Accurate subseasonal to seasonal (S2S) forecasts of Indian Summer Monsoon Rainfall (ISMR) are vital for agricultural planning, water resource management, and disaster risk reduction. Conventional post-processing techniques for S2S General Circulation Model (GCM) forecasts predominantly rely on linear methods, which often exhibit limited predictive skill. In this study, we investigate the use of deep learning–specifically, U-Net convolutional neural networks (CNNs)–for improving probabilistic ISMR forecasts. We apply U-Net-based bias correction to three state-of-the-art S2S GCMs: GEFSv12, ECMWF, and IITM ERPv2, training the U-Net models to calibrate tercile probabilities of weekly accumulated rainfall at lead times of 1 to 4 weeks for the monsoon season (June-September) over India. Our results show that the U-Net enhances probabilistic forecast skill, consistently outperforming the baseline Extended Logistic Regression (ELR) approach. Using hindcast data from 1989 to 2022, we observe a statistically significant improvement in weeks 3–4 probabilistic forecast skill, measured by the Ranked Probability Skill Score (RPSS). The extent of improvement, however, varies with the amount of training data available for each GCM, and can reach a 2 p.p. increase, from 2

Forecasting the impacts of climate extremes is challenging, but critical to a range of sectors, including agriculture, water management, public health and safety, infrastructure, energy, national defense, and ecology. Foundational to these forecasts are hindcast model archives, which are routinely used for applications produced for the private sector and agencies, including NASA, NOAA, US Department of State, and the US Department of Defense. Forecast and hindcast archives underpin scientific inquiry funded by these agencies, particularly in relation to forecasting weather and climate extremes, as well as the US NSF. In this article, we catalog the sector-specific decision support systems that depend upon hindcast archives and survey the current state of hindcast archive infrastructure. We find that despite the tremendous amount of investment and dependent decision support systems, the United States hindcast archive is relatively fragile and underfunded, especially when compared with the Copernicus system in Europe. We conclude with recommendations for improving hindcast archive infrastructure to support routine sector-specific applications and improve resilience to climate extremes.

Subseasonal predictions from 2 weeks to 2 months have made significant advancements over the past decade, driven by progress in physical understanding, climate modeling, computational capabilities, and artificial intelligence (AI). These predictions are increasingly in demand due to their potential to provide stakeholders with adequate lead time for effective disaster adaptation, mitigation, and resource management. However, there remain critical gaps in the engagement between prediction providers and service users. Providers often lack insight into the specific needs of users and do not have transferrable strategies to build trust through tailored evaluations and clear confidence levels, which often results in repeatedly devising approaches for each provider–user interaction. Further, users frequently struggle to interpret predictions and are hesitant to make decisions based on these uncertain outcomes. This paper attempts to make “last-mile efforts” by reviewing relevant literature, operational systems, and the most informative communications and engagement strategies with key sectors. It proposes a preliminary framework to standardize the approach for provider–user interaction in the context of subseasonal prediction and services, with potential applicability and extension to seamless prediction systems in the future. Lastly, we underscore future directions for subseasonal predictions, emphasizing the integration of dynamic climate modeling and AI-driven enhancements with large ensemble techniques to improve both reliability and confidence. This review is part of the United Nations Educational Scientific and Cultural Organization (UNESCO) International Decade of Sciences for Sustainable Development (2024–33) and contributes to the Seamless Prediction and Services for Sustainable Natural and Built Environment (SEPRESS) Program (2025–32), an initiative endorsed under this global framework.

