistribution more directly with reservoir operation, irrigation reliability, downstream water distribution, flood risk assessment, and adaptation planning.

Compound processes and interacting controls

Figure 12 and Supplementary Table S4 are central to this subsection. The combination of Supplementary Table S4 and Figures 10–12 indicates that, in most cases, hydrological extremes in CA are driven by multiple processes interacting with one another, rather than by any single process. Figure 12 also illustrates the regions where compound susceptibilities are most evident. These mechanisms involve extreme precipitation, the seasonality of snowmelt, glacier and alpine-runoff contributions, antecedent hydrological conditions, the seasonality of runoff concentration, and human intervention (e.g., water abstraction, irrigation, reservoir operation, river regulation, and land surface change). This multi-driver behavior is particularly relevant in basins supported by mountains, in arid basins, and in transboundary basins, where minor shifts in climate forcing or water allocation can cause substantial changes in hydrology. However, in these systems, climatic and human drivers are not necessarily independent; human alterations of catchments and river systems can affect the magnitude, timing, and downstream transport of responses to climate change (Gulahmadov et al., 2021; Hu et al., 2021).

The interactions can be seen in various contexts, as shown in Figure 12. These include snowmelt–rainfall interactions in Kazakhstan, upstream–downstream human–climate contrasts in Tajikistan, compound heat-rainfall flood extremes in the Tarim River Basin, and hydrological extremes driven by ROS across the wider Tianshan–Pamir area. When assessing anthropogenic influence, the contrast between upstream and downstream is even more significant, as signals from the mountain headwaters may be altered by anthropogenic activities such as water abstraction, irrigation, and reservoir operation as the water flows through intensively managed downstream areas.

These interacting mechanisms can be directly observed in several of the reviewed studies. The very heavy flood in August 2022 was not generated uniformly throughout the Tarim River basin. High temperatures and heat-related processes dominated some river systems; rainfall dominated others, while a combination of heat and rainfall impacted some river systems (Yao et al., 2024). Similarly, the magnitude and timing of flood variations for the alpine Tarim catchments were related to various combinations of precipitation, temperature, melt-level height, and the cryospheric conditions of the catchments (Fang et al., 2023). In addition, Chen M. et al. (2024) demonstrated that the trends of extreme precipitation in nearby arid–mountain regions can be linked to circulation anomalies, moisture transport changes, and large-scale snow-related feedback effects from an atmospheric perspective. Su et al. (2025) demonstrated that flooding could also be intensified in extremely arid desert environments, and the different spatial characteristics of flood–drought severity and flood disaster concentration in the Xinjiang region were also reported by Naibi et al. (2025) and Wang et al. (2023). The Tarim event of August 2022 is the clearest quantitative example, where various sub-basin controls with different drivers during the same event system were observed, such as rainfall-dominated, heat-dominated, and combined heat–rainfall drivers (Yao et al., 2024).

Other analogue evidence indicates that flood intensification can also occur in areas with mean annual precipitation below 100 mm when warming, short-duration precipitation events, and geomorphic sensitivity interact. Related studies also report strong non-linear flood-drought relationships with precipitation extremes and substantial debris-flow instability in high-mountain valleys (Su et al., 2025; Naibi et al., 2025; Sidle et al., 2025a). Much of this analogue evidence comes from adjacent arid–mountain environments rather than CA itself and is used here to strengthen the interpretation of compound hazard processes, not to replace regional evidence from CA. Overall, these results suggest that compound behavior is observed not only at the event scale, when drivers co-occur, but also over longer time scales through interactions among warming, rainfall intensification, cryospheric change, geomorphic sensitivity, and basin-specific runoff responses.

Anthropogenic modification adds another important dimension to these compound controls. In the Kofarnihon River Basin of Tajikistan, Gulahmadov et al. (2021) showed a clear spatial difference in the relative influence of climatic and human drivers. Climate change accounted for 87.96% of runoff change in the upstream part of the basin, while anthropogenic activities accounted for 92.47% of runoff change downstream. This difference illustrates how the hydrological signal generated by climatic forcing in mountain headwaters can be considerably modified as water moves through more intensively managed downstream areas. Similar evidence from the Amu Darya River Basin shows that long-term runoff reduction was mainly influenced by human activities related to population growth, expansion of cropland, reservoir construction, and water management. Between 1960 and 2017, runoff progressively decreased downstream, while the estimated contribution of human activities to runoff reduction exceeded 100% at the analyzed stations because anthropogenic effects countered an opposing climatic contribution (Hu et al., 2021).

These interactions are mediated by land-use change and water infrastructure. The expansion of irrigated agriculture results in greater consumptive water use and changes in downstream discharge, while reservoirs and irrigation infrastructure alter the storage, timing, and redistribution of river flow. Evidence from the Syr Darya Basin, used here as supporting process context because it predates the formal review period, shows that upper-basin discharge can increase with warming and enhanced glacier melt, while irrigation expansion and reservoir operation are associated with decreasing middle- and lower-basin discharge (Zou et al., 2019). The same study reported significant expansion of construction and agricultural land and suggested that reservoir and irrigation–canal operation are also important controls on downstream river discharge. This example demonstrates how increases in runoff or precipitation caused by climate change may not directly translate into equivalent downstream hydrological changes in heavily managed basins.

