Flooding is a frequent occurrence in the coastal areas of Nigeria. The causes of the floods are not fully understood and have been attributed to several factors such as sea level rise, poor drainage systems, and increasing precipitation. This study projects expected changes in sea surface temperature (SST) and sea level rise (SLR) at the coast of Nigeria. The HadISST1 of the Met Office Hadley Center observation SST data version 1 at 1° × 1° and the European Space Agency’s ESACCI version 2.0 at 0.25° × 0.25° between 1993 and 2005 together with the historical Coupled Model Inter-Comparison Project Phase 6 (CMIP6) were used for the bias correction of the projected CMIP6 of the MIROC5 global climate model (GCM) using linear scaling method. The differences between the projected and the historical SST and SLR at the study area for the annual and the seasonal periods for the near future (2021 – 2060) and the far future (2061 – 2100) under SSPs 126, 245, 370 and 585 shows that under SSP 126 during the near future, changes in temperature are expected to range between 0.45 – 0.7°C while it ranged between 0.69 and 1.08°C during the far future. SLR are expected to range between 2.05 and 15.79mm in the near future and between 17.36 and 32.02mm during the far future. The SSTs and the SLRs changes continually increased from SSP 126 to SSP 585. The highest expected changes during the annual were for SSP 585 during the future with changes in SST ranging between 1.8 and 2.65°C and SLR ranging between 34.79 and 63.62mm. Expected changes in SST are lower near shore than at the outer shore. However, expected changes in SLR are higher near shore where the SSTs were lower.

Changing climatic conditions have escalated the risk of compound disaster. An integrated research framework is proposed in this study to quantitatively analyze and assess the risk of future compound flood in the Xijiang River Basin based on external driving factors and internal variables. Under this framework, a multi-model ensemble of 10 preferred CMIP6 GCMs is carried out based on statistical downscaling and Bayesian weighted average method, and the multi-scale variation characteristics of precipitation and runoff during 2020~2099 are analyzed based on the ensemble data. Combined with univariate and multivariate trend analysis considering intra-correlation, the multi-class Copula functions are utilized to estimate the joint probability and return period of compound flood. The results show that: 1) The precipitation and runoff increase by 8.25%, 14.5%, and 34.05%, 55.18%, respectively, comparing to the baseline period under SSP2-4.5 and SSP5-8.5, with both displaying increasing trends at rates of 1.03%/10a, 2.66%/10a, and 2.74%/10a, 4.62%/10a on the interdecadal scale under the two scenarios, respectively. 2) The internal variables of the compound flood represented by the annual maximum peak flow (AMPF) and the annual consecutive maximum 7-day flood volume (AM7dFV) show significant increasing trends under the two scenarios. However, the annual maximum precipitation (AMPre) of the external driving factor does not show a significant trend, while the annual total precipitation (ATPre) of the external driving factor increases significantly under both scenarios. It is noteworthy that both the internal variables and the external driving factors of compound flood show significant increases in the multivariate analysis. 3) The joint variable of compound flood demonstrates a substantially increasing trend under both scenarios, along with an increase in the magnitude of the once-in-a-century flood. If the intra-correlation between multivariate is discounted, the degree of disaster would be underestimated.

This study investigates climate change impacts on flood-triggering mechanisms in Japan using hyper-resolution (150 m, hourly) hydrodynamic simulations from the 5-km d4PDF climate ensemble and the Rainfall–Runoff-Inundation model. Flood events are categorized into typhoon-induced, June–July non-typhoon (NT67, linked to Meiyu-Baiu fronts), and August–September non-typhoon (NT89, autumn rain fronts) events across historical, +2°C, and +4°C warming stages. Historically, typhoons caused ~70% of floods over Japan. At +2°C, typhoon contributions vary (±10%), with notable declines in northern Kanto. By +4°C, northern Japan experiences a 10–20% reduction in typhoon-driven floods. NT67 events, historically responsible for 10–20% of floods (peaking at 60–70% in western Kyushu and parts of Tohoku due to Meiyu-Baiu rainfall), are projected to increase by ~10% nationwide, with amplified growth (+4°C stage) in Hokuriku and northern Tohoku. NT89 events, historically accounting for ~20% of floods (reaching 50% in western Chubu, northwestern Tohoku, and western Hokkaido), are expected to rise by ~20% in northern Japan, particularly along the Pacific coast. The findings highlight a transition in dominant flood drivers: declining typhoon influence contrasts with growing contributions from frontal rainfall systems. This shift increases the complexity of flood mechanisms, challenging prediction accuracy and mitigation strategies. Regional disparities underscore the need for adaptive measures tailored to localized trends, such as enhanced preparedness for non-typhoon floods in northern and Pacific coastal areas. The study provides critical insights for updating flood risk frameworks and informing climate-resilient planning in Japan.