Numerical seasonal predictions provide information about the expected climate conditions for the forthcoming seasons. It is common practice every month in operational centers to produce and issue numerical seasonal predictions for the following 3-month season. These predictions are relevant for various socioeconomic activities such as water resources management, agricultural planning, and hydro-electricity production, to name a few. Precipitation and near-surface air temperature are the most important variables of interest impacting these sectoral activities. The main climate driver of seasonal precipitation anomalies over South America is El Niño–Southern Oscillation, a phenomenon that manifests through ocean-atmosphere interactions in the tropical Pacific, with remote effects observed through atmospheric teleconnections affecting tropical and southeastern South America. Another important driver of climate anomalies is the tropical Atlantic inter-hemispheric meridional sea surface temperature anomaly gradient (also referred to as the tropical Atlantic dipole). This feature modulates the position of the Intertropical Convergence Zone over the tropical Atlantic and therefore has impacts on precipitation over northern and northeastern South America. Developing and operating climate models able to reproduce these phenomena along with the associated teleconnections is key for successfully predicting the seasonal climate over South America. Efforts have been undertaken by the Center for Weather Forecast and Climate Studies (CPTEC) of the National Institute for Space Research, in collaboration with the National Institute of Meteorology (INMET) and Ceará Institute for Meteorology and Water Resources (FUNCEME) in Brazil, and by the International Research Institute for Climate and Society (IRI) of Columbia Climate School, Columbia University in the United States, for developing the capabilities and producing skillful numerical seasonal predictions for South America. Over the years, the scientific community has recognized the importance of (a) combining ensemble predictions from different models to better sample uncertainties in initial conditions and model formulation, and (b) applying procedures for calibrating predictions using historical (past) observations and the corresponding retrospective model predictions. Activities on multi-model ensemble seasonal predictions have been led by CPTEC and IRI for producing well-calibrated predictions for South America, including seasonal precipitation forecasts and retrospective prediction performance. The research developed and predictions produced by CPTEC (in collaboration with INMET and FUNCEME) and by IRI contribute to international activities such as the Global Framework for Climate Services and Regional Climate Outlook Forums (RCOFs) organized in South America under the auspices of the World Meteorological Organization (WMO). As a WMO-designated Global Producing Centre for Seasonal Prediction, CPTEC generates seasonal prediction products with global coverage each month and makes them available, with the corresponding retrospective prediction performance products, through the WMO Lead Centre for Seasonal Prediction Multi-Model Ensemble for use by National Meteorological and Hydrological Services, Regional Climate Centres, and RCOFs. The scientific-based climate predictions produced by CPTEC, INMET, FUNCEME, and IRI support climate-sensitive decision making and response actions in agriculture, food security, disaster risk reduction, energy, health, and water management.

The desert climate of the Arabian Peninsula (AP), marked by sparse rainfall, extreme temperatures, and frequent dust events, significantly impacts its 80-million population, environment, and economy. Rising temperatures and dust incursions exacerbate these harsh conditions, yet the AP's climate is underrepresented in global climate research. Understanding its variability is crucial for improving predictions on subseasonal-to-seasonal timescales and for developing reliable climate change projections. Existing climate models fail to capture the region's unique environment, topography, and land-use changes, leading to poor representation of key processes like local convection, aridity, and moisture transport. To address these gaps, Saudi Arabia established the Climate Change Center (CCC) in 2022, part of the Saudi Vision 2030 initiative. The CCC aims to study climate variability and project future changes using advanced Earth system models developed in collaboration with international partners. This study presents the CCC's roadmap, focusing on its relevance for global climate research and policymaking, including the Saudi and Middle East Green Initiatives. We also discuss regional uncertainties in the IPCC's climate projections for the AP and highlight the development of high-resolution regional models that account for local atmospheric, land, and oceanic processes. The CCC is developing subseasonal-to-seasonal forecasting systems and drought monitoring tools, alongside user-friendly dashboards to offer stakeholders customized climate data. These tools, set for launch in 2025, will aid informed decision-making in addressing extreme weather events and climate-related challenges in Saudi Arabia.