In the CA evidence base, urbanization is a related but less well-quantified pathway. Expansion of impervious and built surfaces may lead to decreased infiltration, shorter runoff response times, and increased runoff concentration, along with greater urban flood susceptibility, while land-cover changes may alter evapotranspiration, soil–water storage, erosion, and hydrological connectivity. Results from the Syr Darya Basin suggest that the expansion of construction land has changed runoff characteristics and has been associated with higher urban flood risk (Zou et al., 2019). Compared with irrigation, water abstraction, and river regulation, however, direct quantitative attribution of flood-related changes to urbanization and land-use change remains quite limited in the reviewed CA literature. These processes are therefore considered important but insufficiently quantified components of the regional flood risk system, rather than uniformly demonstrated dominant drivers.

However, the existing evidence base still considers many of these controls separately. Many studies examine precipitation, snowmelt, runoff redistribution, or water stress individually, while fewer explicitly quantify the interactions between these factors within integrated flood-relevant frameworks. This is especially true for anthropogenic influences. It has been shown that water abstraction, irrigation, reservoir operation, land-use change, and other forms of basin management can significantly alter hydrological outcomes, but the interactions between these elements and changing precipitation extremes, snowmelt timing, and cryospheric runoff are seldom addressed within a common attribution framework. The current literature therefore has one of its main weaknesses not so much in the lack of process knowledge as in the limited integration of already known controls into unified compound process and compound hazard analyses (Gulahmadov et al., 2021; Tursunova et al., 2025b; Hu et al., 2021).

Therefore, the analysis of compound events and processes remains a major priority for future research. Future assessments should go beyond climate-only and/or single-driver approaches and explicitly address how climatic forcing interacts with land-use and land-cover change, urban development, irrigation and abstraction, reservoir operation, and other water management interventions. This limitation underscores the importance of developing frameworks that integrate climate forcing, hydrological response, catchment and infrastructure modification, exposure, vulnerability, and management decisions into a single assessment.

Regional patterns and practical implications

Based on the reviewed literature, hydrological extremes and flood-related hazards in CA are associated with several pathways rather than a single dominant mechanism. Figure 13 offers an integrated synthesis of regional susceptibility to climate-driven hydrological extremes and flood-related hazards, based on dominant climate-driven hazard drivers, strength of evidence, evidence horizon, and key quantitative findings for the main mountain-basin settings in and around CA. Panel A categorizes the dominant driver types, including snowmelt, rainfall extremes, ROS, and compound hazards, whereas Panel B anchors the synthesis in selected quantitative benchmarks. The major regional trends, susceptibilities, and practical consequences derived from the reviewed literature are summarized in Supplementary Table S6. Instead of organizing the evidence primarily around Kazakhstan subregions, the synthesis follows the main regional settings represented in the literature, including region-wide CA studies, individual country settings, and transboundary or mountain-source systems. This structure better reflects the evidence developed in Sections a–g, where regional, shared-basin, and mountain system studies are central to understanding hydrological extremes, runoff instability, flood susceptibility, and adaptation challenges across CA (Didovets et al., 2024; Fallah et al., 2024; Ceresa et al., 2025; Guo et al., 2025; Sadyrov et al., 2025; Yang et al., 2025). However, regional susceptibility should not be considered solely in terms of climatic forcing, since the downstream effects of hydrological changes driven by climate change can be significantly altered by water abstraction, irrigation, reservoir operation, river regulation, land-use change, and other catchment modifications (Gulahmadov et al., 2021; Hu et al., 2021).

Figure 13 should be interpreted as a structured qualitative evidence synthesis matrix rather than a quantitative susceptibility model. The evidence classes in Panel A represent the relative strength and consistency of the reviewed literature for each hazard or driver category, while the asterisks indicate where explicit quantitative metrics were available. Thus, stronger markers indicate stronger or more quantitatively supported evidence, not necessarily greater physical hazard magnitude. The anthropogenic-influence category should likewise be interpreted as evidence of human modification of hydrological response rather than as a uniform measure of human-induced flood magnitude.

This regional synthesis should be interpreted in light of evidence imbalance. Kazakhstan and region-wide CA studies are more strongly represented than those specific to Uzbekistan or Turkmenistan, while several process-level insights on mountain hazards, rainfall-on-snow (ROS), and compound mechanisms rely on transboundary and broader Tianshan–Pamir evidence (Shahgedanova et al., 2024; Ceresa et al., 2025; Tursunova et al., 2025; Yang et al., 2025; Nurbatsina et al., 2025a, 2025b). This is important because many hydrologically critical processes originate in shared mountain headwaters and are subsequently modified downstream by withdrawals, regulation, exposure, and uneven adaptive capacity, making national-scale interpretation alone insufficient in many basin settings (Gulahmadov et al., 2021; Duisebek et al., 2025; Gulakhmadov et al., 2025; Sadyrov et al., 2025). There are also basin-scale lines of evidence showing that the relative importance of climate and human influences can change significantly from upstream to downstream. In the Amu Darya River Basin, long-term runoff reductions were mainly attributed to human activities, particularly reservoir construction associated with cropland expansion and irrigation demand, while climatic effects partly counteracted the observed runoff decline (Hu et al., 2021). This further implies that basin-scale and transboundary frameworks are essential for reliable flood risk interpretation and adaptation planning in CA.

