Process-Informed Subsampling Improves Subseasonal Rainfall Forecasts in Central America.

Subseasonal rainfall forecast skill is critical to support preparedness for hydrometeorological extremes. We assess how a process-informed evaluation, which subsamples forecasting model members based on their ability to represent potential predictors of rainfall, can improve monthly rainfall forecasts within Central America in the following month, using Costa Rica and Guatemala as test cases. We generate a constrained ensemble mean by subsampling 130 members from five dynamic forecasting models in the C3S multimodel ensemble based on their representation of both (a) zonal wind direction and (b) Pacific and Atlantic sea surface temperatures (SSTs), at the time of initialization. Our results show in multiple months and locations increased mean squared error skill by 0.4 and improved detection rates of rainfall extremes. This method is transferrable to other regions driven by slowly-changing processes. Process-informed subsampling is successful because it identifies members that fail to represent the entire rainfall distribution when wind/SST error increases.

Global terrestrial drought and its projected socioeconomic implications under different warming targets.

Droughts are increasingly frequent as the Earth warms, presenting adaptation challenges for ecosystems and human communities worldwide. A strategic environmental assessment (SEA) and the integration of adaptation strategies into policies, plans, and programs (PPP) are two important approaches for enhancing climate resilience and fostering sustainable development. This study developed an innovative approach to strengthen the SEA of droughts by quantifying the impacts of future temperature increases. A novel method for projecting drought events was integrated into the SEA process by leveraging multiple data sources, including atmospheric reanalysis, reconstructions, satellite-based observations, and model simulations. We identified drought conditions using terrestrial water storage (TWS) anomalies and applied a random forest (RF) model for disentangling the drivers behind drought events. We then set two global warming targets (2.0 °C and 2.5 °C) and analyzed drought changes under three shared socioeconomic pathways (SSP126, SSP370, SSP585). In a 2.0 °C warming world, over 50 % of the global surface will face increased drought risk. With an additional 0.5 °C increase, >60 % of the land will be prone to further drought escalation. We utilized copulas to build the joint distribution for drought duration and severity, estimating the joint return periods (JRP) for bivariate drought hazard. In tropical and subtropical regions, JRP reductions exceeding half are projected for >33 % of the regional land surface under 2.0 °C warming and for >50 % under 2.5 °C warming. Finally, we projected the impacts of drought events on population and gross domestic product (GDP). Among the three SSPs, under SSP370, population exposure is highest and GDP exposure is minimal under 2.0 °C warming. Global GDP and population risks from drought are projected to increase by 37 % and 24 %, respectively, as warming continues. This study enhances the accuracy of SEA in addressing drought risks and vulnerabilities, supporting climate-resilient planning and adaptive strategies.

Large anomalies in future extreme precipitation sensitivity driven by atmospheric dynamics.

Increasing atmospheric moisture content is expected to intensify precipitation extremes under climate warming. However, extreme precipitation sensitivity (EPS) to temperature is complicated by the presence of reduced or hook-shaped scaling, and the underlying physical mechanisms remain unclear. Here, by using atmospheric reanalysis and climate model projections, we propose a physical decomposition of EPS into thermodynamic and dynamic components (i.e., the effects of atmospheric moisture and vertical ascent velocity) at a global scale in both historical and future climates. Unlike previous expectations, we find that thermodynamics do not always contribute to precipitation intensification, with the lapse rate effect and the pressure component partly offsetting positive EPS. Large anomalies in future EPS projections (with lower and upper quartiles of -1.9%/°C and 8.0%/°C) are caused by changes in updraft strength (i.e., the dynamic component), with a contrast of positive anomalies over oceans and negative anomalies over land areas. These findings reveal counteracting effects of atmospheric thermodynamics and dynamics on EPS, and underscore the importance of understanding precipitation extremes by decomposing thermodynamic effects into more detailed terms.

Constrained CMIP6 projections indicate less warming and a slower increase in water availability across Asia.

Climate projections are essential for decision-making but contain non-negligible uncertainty. To reduce projection uncertainty over Asia, where half the world's population resides, we develop emergent constraint relationships between simulated temperature (1970-2014) and precipitation (2015-2100) growth rates using 27 CMIP6 models under four Shared Socioeconomic Pathways. Here we show that, with uncertainty successfully narrowed by 12.1-31.0%, constrained future precipitation growth rates are 0.39 ± 0.18 mm year-1 (29.36 mm °C-1, SSP126), 0.70 ± 0.22 mm year-1 (20.03 mm °C-1, SSP245), 1.10 ± 0.33 mm year-1 (17.96 mm °C-1, SSP370) and 1.42 ± 0.35 mm year-1 (17.28 mm °C-1, SSP585), indicating overestimates of 6.0-14.0% by the raw CMIP6 models. Accordingly, future temperature and total evaporation growth rates are also overestimated by 3.4-11.6% and -2.1-13.0%, respectively. The slower warming implies a lower snow cover loss rate by 10.5-40.2%. Overall, we find the projected increase in future water availability is overestimated by CMIP6 over Asia.