Climate-driven ecosystem changes are increasingly affecting the world’s ocean ecosystems, necessitating urgent guidance on adaptation strategies to limit or prevent catastrophic impacts. The Fisheries and Marine Ecosystem Model Intercomparison Project (FishMIP) is a network and framework that provides standardised ensemble projections of the impacts of climate change and fisheries on ocean life and the benefits that it provides to people through fisheries. Since its official launch in 2013 as a small, self-organised project within the larger Inter-Sectoral Impact Model Intercomparison Project, the FishMIP community has grown substantially and contributed to key international policy processes, such as the IPCC AR5 and AR6, and the IPBES Global Biodiversity Assessment. While not without challenges, particularly around comparing heterogeneous ecosystem models, integrating fisheries scenarios, and standardising regional-scale ecosystem models, FishMIP outputs are now being used across a variety of applications (e.g., climate change targets, fisheries management, marine conservation, Sustainable Development Goals). Over the next decade, FishMIP will focus on improving ecosystem model ensembles to provide more robust and policy-relevant projections for different regions of the world under multiple climate and societal change scenarios, and continue to be open to a broad spectrum of marine ecosystem models and modellers. FishMIP also intends to enhance leadership diversity and capacity-building to improve representation of early- and mid-career researchers from under-represented countries and ocean regions. As we look ahead, FishMIP aims to continue enhancing our understanding of how marine life and its contributions to people may change over the coming century at both global and regional scales.

A document by Andy Baker. Click on the document to view its contents.

The Regional Climate Modeling system (RegCM) has undergone a significant evolution over the years, leading for example to the widely used versions RegCM4 and RegCM4-NH. In response to the demand for higher resolution, a new version of the system has been developed, RegCM5, incorporating the non-hydrostatic dynamical core of the MOLOCH weather prediction model. In this paper we assess the RegCM5’s performance for 5 CORDEX-CORE domains, including a pan-European domain at convection-permitting resolution. We find temperature biases generally in the range of -2 to 2 degrees Celsius, higher in the northernmost regions of North America and Asia during winter, linked to cloud water overestimation. Central Asia and the Tibetan Plateau show cold biases, possibly due to sparse station coverage. The model exhibits a prevailing cold bias in maximum temperature and warm bias in minimum temperature, associated with a systematic overestimation of lower-level cloud fraction, especially in winter. Taylor diagrams indicate a high spatial temperature pattern correlation with ERA5 and CRU data, except in South America and the Caribbean region. The precipitation evaluation shows an overestimation in South America, East Asia, and Africa. RegCM5 improves the daily precipitation distribution compared to RegCM4, particularly at high intensities. The analysis of wind fields confirms the model’s ability to simulate monsoon circulations. The assessment of tropical cyclone tracks highlights a strong sensitivity to the tracking algorithms, thus necessitating a careful model interpretation. Over the European region, the convection permitting simulations especially improve the diurnal cycle of precipitation and the hourly precipitation intensities.

Climate change is expected to influence the frequency and severity of biological invasions in a variety of ways, including creating novel introduction pathways, decreasing the resilience of native habitats, inducing range shifts and expansions, and altering phenologies. As such, it is important to gain a better understanding of how invasive species managers incorporate climate change in their management strategies and identify the invasive species that are expected to pose the greatest threat under climate change. To address these questions, the Regional Invasive Species and Climate Change Management Network surveyed invasive species researchers and managers across four regions of the continental U.S. (the Northeast, Southeast, North Central, and Northwest) to determine the invasive species of greatest concern. This analysis will identify and compare the invasive species most frequently reported by researchers and managers for each region and describe their ecologies.