At the region-wide scale, the literature consistently indicates intensifying precipitation extremes, changing runoff seasonality, declining snow storage, and uneven adaptation capacity across CA (Yao et al., 2021; Fallah et al., 2023; Wei et al., 2023; Ceresa et al., 2025). Quantitatively, this signal is reflected in the increase of R95p from 36.64 to 47.84 mm after 1986, the acceleration of its trend from 0.8 to 5.06 mm/10a, and the broader intensification of PRCPTOT and Rx1day under warming (Yao et al., 2021; Wei et al., 2023). Hydrological stress is also an important part of the regional picture. Under high-emission scenarios, approximately 39.01% of the land area and 35.9% of the population in CA may be exposed to severe drought risk, indicating that seasonal runoff instability can coexist with broader water security pressures alongside flood-related hazards (Wu et al., 2025). These climate-induced pressures also interact with water management pressures. Changes in runoff availability, precipitation extremes, and snowmelt timing can be amplified, redistributed, or partially obscured by water abstraction, storage, irrigation demand, and reservoir operation, complicating the attribution of downstream hydrological changes, especially in highly regulated and irrigated basins (Gulahmadov et al., 2021; Hu et al., 2021). Combined, these findings underscore the need for a more comprehensive focus on hydro-climatic instability and human alteration of basin response to understand flood risk in CA, rather than considering individual hazard processes alone.

Country-level evidence further refines this general pattern. Kazakhstan is the most well-represented individual country setting and provides strong evidence for snowmelt-related flooding, runoff redistribution, water stress, and basin-scale hydrological change (Makhmudova et al., 2023, 2025; Tursunova et al., 2025, 2025b; Nurbatsina et al., 2025a, 2025b). Historical peak discharges have exceeded 4,000 m 3 /s in the Yesil basin and 1,700 m 3 /s in the Zhabai basin; floodplain inundation depth has reached up to 639 cm in the Yesil basin; and the flood-inundated area increased by 8.25 km 2 /year in parts of Kazakhstan from 1970 to 2007 (Makhmudova et al., 2023; Wang et al., 2024). Existing indicators also suggest increasing future basin stress, with more than 70% of the West Kazakhstan Region projected to experience severe water stress by 2050 and per-capita water availability expected to decline to 900–1,000 m 3 /person/year (Tursunova et al., 2025b). In such settings, irrigation demand, reservoir operation, and water allocation become increasingly relevant because they determine how climate-driven changes in seasonal water availability are transmitted through managed river systems. This makes Kazakhstan one of the most data-rich reference systems for translating hydro-climatic change into applied flood risk interpretation in CA.

Kyrgyzstan and Tajikistan are more closely associated with mountain-source hydrology, snowmelt sensitivity, ROS intensification, compound mountain hazards, and upstream–downstream contrasts in runoff response (Gulahmadov et al., 2021; Uwamahoro et al., 2021; Kalashnikova et al., 2024; Sidle et al., 2025a; Yang et al., 2025). In Kyrgyzstan, runoff in southern basins is projected to reach 126–202% of normal in Isfayram-Sai and 139–218% in Shakhimardan by 2029–2040, while improved snowmelt representation increases simulated flood peaks by 43.87–49.42% in Issyk-Kul catchments (Uwamahoro et al., 2021; Kalashnikova et al., 2024). In Tajikistan, climate explains 87.96% of runoff change upstream in the Kofarnihon basin, whereas anthropogenic activity explains 92.47% downstream (Gulahmadov et al., 2021). A mostly climate-driven upstream signal that is gradually modified by human water use and basin management is one of the clearest examples in the literature of the interaction between climatic and anthropogenic controls. The contributions of glaciers, snow, and rainfall are also important in high-elevation basins relevant to the broader interpretation of CA. Evidence from neighboring glacier-fed inland basins shows that glacier runoff was responsible for 60.5% of the overall increase in runoff in the Yarkant River Basin between 1975 and 2020 (Li Z. et al., 2025). These are process-dominant source areas where hydro-climatic risk signals can be carried downstream across CA; however, their downstream expression depends partly on the level of abstraction, irrigation, regulation, and other human interventions along the river system.

Uzbekistan is less densely covered in the reviewed literature; however, the available evidence points to uneven susceptibility associated with increased rainfall intensity, heat stress, and downstream water pressure (Ceresa et al., 2025; Zhuo et al., 2025; Opitz-Stapleton et al., 2022). The limited quantitative evidence indicates a warming of approximately 0.17 °C/decade, with more than half of the country projected to experience strong heat stress by the 2050s under RCP4.5, as well as increases in 100-year 24-h rainfall intensity in locations such as Jizzakh, from 1.88 to 2.36 mm/h (Opitz-Stapleton et al., 2022; Zhuo et al., 2025). Irrigation demand, abstraction, reservoir operation, and upstream water management decisions are also significant components of hydrological exposure and water security risk in Uzbekistan, as it is predominantly located downstream in the major transboundary river systems. Results from broader Central Asian evidence in the Syr Darya Basin show that irrigation expansion and reservoir operation can intercept and redistribute runoff even when upper-basin discharge increases, leading to reduced discharge in the middle and lower parts of the basin (Zou et al., 2019). This study predates the formal review period and is therefore used here only as supporting context when interpreting downstream climate signals.