Drought is classified using various indices, such as meteorological, agricultural, and hydrological. However, the use of multiple indices often leads to confusion in water resource management. There is a significant need for a unified index that consolidates multiple dimensions of drought into one framework. This study addresses this gap by developing composite drought indices using autoencoders (ACDIs). Initially, a basic autoencoder model generated an ACDI, revealing an inverse relationship with observed drought indices. This relationship was corrected by multiplying the output by a negative value. Furthermore, the hyperbolic tangent function was selected over the sigmoid activation function due to its superior sensitivity to drought conditions. To improve the accuracy of the generated drought indices, positive and unity constraints were imposed on the weights, resulting in the ACDI-C model. Further simplification led to the sACDI model by excluding the decoding module, deemed redundant for this study. Next, the weights derived from multiple sites across the country were consolidated into a unified weight, applied consistently across all sites. This process culminated in the development of the sACDI1 model. The performance of the models was evaluated using RMSE, MAE, and correlation coefficients. The results demonstrated the superior performance of both the sACDI and sACDI1 models relative to alternative models. Specifically, these simplified models exhibited significantly lower RMSE and MAE values, indicating enhanced predictive accuracy. The sACDI1 model displayed a considerably lower MAE compared to its counterparts. In addition to evaluating predictive accuracy, alarm performance was assessed through key metrics, including the false alarm ratio, probability of detection, and accuracy. The analysis revealed that the simplified models, particularly sACDI1, outperformed the alternatives in alarm accuracy, especially with the SRSI(A) index. Overall, the ACDIs developed in this study provide an efficient method for synthesizing multiple aspects of drought into a single metric.

Anthropogenic climate change is expected to alter the hydrological cycle, which necessitates the implementation of adaptation measures to maintain current levels of water risk, particularly concerning flooding and water scarcity. However, there has been limited exploration of the specific adaptation measures that should be implemented. In this study, we developed three adaptation portfolios, each composed of a combination of measures – reforestation, dam construction, dam capacity enhancement, changes in dam operation rules, flood channels, retention areas, and improved irrigation efficiency – to mitigate future floods and droughts using explicit basin-wide modeling in the Chao Phraya River Basin (CPRB), Thailand, employing the H08 macro-scale model under both (i) historical (1980–1999) and (ii) future (2080–2099) climate scenarios. Future climate scenarios were derived from the bias-corrected outputs of three general circulation models under RCP4.5 and RCP8.5. Of the three portfolios, two were successful in mitigating future water scarcity to current levels but did not eliminate extreme flooding. The third portfolio, which included basin-wide reforestation, substantially reduced future flooding days but increased the number of drought months to 3–6 months per year due to a 7–11% increase in evapotranspiration. Another finding was that the basin could adapt to more than half of the required adaptation to maintain current water risk levels under RCP4.5. However, under the extreme warming scenario of RCP8.5, the adaptation potential dropped to nearly 30%. This reduced adaptability was primarily driven by the challenges in managing flooding, whereas droughts were more manageable. Overall, this study serves as a warning that even highly regulated river systems, like the CPRB with extensive infrastructure, may struggle to adapt to future floods, a challenge that also affects moderately regulated and pristine rivers. Additionally, adaptation effects varied significantly across sub-basins, underscoring the need for detailed spatio-temporal impact modeling that incorporates multiple adaptation measures.