Seasonal drought forecasting is critical for the hyper-arid, water-stressed Arabian Peninsula (AP), where droughts have become more frequent and intense as manifestations of long-term regional aridification. The region depends heavily on cold-season rainfall to sustain agriculture and livestock, making rainfall deficits particularly disruptive in vulnerable areas. Here, forecasts from the North American Multi-Model Ensemble (NMME) and observations are used to produce Standardized Precipitation Index (SPI) data over the AP for 1991–2020, focusing on six overlapping 3-month wet seasons from October–December to March–May. Covariability between AP SPI and Pacific sea surface temperatures (SSTs) is examined using singular value decomposition. The leading mode represents approximately 27–38% of the shared and 37–57% of the individual variability and is linked to ENSO-related SST anomalies. Leading SVD SPI pattern is associated with well-organized 500 mb height anomalies. NMME forecast skill for the 3-month SPI was evaluated across multiple lead times, revealing pronounced spatial and temporal variability through both deterministic and probabilistic metrics. The models demonstrate skill linked to ENSO teleconnections, particularly in La Niña-linked droughts. Deterministic forecast skill, measured through anomaly correlation, is highest at the shortest lead times, where combining observed data with model forecasts provides the greatest added value, and declines sharply at longer leads. Skill is higher in early wet-season periods (October–December, November–January, and December–February) and decreases in later periods, with March–May showing the lowest skill relative to persistence forecasts. Assessments based on reliability diagrams and ranked probability skill score also reveal that probabilistic forecast skill is strongest during the early wet season, decreases in subsequent periods, and is minimal in March–May. These findings suggest that NMME-based seasonal forecasts could offer useful early warning information for drought across the AP, especially at short lead times and during ENSO-active periods.

Numerical subseasonal predictions provide information about the expected meteorological conditions up to about 4 to 6 weeks ahead, although predictions for any period less than a season (3–4 months) are considered subseasonal , fulfilling the gap between numerical weather predictions (for the forthcoming days) and numerical seasonal predictions (for the forthcoming months to seasons). Numerical subseasonal predictions are issued at least once per week, and are relevant for socioeconomic applications, such as in water resources management, agricultural planning, and hydroelectricity production, including in South America. The variables of greatest interest, having major impacts in these sectors, are precipitation and near-surface temperature. An important driver of subseasonal variability over South America is the Madden-Julian oscillation (MJO), a large-scale phenomenon characterized by anomalous convective activity starting over the equatorial Indian Ocean. The MJO travels eastward generating a wave-number-one-type tropical pattern, where half of the globe experiences favorable conditions for convection and the other half experiences unfavorable conditions. A full MJO cycle around the globe takes about 30 to 60 days to be completed and is divided into 8 phases. Some MJO phases contribute favorably while other phases contribute unfavorably for the organization of clouds and the associated precipitation over South America. Another relevant subseasonal feature over South America is the dipole between the southeastern South America region and the central-east portion of Brazil, the latter located over the region where the South Atlantic Convergence Zone (SACZ) manifests. When the SACZ is active (inactive), the central-east portion of Brazil experiences both enhanced (reduced) convection and precipitation, while southeastern South America, at the opposite pole of the dipole, experiences inhibited (enhanced) convection and reduced (increased) precipitation. Developing and operating climate models able to reproduce such phenomena, and generating and making available the associated prediction products, is key for enabling the issuance of subseasonal predictions over South America. Efforts have been undertaken by the Center for Weather Forecast and Climate Studies (CPTEC) of the National Institute for Space Research (INPE), in collaboration with the Ceará Institute for Meteorology and Water Resources (FUNCEME) in Brazil, and by the International Research Institute for Climate and Society (IRI) of the Columbia Climate School, Columbia University in the United States, to develop capabilities and produce numerical subseasonal predictions for South America. This includes the development of the first generation of subseasonal predictions produced with the CPTEC global model and activities on multimodel ensemble subseasonal predictions led by CPTEC, FUNCEME, and IRI for producing well-calibrated predictions for South America.