In contrast, evidence for Turkmenistan is much more limited, and the available information supports only cautious interpretation of future flood risk changes at the national scale. This represents a major geographic evidence gap and limits comprehensive risk assessment across CA, especially where shared headwaters, climate-forcing uncertainty, basin connectivity, and compound runoff-generation mechanisms are important (Shahgedanova et al., 2024; Duisebek et al., 2025; Sadyrov et al., 2025; Yang et al., 2025). The limited evidence is even more pronounced for explicit attribution of flood-related changes to urbanization, land-use change, and water infrastructure in Turkmenistan and several other underrepresented CA settings.

Overall, three regional patterns emerge: relatively strong multi-driver evidence for Kazakhstan and region-wide CA; process-dominant mountain-source evidence for Kyrgyzstan, Tajikistan, and the broader Tianshan–Pamir system; and weaker evidence for Uzbekistan, Turkmenistan, and some arid inland areas. Across these settings, anthropogenic influence also varies from one area to another. Water abstraction, irrigation, and river regulation are better documented than urbanization and land-use change, while the interaction between climatic forcing and human modification is most evident in heavily managed downstream and transboundary basins. These patterns indicate that knowledge of flood risk in CA is spatially structured, process-based, and constrained by data availability.

The reviewed literature demonstrates a clear distinction between lowland and steppe systems, on one hand, and mountain or mountain-fed systems, on the other. Lowland and steppe river basins are more often associated with spring floods caused by snowmelt, seasonal redistribution of runoff, and imbalances in water availability at different times of the year. These areas are also generally more affected by direct human modification through reservoirs, irrigation systems, water abstraction, agricultural land use, and urban development, which can alter runoff storage, timing, concentration, and downstream exposure. Evidence from the Syr Darya Basin, for example, shows a substantial expansion of construction and agricultural land and indicates that irrigation and reservoir operation significantly influenced river discharge in the middle and lower parts of the basin (Zou et al., 2019). In contrast, mountain and mountain-fed systems are characterized by a more complex interaction of processes, including increased ROS, the transition from snow to rain, glacier influence, susceptibility to flash floods, and higher uncertainty regarding climate forcing (Makhmudova et al., 2023; Shahgedanova et al., 2024; Sadyrov et al., 2025; Tursunova et al., 2025; Yang et al., 2025). High-altitude areas may also be exposed to more localized cryosphere-related hazards, such as GLOFs and cascading processes linked to glacier and permafrost changes. However, these hazards are still relatively underrepresented in the reviewed CA literature (Daiyrov and Narama, 2021; Chen M. et al., 2024). In the broader Tian Shan–Pamir system, ROS frequency is projected to increase by approximately 49%, annual total ROS intensity by approximately 66%, while the contribution of snowmelt to runoff may decrease by approximately 11–12 percentage points. Mountain hazard zones also exhibit pronounced local vulnerability. For example, the Gunt Valley contains 49 debris flow cones, of which 34 are unstable, and 63% receive avalanche deposits (Sidle et al., 2025a; Yang et al., 2025). However, data from adjacent mountain areas indicate that warming does not necessarily mean a uniform decline in the role of snowmelt. In some basins, snowmelt may remain the dominant factor in flood formation even with an increasing share of rainfall (Niu et al., 2025). Data uncertainty is also an important feature of the regional evidence base, particularly in mountainous and transboundary basins where reliable hazard interpretation is limited by the quality of precipitation data, snow process representation, and altitude-dependent factors (Song et al., 2022; Duisebek et al., 2025; Yang et al., 2025).

From a practical perspective, these findings imply that flood risk management strategies in CA should be differentiated by hydro-climatic regime, including (i) early-warning systems for snowmelt-dominated basins, (ii) flash-flood monitoring in rainfall- and mountain-driven systems, and (iii) integrated seasonal water management in regions prone to runoff redistribution. In high-altitude cryospheric environments, these measures must be coupled with targeted monitoring of potentially hazardous glacial lakes and unstable terrain, along with early-warning systems for rapidly developing hazards in upstream areas (Daiyrov and Narama, 2021; Chen M. et al., 2024).

For regulated, irrigated, and increasingly developed lowland basins, adaptation should also include reservoir operation rules, irrigation and abstraction scenarios, land-use and urban development changes, and upstream–downstream water allocation effects. This is especially important because changes in extreme precipitation, snowmelt, or headwater runoff may not be directly reflected in comparable downstream changes, where human water use and infrastructure greatly influence river discharge (Gulahmadov et al., 2021; Hu et al., 2021; Zou et al., 2019). From an applied perspective, effective adaptation in CA will require stronger coordination among hydro-climatic monitoring, basin-scale forecasting, floodplain management, irrigation planning, reservoir and infrastructure operation, land-use planning, urban development, and cross-border water coordination (Ceresa et al., 2025; Issakov et al., 2025; Sidle et al., 2025a). Yet, the literature is far more developed in diagnosing regional hazard patterns than in translating them into operational planning thresholds, infrastructure screening criteria, or decision-ready frameworks for flood risk management, indicating a persistent gap between hydro-climatic understanding and applied flood risk management. This gap is especially significant for anthropogenic drivers, as the impacts of irrigation and regulation are better documented than those of land-use change and urbanization, while relatively few CA studies quantify the contributions of both climatic and anthropogenic drivers within flood risk assessments.