Extreme climate events, particularly droughts, have become increasingly frequent and unpredictable due to climate change, leading to significant economic losses and adverse impacts on agricultural production. This study introduces a novel Agricultural Water Deficit Index (AWDI) for detecting agricultural drought. AWDI has the advantages of 1) considering agricultural drought directly from the angle of crop water deficit; 2) including the effects of local irrigation water supply; 3) comparable on spatiotemporal scale. In comparison to the widely used agricultural drought indices, the Soil Moisture Anomaly Index (SMA) and the Vegetation Health Index (VHI), AWDI shows a stronger correlation with SMA than VHI. Approximately 63% of the study area exhibits correlation values between 0.3 to 0.6, and 10% shows values between 0.6 and 1.0. Additionally, the performance comparison during drought events indicates a higher consistency between AWDI and SMA than between VHI and SMA. Moreover, AWDI shows a higher average correlation with crop yield anomalies across seven provinces in the basin, underscoring its reliability as a tool for monitoring agricultural drought. This research highlights the importance of incorporating local water supply information into agricultural drought assessments, offering valuable insights for improving agricultural drought detecting.

Rapid/flash droughts, driven by climate warming, lead to the swift and severe drying of the land, with widespread concerns about their damaging effects on the nation's economy. During these periods, reduction in precipitation deficits is frequently followed by abrupt increases in evaporative demand, which causes significant drops in soil moisture and discernible effects on agricultural production and the environment. Several recent publications have emphasized the need for a better understanding of rapid drought conditions to manage their effects effectively. However, progress in assessing these conditions has been hindered by the absence of consistent definitions. There are several factors and climatic forces that are typically connected to the development of flash droughts, thus it's conceivable that no one definition will fully encapsulate the phenomenon. But it's imperative to ensure that flash droughts which lasts for short period of time can be recognized and differentiated from more traditional or longer duration drought occurrences due to their quick onset, quick intensification, and severe nature. With the increasing use of rapid /flash drought term within the research community, this study explores the extent to which pentad-scale precipitation series across India can be used to represent historical flash droughts (1951 to 2022), providing a simple framework for the phenomenon. The results categorize rapid/flash droughts at various hotspot locations across India, which are linked to acute precipitation deficits, a key meteorological variable. The findings of this study can be further utilized in the accurate prediction of flash/rapid drought with the robust evidence from precipitation series in identifying flash drought episodes across the nation. As a result, our findings suggest that continuous monitoring of rapid drought conditions and their drivers is essential for effective preparedness. Additionally, it is necessary to observe the variation in other hydro-meteorological variables during the flash drought period.

Droughts cause water shortages and have a significant impact on vegetation, with the duration of the drought playing a crucial role. Short-term droughts allow vegetation to recover, but prolonged droughts can cause irreversible damage. This study examines how vegetation responds to different drought durations in the southern part of South Korea from 2014 to 2023 using simulated soil moisture, leaf area index (LAI), and gross primary production (GPP) from the Weather Research and Forecasting Model Hydrological modeling system (WRF-Hydro) and remotely sensed solar-induced chlorophyll fluorescence (SIF) from Global OCO-2 SIF (GOSIF), soil moisture from the European Space Agency Climate Change Initiative (ESA-CCI), and evapotranspiration (ET) from the Moderate Resolution Imaging Spectroradiometer (MODIS). Droughts were assessed based on soil moisture changes, with classification based on drought onset duration as flash droughts (<6 pentads), general droughts (7-12 pentads), and creep droughts (>12 pentads). The results show that general droughts led to an initial decline in vegetation indices such as LAI and GPP, but these indices began to increase as the drought ended, indicating a recovery aligned with the termination of drought conditions. In contrast, creep droughts resulted in greater declines in vegetation indices and significantly slower recovery. As drought duration increased, vegetation recovery declined, with prolonged drought leading to potential vegetation mortality. Understanding how different types of droughts influence vegetation and ecosystem dynamics is crucial for predicting ecological responses to future climate variability.AcknowledgementsThis work was supported by the Basic Science Research Program through the National Research Foundation of Korea, which was funded by the Ministry of Science, ICT & Future Planning (RS-2024-00457308), supported by Korea Environment Industry & Technology Institute (KEITI) through Water Management Program for Drought, funded by Korea Ministry of Environment (MOE)(RS-2023-0023194).