Soil respiration (Rs) is the second largest carbon dioxide (CO2) flux in terrestrial ecosystems, and it provides an average of 30-90% to ecosystem respiration (Reco). In semi-arid ecosystems, there is a considerable need to expand our knowledge on Rs trends. Chaparral, a semi-arid Mediterranean plant community in California, has the potential to act a sink, which is an essential ecosystem to mitigate climate change. However, Rs responses to meteorological variables remain uncertain in these regions and no studies have quantified how much Rs attributes to Reco in chaparral shrublands. Our study analyzed continuous field Rs data in chaparral shrublands, the effects of soil temperature (Ts) and soil water content (SWC), and its contribution to Reco. Our study incorporated long-term Rs data collected by automated chambers and net ecosystem exchange (NEE) measurements collected by the eddy covariance technique from June 2020 to May 2021 in a chaparral stand in San Diego, California. The results suggest SWC was the strongest driver of Rs, whereas Ts was only a significant control when soil was wet, and temperatures were mild. Monthly Rs/Reco ratios, which described the contribution of Rs to Reco, were highest during the January and February, likely due to the reduced aboveground respiration. Whereas Rs/Reco ratios were lowest when SWC was the driest and Rs was reduced. The results from this study improve our understanding in Rs response to climatic conditions and emphasize the importance of Rs by quantifying its contribution to Reco in chaparral shrublands.

Understanding the vertical structure of atmospheric aerosol is important for solar radiometric study across all spectra. This work pertains to three relevant Southern Hemispheric regions of interest: Southeast Atlantic (SEA), Amazonia (AMZ), and Southeast Pacific (SEP), where seasonal biomass burning events produce smoke plumes of climatic interest. We make use of our previously validated aerosol typology based in AERONET retrieved optical properties, to identify each individual measurement classified as biomass burning within the geographic region of interest. The data is trimmed to select only those classifications measured within the recognized fire-dominant season of each geographic region. We employ The European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis of Copernicus Atmosphere Monitoring Service (CAMS) Carbon Monoxide (CO) data to construct canonical vertical CO profiles (characteristic "shape" curves) from records in the historic burn season window, and within the geographic rectangular boundaries of interest. These canonical CO curves then proxy for AOD curves, constrained to be distributed vertically such that their integrated sum matches to specific Bulk Columnar AOD (BCA) values as determined in the corresponding AERONET record. The layer CO values are normalized to fractional coefficients of the columnar CO total for each canonical profile. These coefficients then are distributed as AOD layer coefficients of the bulk columnar AOD; thus, preserving the canonical profile shapes. This results in vertically resolved AOD profiles for specific geo-region which can be fed into a Radiative Transfer model to result in Total Layered Heating Rates (TLHR) and Aerosol Layered Heating Rates (ALHR) expressed in K/day. We found for example: smoke aerosol plumes in the SEA during the August to October season tend to bi-modally develop between a characteristically higher plume or a lower plume, separated by approximately 1 km vertically. Layer Heating rates develop accordingly. We present methodology, developments, and some cases of these studies specifically for the Southeast Atlantic (SEA) region dominated by seasonal wildfire in Sub-Sahel Africa.

Accurate land cover data can provide powerful insight into characterizing the effects of climate change. Remote sensing satellites enable state-of-the-art land cover measurements, but data collected on the Earth's surface offers a new perspective on land cover characteristics through its more localized scope. Areas that may be generalized to a single pixel in a remote sensing satellite’s data products can be observed at a more granular level through on-site data collection strategies. Low-cost sensors, such as NASA’s Science and Technology Education for Land/Life Assessment (STELLA), make such granular data collection more cost-effective. Citizen Science programs, like the Global Learning and Observations to Benefit the Environment (GLOBE) Program, provide a blueprint for reliably scaling this type of on-site data collection. STELLA is an open-source platform that allows volunteers, like Citizen Scientists, to measure electromagnetic waves to calculate irradiance and temperature in a specific Area Of Interest (AOI). STELLA is cost-effective, as its kit can be assembled by any user with access to makerspace tools commonly found in educational institutions, like 3D printers and soldering irons. This work presents a comparative analysis of the land cover measurements recorded by the STELLA sensor and remote sensing satellites, such as LANDSAT9. Surface temperatures were recorded hourly using the STELLA sensor on four different types of land cover within a 500-square-meter area in Reno, Nevada. The results indicate a statistically significant discrepancy between measurements recorded by the STELLA sensor and LANDSAT, highlighting an untapped data trove in localized sensor measurements. Additionally, we present a data collection control flow for Citizen Science volunteers to record reliable STELLA sensor data. We demonstrate how such Citizen Science data can provide a valuable alternative perspective when compared to its state-of-the-art counterparts, rendering it a valuable tool for future studies. AGU23 Poster LinkAGU23 Abstract Link