Evidence quality and methodological limitations

The reviewed literature demonstrates a general increase in scope and analytical sophistication across CA. Since 2021, studies have expanded in terms of research context, including country-level, basin-scale, transboundary, and mountain system analyses, along with greater use of future climate scenarios, particularly those based on CMIP6 climate models (Nurbatsina et al., 2025b) Recent work also moves beyond trend detection by integrating hydrological modeling, uncertainty assessment, precipitation product evaluation, remote sensing, and basin-scale climate impact analysis (Song et al., 2022; Aitzhanova et al., 2025; Duisebek et al., 2025). These advances have improved understanding of runoff timing, flood sensitivity, water stress, basin variability, and mountain hazard processes. Further progress is also reflected in improved model performance, especially where bias-corrected forcing and hybrid modeling approaches have enhanced the skill of runoff and flood simulations in mountain-fed basins (Ren et al., 2023; Li Z. et al., 2024).

Despite these advances, important challenges remain because of limitations in hydro-climatic data. The reliability of trend detection, model calibration, and regional comparison is diminished in areas with sparse observations, particularly in mountainous regions, headwaters, and remote basins (Naqi et al., 2025; You et al., 2020). This creates specific issues in regions with significant elevation-dependent precipitation, glacier influence, and snowmelt control on hydrological response (Yao et al., 2021). Precipitation and snowfall records also show wide variation in terms of terrain, season, and elevation. GPCC was found to be the most successful in arid CA when compared with station observations for long-term precipitation, while ERA5-Land, GPM IMERG, and PERSIANN-CDR showed overestimation biases of 20–60% in some configurations (Song et al., 2022). The majority of products demonstrated correlations greater than 0.7 with gauge observations in the Ili Basin, although in a few seasonal cases, PERSIANN had correlation values below 0.6 (Duisebek et al., 2025). However, uncertainties in snowfall products are also significant in high-mountain areas (Ding et al., 2024).

Other challenges relate to methodological inconsistencies. The available literature includes observational trend analysis, hydrological simulations, climate scenario modeling, remote sensing studies, and dataset evaluations, but with varying assumptions, bias correction methods, validation depths, spatial scales, and hazard definitions. Some historical studies are aggregated with projections simulated by the models of the Coupled Model Intercomparison Project, including CMIP5 or CMIP6, which complicates direct comparison (Yao et al., 2021; Gulakhmadov et al., 2025). Future-oriented studies are particularly diverse, involving basin-scale CMIP6 hydrological modeling, broader climatic indicators, earlier CMIP5-based projections, and indirect statistical relationships derived from climate data (Nurbatsina et al., 2025b). Consequently, the literature is rich in information but methodologically uneven, requiring careful handling when used for regional synthesis (Yapiyev et al., 2025). Additionally, a literature search limited to English-language publications may not have captured some relevant studies published in Russian or in the national languages of Central Asian countries. While this approach ensured uniformity and transparency in the screening and evidence synthesis process, it may have introduced language bias, leading to the exclusion of some locally published evidence.

The evidence base is also geographically and thematically imbalanced. Coverage is concentrated in region-wide CA studies and Kazakhstan, while Uzbekistan and especially Turkmenistan remain underrepresented. Kyrgyzstan and Tajikistan are more often examined through mountain-basin, rainfall-on-snow, snowmelt, or shared-headwater studies than through broad national syntheses. Many transboundary systems cannot be adequately interpreted within national boundaries alone because downstream hazards are influenced by shared headwaters, upstream–downstream contrasts, climate, topography, water use, regulation, and basin management decisions (Tursunova et al., 2025b; Aitzhanova et al., 2025; Nurbatsina et al., 2025a; Birimbayeva et al., 2025).

Thematic imbalance is equally evident. The literature is relatively strong on hydro-climatic drivers, extreme precipitation, snowmelt timing, runoff redistribution, mountain hazard processes, and water stress, but weaker on direct flood risk translation. Fewer studies address inundation, infrastructure-focused flood risk assessment, vulnerability, exposure, operational hazard mapping, or adaptation-oriented thresholds. Although recent work has begun to examine floodplain degradation, human–flood proximity, evacuation benefits, nature-based solutions, and adaptation heterogeneity, these areas remain weakly integrated with physical process studies (Aitzhanova et al., 2025; Sidle et al., 2025a; Yapiyev et al., 2025). These gaps limit the practical use of current evidence for basin planning, mountain hazard management, and locally targeted adaptation.

In general, the evidence base is stronger for diagnosing hydro-climatic change than for producing decision-ready knowledge on flood and hydrological risk. Many studies document changes in precipitation extremes, snowmelt timing, runoff seasonality, mountain hazards, and water stress, but fewer studies link these changes to exposure, vulnerability, infrastructure sensitivity, flood management priorities, operational tools, or integrated risk products (Wasti et al., 2022; Tursunova et al., 2025b). Similar gaps can be observed in forecasting and predictive systems. Although machine learning-based hazard models are becoming increasingly feasible in High Mountain Asia, operational forecasting approaches remain underdocumented in the CA-specific literature. These limitations inform the gap-to-priority framework outlined in Figure 14.