Climate change is expected to cause regional deviations from historical climate patterns. To compare past and future conditions accurately, climatological adjustments are necessary. This study explores the sensitivity of future drought characteristics to climatological adjustments in Türkiye, using 56 GCM-RCM pairs at 0.11° spatial resolution from the CMIP5 RCP8.5 scenarios obtained from the CORDEX EUR-11 domain datasets. The Standardized Precipitation Index (SPI) was employed to assess drought variability, and an additive adjustment was applied to minimize differences between reference and future climatological periods. The results indicate that long-term droughts are projected, particularly towards the end of the 21st century (2061-2100). The projected average drought severity is -1.94 (severe drought) with an average drought duration reaching 458 months over the study period. From 1970 to 2060, drought events occurred approximately 35-45% of the time, while this frequency is projected to rise to 82% between 2061 and 2100. The analysis reveals that droughts will intensify during the near future (2020-2060), with increased length and severity, while the last four decades of the century (2061-2100) will experience drought conditions with significantly longer durations and higher severity than seen in the past century. This study highlights the significant impact of climate change on drought characteristics across Türkiye, where no basin will exhibit the same drought patterns by the end of the century as observed historically.

The increasing frequency and severity of global droughts present critical challenges to agriculture and climate resource management, highlighting the need for consistent and comprehensive drought data. To address this, we introduce the Near-global Combined Drought Monitoring (NEC-DROMO) dataset, which integrates key agro-climatological variables to provide robust insights into drought conditions. Spanning 2002 to 2021 at a spatial resolution of 0.25 degrees, NEC-DROMO combines soil moisture (SM) and vegetation water content (VWC) to capture agricultural stress, along with rainfall and temperature, to provide a climatological context. NEC-DROMO delivers monthly drought data for near-global regions.
In developing the NEC-DROMO, agro-climatological variables were integrated using Principal Component Analysis (PCA) to assign weights to each input variable, making it a robust tool for monitoring and assessing both historical and current drought conditions. Validation against independent satellite-based and in-situ methods, including the Standardized Precipitation Index and Geocoded Disaster dataset, confirm NEC-DROMO’s reliability in accurately capturing global drought patterns. Building upon the ECoHydrological Land Reanalysis (ECHLA) dataset and using its VWC, SM, and temperature data, the NEC-DROMO represents a significant advancement in global drought monitoring, supporting diverse applications such as policymaking, food security planning, and disaster risk assessment.

Flood detention basin can reduce the downstream flood risk effectively by temporarily storing floodwater, which has been applied in many countries. However, as the economic development within detention basins, it has been challenging to balance the safety of downstream flood protection area and the economic losses of detention basins. This calls for a fundamental theory and quantitative rules for determining the flood discharge diverted into detention basins, as well as the remaining flood discharge to downstream. This study develops an optimization model to determine the optimal operation of detention basins. We analytically derive the economic principles for optimal flood discharge allocation between the detention basin and downstream protection area under different hydrological and economic scenarios. The derived risk-adjusted safety threshold (RAST) of downstream protected areas is a key control indicator in determining whether flood diversion is needed. Based on these economic principles, optimal rules are proposed for real time operation. Using a real case of Haihe River Basin in China, we verify that the proposed economic principles and optimal rules can reduce the total expected economic losses of the detention basin and downstream protection area. Furthermore, the study analyzes the impacts of key parameters, including economic losses, physical constraints, forecasting uncertainty, on the optimal operation effectiveness. The proposed economic principles and optimal rules for utilizing flood detention basin can offer valuable insights for addressing similar issues in flood risk management.