It has been found that wintertime mixed-phase cloud properties can present significant differences based on the degree of interaction with air masses coming from locations with reduced sea ice concentration or high presence of sea ice leads. When these air masses are represented by the water vapor transport (WVT) which can interact with the clouds, the properties of the clouds show contrasting differences with respect to cases where the WVT is not interacting with the cloud, i.e. it is not coupled to the cloud. These findings have been reported first for the analysis of the MOSAiC expedition dataset from 2019 to 2020 in the central Arctic \cite{Shupe_2022,Saavedra_Garfias_2023}. In the present contribution, we expand that analysis to long-term measurements (2012-2022) at the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) at the North Slope Alaska (NSA) site in Utqiaǵvik, Alaska. Based on those 10 years of characterized cloud and  sea ice properties, statistically more robust analysis is performed to support or contradict the MOSAiC results. Furthermore, the statistically richer data set from NSA allows to narrow down cases where the properties or coupled clouds to WVT are substantially dissimilar to decoupled cases. Among those are the increase of liquid water path correlated to a decrease of sea ice concentration and ice water paths which are not exhibiting an influence by sea ice concentration. The thermodynamic phase of the clouds also exposes differences based on the state of coupling among the cloud--WVT--sea ice system. These results are put into consideration for the modeling community since sea ice leads are not explicitly resolved in such models, thus the sea ice leads or polynyas effects to processes responsible for mixed-phase cloud formation/dissipation and thermodynamic phase balance are of considerable interest for the parametrization of energy exchange between the surface and the atmosphere in the Arctic.AGU 2023 Session Selection: A093. Microphysical and Macrophysical Properties and Processes of Ice and Mixed-Phase Clouds: Linking in Situ and Remote Sensing Observations and Multiscale Models.

Based on wintertime observations during the MOSAiC expedition in 2019-2020 \cite{Shupe_2022}, it has been found that Arctic cloud properties show significant differences when clouds are coupled to the fluxes of water vapor transport (WVT) coming from upwind regions of sea ice leads \cite{Saavedra_Garfias_2023,saavedragarfias2023}. Mixed-phase clouds (MPC) were characterized by the Cloudnet algorithm using observations from the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) mobile facility and the Leibniz Institute for Tropospheric Research (TROPOS) OCEANet facility, both on board the RV Polarstern . A coupling mechanism to entangle the upwind sea ice leads via the water vapor transport entraintment to the cloud layer has been proposed to successfully identify differences of MPC properties under and without the influence of WVT. For MPC below 3 km liquid water path was found to be increasingly influenced by sea ice lead fraction whereas ice water path was not significantly different in the presence of sea ice leads. However, the ice water fraction, defined as the fraction of ice water path to the total water path, was exhibiting distinguishable asymmetries for cases of MPC coupled to WVT versus decoupled cases. Mainly, the ice water fractions of MPC coupled to WVT were monotonically increasing with decreasing cloud top temperature, while the decoupled cases show increases and decreases in ice water fraction at some specific temperature ranges. The dissimilar behavior of ice water fraction suggests that WVT could importantly influence the processes responsible for heterogeneous ice formation and solid precipitation, therefore coupled MPC and the ice water fraction was also analyzed as a function of snowfall rates at ground. These characteristics are presented based on case studies where WVT back trajectories are available to have a deeper understanding of the interaction processes with sea ice leads that drives the cloud coupled/decoupled differences. Moreover the statistics of our findings based on the whole MOSAiC wintertime period will be put into  consideration.\cite{von_Albedyll_2023}\cite{Shupe_2022}\cite{saavedragarfias2022}AGU 2023 Session Selection: C014. Coupled-system Processes of the Central Arctic Atmosphere-Sea Ice-Ocean System: Harnessing Field Observations and Advancing Models.