Research gaps and future priorities

Although the literature on hydrological extremes and flood-related hazards in CA has grown considerably, several important knowledge gaps remain. These gaps relate not only to the number of available studies but also to the conceptualization of hydro-climatic processes, their consistent evaluation across different regional contexts, and the effective transfer of knowledge into adaptation-relevant risk assessment. As summarized in Figure 14, improving comparability, strengthening mountain and transboundary evidence, incorporating multiple drivers, and better connecting hydro-climatic understanding with flood risk management needs will be key to future progress (Ceresa et al., 2025; Issakov et al., 2025; Yang et al., 2025). Another limitation across the reviewed literature is the relatively limited integration of anthropogenic drivers, particularly land-use change, urban development, irrigation, water abstraction, reservoir regulation, and their interactions with changing climatic conditions. Thus, the priority is not only to increase the number of studies but also to make them more comparable, process-integrated, observationally robust, and decision-relevant.

A first priority is the development of integrated multi-driver assessments. Many existing publications examine extreme precipitation, snowmelt timing, glacier contribution, runoff redistribution, water stress, mountain hazards, or adaptation separately. However, emerging evidence indicates that hydrological extremes in CA frequently result from combined conditions involving rainfall, snowmelt, temperature anomalies, antecedent conditions, basin sensitivity, and anthropogenic pressure (Tursunova et al., 2025b). Therefore, future work should develop attribution frameworks that connect precipitation, snow and glacier processes, runoff timing, antecedent wetness, catchment sensitivity, and human modification within a single hazard-relevant design (Makhmudova et al., 2025). These frameworks should clearly separate the relative impacts of climatic forcing and anthropogenic interventions and also evaluate how these contributions vary spatially between relatively natural headwater areas and increasingly regulated and developed downstream reaches. The Kofarnihon River Basin demonstrated that anthropogenic activities accounted for 92.47% of runoff change in the downstream part of the basin, while 87.96% of the runoff change in the upstream part was attributed to climate change (Gulahmadov et al., 2021). Additionally, according to a long-term basin-scale analysis of the Amu Darya, runoff reductions can be primarily attributed to human activities, such as reservoir construction, cropland expansion, and irrigation demand, even when climatic influences occur simultaneously (Hu et al., 2021). This is particularly important for mountain-fed and inland arid systems, where rainfall, melt contribution, runoff timing, and basin-specific sensitivity are closely linked (Fang et al., 2023; Yao et al., 2024; Yang et al., 2025). Sequence-based deep learning methods, such as LSTM and Transformer networks, could help capture non-linear dynamics among precipitation, snowpack evolution, temperature anomalies, antecedent moisture, and runoff timing. This is important for identifying compound flood-generating conditions in mountain-fed and transboundary basins. Where data are available, these approaches should also consider reservoir operation, water abstraction, irrigation demand, and land-surface change, so that predictive models can better represent managed basins rather than treating them as natural systems. These models have already proven useful for flood forecasting and streamflow prediction in settings with temporally dependent and non-linear hydrological responses (Li W. et al., 2024; Nearing et al., 2024).

A second priority concerns hydro-climatic observations, forcing data, and validation. Monitoring in high-mountain areas remains sparse, while uncertainty in precipitation and snow-related datasets continues to affect regional interpretation. Monitoring in mountain and transboundary regions should be improved, and reanalysis-derived products need to be evaluated against the limited available observations (Yao et al., 2021; Song et al., 2022; Duisebek et al., 2025). Additionally, monitoring should be extended to high-altitude areas for cryosphere-specific hazards that tend to be underrepresented in the CA evidence base, including potentially hazardous glacial lakes and permafrost-related instability. Improved observations of these processes would support a more explicit assessment of GLOFs and cascading mountain hazards and facilitate their inclusion in flood risk assessment and early-warning systems (Daiyrov and Narama, 2021; Chen P. et al., 2024). Future research should also explicitly quantify how input-data uncertainty affects hydrological conclusions. A systematic evaluation of snowfall and snow-cover datasets is also needed in high-mountain contexts, where forcing errors can directly influence the interpretation of flood processes and runoff seasonality (Ding et al., 2024; Naqi et al., 2025). Hybrid data-fusion frameworks that integrate satellite observations, reanalysis products, and sparse in-situ measurements may help improve precipitation, snow-depth, and hydro-climatic forcing quality in poorly gauged mountain regions, where observational gaps and elevation-dependent biases remain major sources of uncertainty (Lyu and Yong, 2025; Qiao et al., 2025).

A third priority is improved flood characterization. Compared with the growing number of studies on extreme precipitation, snowmelt change, and runoff redistribution, relatively little attention has been paid to multidimensional flood properties such as frequency, duration, timing, antecedent wetness, and event sequencing. This limits the understanding of why similar climatic signals may produce different flood responses across basins. Future studies should therefore extend beyond event occurrence and runoff shifts toward multidimensional flood analysis that incorporates preconditioning effects, flood timing, duration, and basin wetness memory (Liu et al., 2022; Makhmudova et al., 2023; Yao et al., 2024; Yang et al., 2025). Such analyses should also consider whether differences in flood response arise partly from human modifications of catchment storage, flow pathways, and downstream exposure rather than climatic forcing alone. Event-based machine learning and hybrid hydrological–statistical frameworks could support this priority by classifying flood typologies, detecting sequencing behavior, and distinguishing rainfall-driven, snowmelt-driven, ROS-related, and compound flood events across diverse basin settings (Zhou et al., 2023).