Changing climatic conditions have escalated the risk of compound disaster. An integrated research framework is proposed in this study to quantitatively analyze and assess the risk of future compound flood in the Xijiang River Basin based on external driving factors and internal variables. Under this framework, a multi-model ensemble of 10 preferred CMIP6 GCMs is carried out based on statistical downscaling and Bayesian weighted average method, and the multi-scale variation characteristics of precipitation and runoff during 2020~2099 are analyzed based on the ensemble data. Combined with univariate and multivariate trend analysis considering intra-correlation, the multi-class Copula functions are utilized to estimate the joint probability and return period of compound flood. The results show that: 1) The precipitation and runoff increase by 8.25%, 14.5%, and 34.05%, 55.18%, respectively, comparing to the baseline period under SSP2-4.5 and SSP5-8.5, with both displaying increasing trends at rates of 1.03%/10a, 2.66%/10a, and 2.74%/10a, 4.62%/10a on the interdecadal scale under the two scenarios, respectively. 2) The internal variables of the compound flood represented by the annual maximum peak flow (AMPF) and the annual consecutive maximum 7-day flood volume (AM7dFV) show significant increasing trends under the two scenarios. However, the annual maximum precipitation (AMPre) of the external driving factor does not show a significant trend, while the annual total precipitation (ATPre) of the external driving factor increases significantly under both scenarios. It is noteworthy that both the internal variables and the external driving factors of compound flood show significant increases in the multivariate analysis. 3) The joint variable of compound flood demonstrates a substantially increasing trend under both scenarios, along with an increase in the magnitude of the once-in-a-century flood. If the intra-correlation between multivariate is discounted, the degree of disaster would be underestimated.

The formation of floods, as a complex physical process, exhibits dynamic changes in its driving factors over time and space under climate change. Due to the black-box nature of deep learning, its use alone does not enhance understanding of hydrological processes. The challenge lies in employing deep learning to uncover new knowledge on flood formation mechanism. This study proposes an interpretable framework for deep learning flood modeling that employs interpretability techniques to elucidate the inner workings of a peak-sensitive Informer, revealing the dynamic response of floods to driving factors in 482 watersheds across the United States. Accurate simulation is a prerequisite for interpretability techniques to provide reliable information. The study reveals that comparing the Informer with Transformer and LSTM, the former showed superior performance in peak flood simulation (NSE over 0.6 in 70% of watersheds). By interpreting Informer’s decision-making process, three primary flood-inducing patterns were identified: precipitation, excess soil water, and snowmelt. The controlling effect of dominant factors is regional, and their impact on floods in time steps shows significant differences, challenging the traditional understanding that variables closer to the timing of flood event occurrence have a greater impact. Over 40% of watersheds exhibited shifts in dominant driving factors between 1981-2020, with precipitation-dominated watersheds undergoing more significant changes, corroborating climate change responses. Additionally, the study unveils the interplay and dynamic shifts among variables. These findings suggest that interpretable deep learning, through reverse deduction, transforms data-driven models from merely fitting nonlinear relationships to effective tools for enhancing understanding of hydrological characteristics.

Forecasting drought events with long-term persistence has remained a complex challenge due to the lack of a sustained relationship between precipitation, the primary driving force of drought, and climate variables over extended periods. A promising approach involves leveraging global climate variables, such as sea surface temperature (SST) and mean sea level pressure (MSLP), to forecast precipitation patterns, enhancing the accuracy of drought predictions. This study aimed to develop models for forecasting accumulated seasonal precipitation (ASP) in South Korea, utilizing globally gridded climate variables, including SST and MSLP. A principal challenge in constructing a robust forecast model is managing the numerous predictors derived from globally gridded climate data, particularly when the observed response variable (ASP) is constrained by limited historical records. To address this issue, two methodologically distinct models, including Least Absolute Shrinkage and Selection Operator (LASSO) and Random Forest (RF), were tested and compared. The LASSO model uses regularization to select relevant predictors by shrinking less informative regression coefficients to zero, while the RF model, an ensemble learning approach, constructs multiple decision trees that collectively incorporate all potential predictors, enabling a flexible data representation. Globally gridded climate variables, including the Difference of Climate Index (DfCI), derived by subtracting the climate variable values of one location from those of another, SST, MSLP, and their combined interactions, were employed as candidate predictors. The final set of predictors was determined through cross-correlation thresholding for each individual season, ensuring the inclusion of the most statistically relevant variables for forecasting. The results indicated that the LASSO model, incorporating combined MSLP-SST predictors, consistently outperformed other configurations, yielding the best forecasting performance across all seasons. This study offers insights for future research, particularly in the application of globally gridded climate variables to hydrological and environmental data analysis, where available observational records are limited.