A document by Suqin Duan. Click on the document to view its contents.

A document by Sandy Hardian Susanto Herho. Click on the document to view its contents.

Under anthropogenic warming, future changes to climate variability beyond specific modes such as the El Niño-Southern Oscillation (ENSO) have not been well-characterized. In the Community Earth System Model version 2 Large Ensemble (CESM2-LE) climate model, the future change to sea surface temperature (SST) variability is spatially heterogeneous. We examined these projected changes (between 1960-2000 and 2060-2100) in the North Pacific using a local linear stochastic-deterministic model, which allowed us to quantify the effect of changes to three drivers on SST variability: ocean “memory” (the SST damping timescale), ENSO teleconnections, and stochastic noise forcing. The ocean memory declines in most areas, but lengthens in the central North Pacific. This change is primarily due to changes in air-sea feedbacks and ocean damping, with the shallowing mixed layer depth playing a secondary role. An eastward shift of the ENSO teleconnection pattern is primarily responsible for the pattern of SST variance change.

The last glacial inception (LGI) marks the transition from the interglacial warm climate to the glacial period with extensive Northern Hemisphere ice sheets and colder climate. This transition is initiated by decreasing summer insolation but requires positive feedbacks to stimulate the appearance of perennial snow. We perform simulations of LGI with climate model AWI-ESM-2.1, forced by the radiative and greenhouse gas forcing of 115,000 years before present. To compare with the preindustrial (PI) simulation, we use a consistent definition of the seasons during the LGI and the PI and evaluating model output on an angular astronomical calendar. Our study reveals a prominent role of sea ice in the albedo feedback to amplify the delayed climate siregnal at polar latitudes. Through a radiative budget analysis, we examine that the ice-albedo feedback exceeds the shortwave radiative forcing, contributing to the cooling and high latitude snow built-up during LGI.

It is well-established that positive feedbacks between permafrost degradation and the release of soil carbon into the atmosphere impacts land-atmosphere interactions, disrupts the global carbon cycle, and accelerates climate change. The widespread distribution of thawing permafrost is causing a cascade of geophysical and biochemical disturbances with global impact. Currently, few earth system models account for permafrost carbon feedback mechanisms. This research identifies, interprets, and explains the feedback sensitivities attributed to permafrost degradation and terrestrial carbon cycling imbalance with in-situ and flux tower measurements, remote sensing observations, process-based modeling simulations, and deep learning architecture. We defined and formulated high-resolution polymodal datasets with multitemporal extents and hyperspatiospectral fidelity (i.e., 12.4 million parameters with 13.1 million in situ data points, 2.84 billion ground-controlled remotely sensed data points, and 36.58 million model-based simulation outputs to computationally reflect the state space of the earth system), simulated the non-linear feedback mechanisms attributed to permafrost degradation and carbon cycle perturbation across Alaska with a process-constrained deep learning architecture composed of cascading stacks of convolutionally layered memory-encoded recurrent neural networks (i.e., GeoCryoAI), and interpreted historical and future emulations of freeze-thaw dynamics and the permafrost carbon feedback with a suite of evaluation and performance metrics (e.g., cross-entropic loss, root-mean-square deviation, accuracy). This framework introduces ecological memory components and effectively learns subtle spatiotemporal covariate complexities in high-latitude ecosystems by emulating permafrost degradation and carbon flux dynamics across Alaska with high precision and minimal loss (RMSE: 1.007cm, 0.694nmolCH4m-2s-1, 0.213µmolCO2m-2s-1). This methodology and findings offer significant insight about the permafrost carbon feedback by informing scientists and the public on how climate change is accelerating, strategies to ameliorate the impact of permafrost degradation on the global carbon cycle, and to what extent these connections matter in space and time.