A fourth priority is better integration of catchment and regulation effects. Basin properties, terrestrial storage, dam and reservoir regulation, irrigation withdrawals, land cover, soils, topography, and land degradation are not yet incorporated consistently, despite their importance in shaping flood response across CA (Gulahmadov et al., 2021; Liu et al., 2022). This gap should extend beyond regulation alone to incorporate land-use and land-cover change, expansion of cropland, urbanization, and infrastructure growth. For instance, construction and residential land in the Kofarnihon Basin increased from 248.63 km 2 in 1990 to 685.45 km 2 in 2015, while the dominant anthropogenic contribution to downstream runoff change reached 92.47% (Gulahmadov et al., 2021). This provides direct evidence from CA that land-use transformation and hydrological change can occur together within the same basin.

Cropland expansion and reservoir construction for irrigation are also important human controls on runoff, with human activities contributing more than 100% of the estimated runoff reduction at the analyzed stations because climatic effects partly counteracted the observed runoff decline (Hu et al., 2021). These results show why future attribution studies should consider climatic forcing, land-cover change, infrastructure development, irrigation demand, and reservoir operation together rather than treating all anthropogenic effects as a single residual category of “anthropogenic influence.” Future studies should clarify how natural and managed basin properties amplify or buffer climate-driven hydrological extremes, especially in transboundary and downstream settings where hydrological outcomes reflect regulation, abstraction, and basin management decisions as well as climate forcing (Ceresa et al., 2025; Issakov et al., 2025). Particular attention should also be given to the effects of reservoirs on hydrograph timing and seasonal flow redistribution. Large reservoirs in the upper Syr Darya and Amu Darya can reduce spring and summer peak flows while increasing autumn and winter flows, demonstrating that infrastructure can substantially change the seasonal timing of hydrological variability. Greater distinction is also needed in glacier- and snow-fed basins between total runoff change and changes in contributing components (Li Z. et al., 2025).

Urbanization is another related but still underdeveloped field of study. Expansion of built surfaces can affect infiltration, runoff concentration, drainage response, and exposure, but explicit quantitative attribution of flood-related changes to urbanization is less developed in the CA literature than analyses of irrigation, water abstraction, and reservoir regulation. Future studies should therefore separate the effects of urban growth on flood generation from changes in exposure and vulnerability and combine land-use scenarios with climate-driven projections of precipitation and runoff. At present, urbanization should be considered an important research gap rather than a uniformly demonstrated dominant flood driver across CA because the available evidence remains relatively limited within the formal review sample.

The fifth priority is stronger country-specific and cross-basin synthesis. While the evidence base is growing, many studies have been conducted at the basin scale, on particular events, or on a limited number of process domains (Yapiyev et al., 2025). More systematic comparisons are needed across snowmelt-dominated lowland systems, water-stressed western basins, mountain-fed catchments, and transboundary river systems (Nurbatsina et al., 2025b). Such comparisons should also differentiate relatively natural basins from heavily regulated, irrigated, urbanized, or otherwise modified catchments because similar climate forcing may produce substantially different hydrological responses under varying levels of human intervention. This would reduce reliance on generalized regional evidence and provide a clearer understanding of flood sensitivity, seasonal runoff change, and hydrological risk across CA.

A sixth priority is the enhanced integration of transboundary systems into climate-risk research. Many river systems in CA are shaped by shared headwaters, upstream water use, reservoir regulation, downstream exposure, allocation pressures, and varying capacities for adaptation. Future research should link hydrological change more explicitly with transboundary governance, basin management, and shared-basin adaptation challenges, especially where climate-driven water variability intersects with competing demands from different sectors (Gulahmadov et al., 2021; Wang et al., 2024; Tursunova et al., 2025b). This requires clear upstream–downstream attribution because climatic changes generated in mountain-source regions can be amplified, reduced, delayed, or redistributed by reservoirs, irrigation systems, water abstraction, land-use change, and downstream development. The Kofarnihon and Amu Darya studies show how the main control on runoff can shift from climate upstream to human intervention downstream (Gulahmadov et al., 2021; Hu et al., 2021). This is particularly important because many key hydrological processes do not align with national borders (Ceresa et al., 2025; Duisebek et al., 2025; Gulakhmadov et al., 2025; Sadyrov et al., 2025).

A seventh priority is to move from hydro-climatic diagnosis toward adaptation-relevant risk assessment. Future research should link climate-sensitive hydrological change more consistently with settlement patterns, infrastructure systems, irrigation dependence, floodplain exposure, disaster risk reduction, and operational planning. Land-use planning, urban development, reservoir operating rules, irrigation demand, and water-allocation decisions should therefore be incorporated alongside climate projections when evaluating future flood and runoff risk. A common requirement is to move from identifying hazard-relevant changes toward defining actionable thresholds, vulnerable locations, and adaptation pathways (Wang et al., 2024; Ceresa et al., 2025; Sidle et al., 2025b). In regulated basins, these thresholds should ideally be assessed under both climate and management scenarios so that adaptation options can separate unavoidable climate-driven changes from hydrological impacts that may still be reduced through operational or land management decisions. This transition could also be strengthened through more explicit testing of predictive and data-driven tools, including machine learning-based hazard frameworks, which remain underused in CA-specific research. Recent AI-based flood forecasting studies indicate that LSTM-based models can improve prediction in ungauged watersheds and extend warning lead times, suggesting clear potential for data-sparse CA basins if these models are adapted using regional hydrological observations, snow and cryosphere information, and basin-specific constraints (Nearing et al., 2024).

