The accessibility of water resources for human consumption and ecosystems depends on the spatio-temporal distribution of precipitation and evaporation. Changes in these characteristics due to climate change may alter water availability (WA), impacting humans and the biosphere. Previous studies have examined trends in annual mean and seasonal precipitation variations, as well as extreme events and changes in evaporation. However, few studies have examined concomitant changes in both annual mean and seasonal variation of precipitation and evaporation. Global climate classifications rarely consider seasonal variation non-parametrically, despite its complex variation across global land regions. Analyzing collective changes in hydrological annual means and seasonal variations can better inform assessments of societal and ecological vulnerability to future WA changes. For example, increased seasonal precipitation variability might disrupt continuous atmospheric water supply, leading to extended dry periods in regions with unimodal precipitation distribution. In high-precipitation regimes, this could lead to floods. Increased seasonal evaporation variation might also alter the monthly terrestrial water budget. This study examines spatially aggregated future projections for nine distinct regimes, characterizing joint changes in annual mean and seasonal precipitation and evaporation.

Numerous studies have individually investigated trends in annual mean precipitation (6-7), seasonal variations (8-9), and extreme precipitation events (10-11). Similarly, research has explored changes in evaporation characteristics (12-14). The combined monthly distribution of precipitation and evaporation significantly impacts regional hydrology (15-16), crop yields (17-18), and ecology (19-20). However, studies examining the combined effect of changes in both annual means and seasonal variations in precipitation and evaporation are scarce. Existing global climate classifications (21-23) used in WA studies often neglect seasonal variation from a non-parametric perspective, even though it varies significantly across global regions (24-25). This study addresses this gap by employing a non-parametric approach to analyze the combined effects.

The study first classifies global land regions into distinct hydroclimatic regimes based on annual means and seasonal variations of precipitation using data from the Global Precipitation Climatology Centre (GPCC). Seasonality is quantified using apportionment entropy (AE), a non-parametric measure capturing higher-order statistics, unlike parametric methods. Higher AE values indicate lower seasonal variation. Nine regimes are classified based on percentiles (30th and 70th) of AE and annual mean precipitation (1971-2000). Four critical zones (high/low precipitation and high/low AE) illustrate extreme scenarios. Precipitation and evaporation projections from Coupled Model Intercomparison Project Phase 5 (CMIP5) models are aggregated over these nine regimes. Linear trends over the 21st century (2005-2100) are evaluated using Bayesian model averaging (BMA) for annual means and seasonal variation in precipitation and evaporation under three RCP scenarios (2.6, 4.5, and 8.5). The Theil-Sen estimator is used to determine linear trends. BMA weights are estimated using a maximum log-likelihood function and Markov Chain Monte Carlo algorithm. The performance of the BMA multi-model ensemble is assessed using Pearson correlation. Monthly available water (WA) is determined as the difference between precipitation and evaporation. Changes in monthly WA distribution are analyzed, and the role of wet and dry season components in altering WA is investigated.

The analysis reveals an increase in annual mean precipitation in all nine regimes across all RCP scenarios, with the magnitude of the increase proportional to the emission forcing. Similarly, annual mean evaporation increases in all regimes. However, only four of the nine regimes show increased seasonal precipitation variation; in these cases, the increase in variability is particularly pronounced in regimes that already exhibit high seasonal variability. This indicates a pattern of "seasonally variable regimes becoming more variable." Regimes with low seasonality in precipitation experience increased wet-season precipitation. Changes in evaporation seasonality are less pronounced; a decrease in seasonal evaporation variation is observed in several regimes. The projected monthly distribution of WA demonstrates that wet seasons become wetter, especially in regimes with consistent water supply. Regimes with low precipitation exhibit minimal changes in monthly WA. Analyzing seasonal changes in precipitation and evaporation reveals that in regions with high AE (consistent water supply), evaporation increases more than in regions with moderate or low AE, particularly in RCP 8.5. This increase is significant in the wet season but not the dry season. Increases in WA are evident in wet seasons, primarily due to precipitation increases exceeding evaporation increases. High AE regions exhibit greater increases in WA, implying increased flood risk potential.

The study's findings demonstrate that coupled changes in seasonal variation and annual mean precipitation and evaporation significantly impact the spatio-temporal distribution of WA. The increase in precipitation and evaporation across all regimes, particularly under RCP 8.5, aligns with previous studies. The increase in seasonal precipitation variation, especially in already highly variable regimes, is consistent with observations of wetter wet seasons and drier dry seasons. The decrease in seasonal evaporation variation contrasts with the precipitation trends. Increased WA in wet seasons, especially in regions with less seasonal variation, highlights potential implications for flood risk and water resource management. The study emphasizes the interplay between precipitation and evaporation in shaping WA changes and highlights the regime-dependent nature of these changes.

This research reveals significant impacts of coupled changes in seasonal and annual mean precipitation and evaporation on global water availability. A non-parametric analysis framework, classifying global regions into nine hydroclimatic regimes, shows increasing annual precipitation and evaporation across all regions, with increased seasonal precipitation variation in already variable regimes. Wet seasons become wetter, particularly in regions with consistent water supply. This study provides a valuable framework for assessing future water resource management and flood risk, especially considering the observed regional differences. Further research could investigate the implications of these findings for extreme hydrological events and specific water management strategies.

The study uses CMIP5 models, which may not perfectly capture all aspects of climate change impacts on hydrological extremes. The spatial aggregation of data over regimes might obscure regional variations. The focus on precipitation and evaporation as proxies for water availability neglects other factors influencing water resources. Lastly, this study does not directly address the impacts on specific societal or ecological systems.