We explore the relationships amongst duration, total amount, mean and maximum hourly intensity of rainfall and spatial scale for more than a thousand rain gauges covering a diverse range of tropical and subtropical (30°N-30°S) climates, from arid (< = 200 mm year⁻1) to very wet (> 3000–4000 mm year⁻1). We find that the interannual variation of seasonal (3-month) amounts is primarily driven by wet hour frequency rather than the mean intensity of rainfall. A total of 3.5 million local wet spells (WS: consecutive wet hours with at least 1 mm of rain) are then systematically analyzed. WS lasting 5 h or less account for 80

California experienced a historic run of nine consecutive landfalling atmospheric rivers (ARs) in three weeks' time during winter 2022/23. Following three years of drought from 2020 to 2022, intense landfalling ARs across California in December 2022-January 2023 were responsible for bringing reservoirs back to historical averages and producing damaging floods and debris flows. In recent years, the Center for Western Weather and Water Extremes and collaborating institutions have developed and routinely provided to end users peer-reviewed experimental seasonal (1-6 month lead time) and subseasonal (2-6 week lead time) prediction tools for western U.S. ARs, circulation regimes, and precipitation. Here, we evaluate the performance of experimental seasonal precipitation forecasts for winter 2022/23, along with experimental subseasonal AR activity and circulation forecasts during the December 2022 regime shift from dry conditions to persistent troughing and record AR-driven wetness over the western United States. Experimental seasonal precipitation forecasts were too dry across Southern California (likely due to their overreliance on La Nina), and the observed above-normal precipitation across Northern and Central California was underpredicted. However, experimental subseasonal forecasts skillfully captured the regime shift from dry to wet conditions in late December 2022 at 2-3 week lead time. During this time, an active MJO shift from phases 4 and 5 to 6 and 7 occurred, which historically tilts the odds toward increased AR activity over California. New experimental seasonal and subseasonal synthesis forecast products, designed to aggregate information across institutions and methods, are introduced in the context of this historic winter to provide situational awareness guidance to western U.S. water managers.

This paper provides an updated assessment of the “International Research Institute for Climate and Society's (IRI) El Niño Southern Oscillation (ENSO) Predictions Plume". We evaluate 247 real-time forecasts of the Niño 3.4 index from February 2002 to August 2022 and examine multimodal means of dynamical (DYN) and statistical (STAT) models separately. Forecast skill diminishes as lead time increases in both DYN and STAT forecasts, with peak accuracy occurring post-northern hemisphere spring predictability barrier and preceding seasons. The DYN forecasts outperform STAT forecasts with a pronounced advantage in forecasts initiated from late boreal winter through spring. The analysis uncovers an asymmetry in predicting the onset of cold and warm ENSO episodes, with warm episode onsets being better forecasted than cold onsets in both DYN and STAT models. The DYN forecasts are found to be valuable for predicting warm and cold ENSO episode onsets several months in advance, while STAT forecasts are less informative about ENSO phase transitions.

WWRP/WCRP S2S Summit 2023 What: More than 190 scientists from 29 countries met to celebrate 10 years of the Subseasonal to Seasonal (S2S) Prediction project and look to the future of S2S prediction. When: 3-7 July 2023 Where: University of Reading, United Kingdom

In 2022, a record-breaking monsoon caused flooding throughout Pakistan, particularly in the southern regions, resulting in deaths, property losses, and severe crop damage, affecting the food supply chain that could last for years. This study assesses the accuracy of sub-seasonal calibrated probabilistic rainfall forecasts for Pakistan. The evaluation focuses on forecasts initialized throughout the summer monsoon season (June–September) and utilizes the European Center for Medium Range Forecast (ECMWF) ensemble prediction system. Forecasts are calibrated using a canonical correlation analysis (CCA) and evaluated using cross-validated hindcasts from 2002 to 2021. The calibrated hindcasts exhibit positive ranked probability skill score and are reliable for weeks 1 (days 1–7), 2 (days 8–14), and 3 – 4 (days 15–28), lead times. In the extraordinary monsoon season of 2022, tercile-category probabilistic forecasts provided useful information up to 4 weeks ahead. Furthermore, the occurrences of intense monsoon rainfall in the highly affected southern region of Pakistan were forecasted reasonably well up to 2 weeks in advance. The ECMWF model's ability to predict sub-seasonal monsoon rainfall in Pakistan during 2022 is attributed to the model’s successful prediction of monsoonal intra-seasonal oscillations.