Finally, there is a need for more consistent scenario practice and greater transparency when dealing with uncertainty in future-oriented research. The existing literature continues to mix methods based on CMIP5 and CMIP6 projections, sometimes inconsistently, with varying approaches to downscaling, bias correction, validation, and ensemble techniques (Nurbatsina et al., 2025b; Gulakhmadov et al., 2025). Increased harmonization of future scenario methodologies, along with clearer reporting of forcing uncertainty, model selection, model calibration assumptions, and ensemble design, would enhance comparability and usefulness for adaptation planning across CA (Fallah et al., 2023, 2024; Yang et al., 2025). This should also include a clearer assessment of how uncertainty is transferred from climate forcing into hydrological and flood risk outputs. For managed basins, uncertainty analysis should consider assumptions about future irrigation demand, reservoir operation, land-use change, and infrastructure development, as these factors can significantly influence downstream responses under the same climatic forcing.

All the priorities listed above are mutually reinforcing, but three stand out as immediate concerns for adaptation planning in CA. Improving hydro-climatic observations, precipitation and snow datasets, cryosphere-specific monitoring, and validation in mountain headwaters and transboundary basins is essential, as these form a basic foundation for flood forecasting, hydrological modeling, and climate-impact assessment. Second, better integration of hydro-climatic processes in adaptation-oriented flood risk assessment is key to making scientific knowledge operational for reservoir management, infrastructure planning, and disaster risk reduction. Third, climatic and anthropogenic drivers should be assessed together, particularly in regulated and irrigated transboundary basins, where runoff generated in climate-sensitive mountain headwaters can be significantly modified downstream by reservoirs, water abstraction, irrigation, land-use change, and development. Advances in these two areas would also facilitate progress in the other research priorities because they would provide a more solid observational basis and a clearer pathway from process understanding to practice.

The present review not only identifies individual research gaps but also allows for a structured evidence-based synthesis that combines hydro-climatic drivers, flood-generating mechanisms, regional patterns, methodological quality, and adaptation relevance within a common interpretative framework. By considering anthropogenic modification alongside climatic forcing, the synthesis also highlights the need to move beyond climate-only interpretations of hydrological change, especially in intensively managed downstream and transboundary systems. The review establishes a basis for making future studies more consistent, enabling cross-basin and cross-country comparisons, and strengthening the development of better and more adaptation-focused flood risk assessment frameworks throughout CA.

Taken together, these priorities indicate that future research should move toward more integrated, better-validated, and more adaptation-oriented assessments of hydrological extremes and flood-related hazards across CA, with a stronger emphasis on compound-process understanding, climate–human attribution, land-use and infrastructure effects, predictive capability, and decision support relevance.

This review shows that hydrological extremes and flood-related hazards across CA are shaped by interacting hydro-climatic and human controls rather than by any single dominant mechanism. Across the reviewed literature, the most robust signals include the intensification of extreme precipitation, declining snow storage, shifts in snowmelt timing, increasing ROS influence, and stronger seasonal redistribution of runoff. However, these signals are expressed unevenly across the region. Lowland and steppe systems are more strongly associated with snowmelt-dominated spring flooding, runoff redistribution, and seasonal water imbalance, whereas mountain and mountain-fed systems are characterized by greater process complexity, including snow-to-rain transitions, shifts in glacier and snow contributions, ROS intensification, flash-flood susceptibility, and greater uncertainty in forcing.

A central conclusion of the review is that the regional evidence base is now much stronger at identifying how hydro-climatic systems are changing than at determining how those changes should be translated into hydrologically grounded flood risk reduction, infrastructure planning, and adaptation strategies. Geographic imbalance, data scarcity in mountain and transboundary settings, uncertainty in forcing, and methodological inconsistency remain major constraints. Future progress will require moving beyond fragmented process-based studies toward integrated, basin-scale, and decision-relevant assessments, supported by more coherent scenario practices, improved high-mountain monitoring, stronger attribution of runoff-regime changes and runoff components, clearer treatment of regulation effects, and closer linkage between hydrological process understanding and flood-relevant planning. Such advances are essential for improving basin-scale flood interpretation, forecasting, water allocation planning, transboundary coordination, and long-term climate adaptation across CA.

In addition to summarizing the most recent literature, the review provides a structured synthesis of the evidence using an evidence-based approach that integrates climate drivers, hydrological processes, flood-generating mechanisms, regional variability, methodological quality, and the relevance of adaptation. Scientific robustness, methodological consistency, and regional representativeness of the available evidence are systematically evaluated to determine scientific priorities for advancing hydrological research in CA and to develop an evidence base for future hydrological research, basin-scale flood risk assessment, transboundary water resource management, and climate change adaptation in CA. The results are intended not only for future scientific research but also for evidence-based decision-making by national water authorities, river basin organizations, infrastructure and urban planners, and policymakers for disaster risk reduction and climate adaptation.