In South Asia (SA), the boreal summer monsoon (June to September; JJAS) and the El Niño-Southern Oscillation (ENSO) are connected, though different areas in SA respond differently to ENSO. In this paper, a new 41-year (1981 to 2021) high-resolution gridded rainfall dataset (ENACTS-BMD; Enhancing National Climate Services for Bangladesh Meteorological Department) is used to investigate the linkage between the Bangladesh Summer Monsoon Rainfall (BSMR) and ENSO. Observed BSMR shows a weak positive correlation ( r = + 0.21, not statistically significant at the 5% level) with sea surface temperatures (SST) in the central-eastern (Niño3.4) Pacific region. Among the eight El Niño events, seven of them corresponded to above-normal BSMR. However, during the 11 La Niña events, the relationship was more varied, with above-normal BSMR occurring in seven instances. These findings highlight an asymmetric relationship between BSMR and ENSO. Furthermore, BSMR is negatively correlated ( r = − 0.47 statistically significant at the 5% level) with Indian Summer Monsoon Rainfall (ISMR: 75°–85 o E, 18°–30 o N). The potential physical mechanism can be outlined as follows: during El Niño, the Walker circulation tends to be weakened, resulting in a weakening of the summer monsoon circulation, which in turn reduces the intensity of easterly winds along the Bangladesh Himalayan foothills. Subsequently, a lower-level anomalous cyclonic circulation is established, facilitating the convergence of moisture within the boundary layer. This, in turn, leads to intensified rainfall over Bangladesh and the surrounding regions during El Niño. Seasonal forecast models do not adequately capture BSMR and ENSO, BSMR and circulation, and BSMR and ISMR inverse correlations. While the observed BSMR-ENSO relationship is complex and teleconnections are weak, awareness of the inverse relationship with ISMR and the incorrect model behavior could be useful in the context of seasonal BSMR predictions.

The Asian summer monsoon (ASM) has a considerable impact on human lives in the most populated region in the world. Thus, its seasonal prediction is a high-profile application in Earth Science. However, the prediction skill of the regional ASM variability has long been limited due to a formidable difficulty in accurately simulating the complex interactions of the atmosphere-ocean variability and its remote influence on regional climate in numerical models. This study updates the current status and assesses progress in the ASM seasonal prediction performance. This study evaluated the seasonal prediction skill of two generations of models in hindcast data archived by the WCRP Climate-system Historical Forecast Project (CHFP) and Copernicus Climate Change Service (C3S). A special focus was put on the representation of the predominant teleconnections associated with the ENSO and Indian Ocean variability. It was found that the latest seasonal prediction systems (C3S) generally outperform previous-generation systems (CHFP) in terms of the reproducibility of the observed precipitation climatology and the prediction skill of the interannual variability of seasonal precipitation over the ASM region. Furthermore, the results suggested that the improvement of the prediction skill of the ASM likely stems from the improved representation of the monsoon climatology and teleconnections in the models. These analyses highlight the steady progress of the atmosphere-ocean coupled modelling and promise future improvements in the seasonal ASM prediction.

A global multimodel probabilistic subseasonal forecast system for precipitation and near-surface tempera-ture is developed based on three NOAA ensemble prediction systems that make their forecasts available publicly in real time as part of the Subseasonal Experiment (SubX). The weekly and biweekly ensemble means of precipitation and tem-perature of each model are individually calibrated at each grid point using extended logistic regression, prior to forming equal-weighted multimodel ensemble (MME) probabilistic forecasts. Reforecast skill of week-3-4 precipitation and tem-perature is assessed in terms of the cross-validated ranked probability skill score (RPSS) and reliability diagram. The multi -model reforecasts are shown to be well calibrated for both variables. Precipitation is moderately skillful over many tropical land regions, including Latin America, sub-Saharan Africa and Southeast Asia, and over subtropical South America, Africa, and Australia. Near-surface temperature skill is considerably higher than for precipitation and extends into the ex-tratropics as well. The multimodel RPSS skill of both precipitation and temperature is shown to exceed that of any of the constituent models over Indonesia, South Asia, South America, and East Africa in all seasons. An example real-time week-3-4 global forecast for 13-26 November 2021 is illustrated and shown to bear the hallmarks of the combined influen-ces of a moderate Madden-Julian oscillation event as well as weak-moderate ongoing La Nina event.

This paper assesses the skill of the Saudi-King Abdulaziz University coupled ocean–atmosphere Global Climate Model, namely Saudi-KAU CGCM, in forecasting the El Niño-Southern Oscillation (ENSO)-related sea surface temperature. The model performance is evaluated based on a reforecast of 38 years from 1982 to 2019, with 20 ensemble members of 12-month integrations. The analysis is executed on ensemble mean data separately for boreal winter (December to February: DJF), spring (March to May: MAM), summer (June to August: JJA), and autumn (September to November: SON) seasons. It is found that the Saudi-KAU model mimics the observed climatological pattern and variability of the SST in the tropical Pacific region. A cold bias of about 0.5–1.0 °C is noted in the ENSO region during all seasons at 1-month lead times. A statistically significant positive correlation coefficient is observed for the predicted SST anomalies in the tropical Pacific Ocean that lasts out to 6 months. Across varying times of the year and lead times, the model shows higher skill for autumn and winter target seasons than for spring or summer ones. The skill of the Saudi-KAU model in predicting Niño 3.4 index is comparable to that of state-of-the-art models available in the Copernicus Climate Change Service (C3S) and North American Multi-Model Ensemble (NMME) projects. The ENSO skill demonstrated in this study is potentially useful for regional climate services providing early warning for precipitation and temperature variations on sub-seasonal to seasonal time scales.

Accurate lesion segmentation is critical in stroke rehabilitation research for the quantification of lesion burden and accurate image processing. Current automated lesion segmentation methods for T1-weighted (T1w) MRIs, commonly used in stroke research, lack accuracy and reliability. Manual segmentation remains the gold standard, but it is time-consuming, subjective, and requires neuroanatomical expertise. We previously released an open-source dataset of stroke T1w MRIs and manually-segmented lesion masks (ATLAS v1.2, N = 304) to encourage the development of better algorithms. However, many methods developed with ATLAS v1.2 report low accuracy, are not publicly accessible or are improperly validated, limiting their utility to the field. Here we present ATLAS v2.0 (N = 1271), a larger dataset of T1w MRIs and manually segmented lesion masks that includes training (n = 655), test (hidden masks, n = 300), and generalizability (hidden MRIs and masks, n = 316) datasets. Algorithm development using this larger sample should lead to more robust solutions; the hidden datasets allow for unbiased performance evaluation via segmentation challenges. We anticipate that ATLAS v2.0 will lead to improved algorithms, facilitating large-scale stroke research.

There is a high demand and expectation for subseasonal to seasonal (S2S) prediction, which provides forecasts beyond 2 weeks, but less than 3 months ahead. To assess the potential benefit of artificial intelligence (AI) methods for S2S prediction through better postprocessing of ensemble prediction system outputs, the World Meteorological Organization (WMO) coordinated a prize challenge in 2021 to improve subseasonal prediction. The goal of this competition was to produce the most skillful forecasts of precipitation and 2-m temperature globally averaged over forecast weeks 3 and 4 and over weeks 5 and 6 for the year 2020 using artificial intelligence techniques. The top three submissions, described in this article, succeeded in producing S2S forecasts significantly more skillful than the bias-corrected ECMWF operational reference forecasts, particularly for precipitation, through improved calibration of the ECMWF raw forecast outputs or multimodel combination. These forecast improvements should benefit the use of S2S forecasts in applications.