It is well-established that positive feedbacks between permafrost degradation and the release of soil carbon into the atmosphere impacts land-atmosphere interactions, disrupts the global carbon cycle, and accelerates climate change. The widespread distribution of thawing permafrost is causing a cascade of geophysical and biochemical disturbances with global impact. Currently, few earth system models account for permafrost carbon feedback mechanisms. This research identifies, interprets, and explains the feedback sensitivities attributed to permafrost degradation and terrestrial carbon cycling imbalance with in situ and flux tower measurements, remote sensing observations, process-based modeling simulations, and deep learning architecture. We defined and formulated high-resolution polymodal datasets with multitemporal extents and hyperspatiospectral fidelity (i.e., 12.4 million parameters with 13.1 million in situ data points, 2.84 billion ground-controlled remotely sensed data points, and 36.58 million model-based simulation outputs to computationally reflect the state space of the earth system), simulated the non-linear feedback mechanisms attributed to permafrost degradation and carbon cycle perturbation across Alaska with a process-constrained deep learning architecture composed of cascading stacks of convolutionally layered memory-encoded recurrent neural networks (i.e., GeoCryoAI), and interpreted historical and future emulations of freeze-thaw dynamics and the permafrost carbon feedback with a suite of evaluation and performance metrics (e.g., cross-entropic loss, root-mean-square deviation, accuracy). This framework introduces ecological memory components and effectively learns subtle spatiotemporal covariate complexities in high-latitude ecosystems by emulating permafrost degradation and carbon flux dynamics across Alaska with high precision and minimal loss (RMSE: 1.007cm, 0.694nmolCH4m-2s-1, 0.213µmolCO2m-2s-1). This methodology and findings offer significant insight about the permafrost carbon feedback by informing scientists and the public on how climate change is accelerating, strategies to ameliorate the impact of permafrost degradation on the global carbon cycle, and to what extent these connections matter in space and time.

A document by Md. Al-Amin. Click on the document to view its contents.

“Pressure and time.” A momentous quote in a compelling movie from a few decades ago interestingly pointed at some of the ingredients that contributed to shaping the Earth. The movie set off from how to seep through masses that appeared just too vast to be shakable or vulnerable – if not by deciphering their inner core. The planetary size and time frame of the Earth may have elicited a perception of a durable, unbuckling living environment – just because “pressure and time” to really affect it would have been out of human reach – supposedly. However, the Earth and environmental sciences have long striven to alert contemporary societies that this is just not the case, as humans have been well exerting scattered yet ubiquitous, planetary-scale pressure over a relatively brief time – with consequential, durable effects. Rising global population, long-term migration shifts of continental extents – due to risks, climate, resources – and unpredicted factors – from vulnerabilities to instabilities – pressure on the environment (natural and built) in unprecedented scale throughout human history. The Earth sciences were born out of deciphering ancient life forms teeming in an aboriginal environment, unfolding on a planet that could be explained only by looking at the Solar system – and at the inception of the Universe.Cross-disciplinary by nature, the Earth and environmental sciences offer crucial tools to gauge location, economic turnout, and societal costs of those very resources and fragilities. They also are pivotal co-actors of intellectual stewardship bridging the gulf with sister disciplines well beyond the remits of the physical sciences. From economics to philosophy, and from history to literature, multiple, diverse and concurring threats call for resourceful, multi-faceted mind- and skill-sets where no single hazard may be really treated apart – not on societal terms.Adapting a famous statement from the 20th century, evolution in a time of poly-crises, multiple hazards, and accrued vulnerabilities is not going to be a dinner party for contemporary societies – especially as they dwell a world perceived as increasingly richer in risks and poorer in resources, with a growing population and across instabilities. Human Earth sciences offer a bridge towards our collective future – as societies, continents, planets.Earth-prints @ INGV  

Ice Floe Tracker is an open-source tool designed to retrieve floe-scale sea ice motion in the Arctic marginal ice zone during spring and summer.  Ice Floe Tracker enables observation of the floe size distribution, ice floe rotation rates, small-scale variation in floe velocity, and individual floe trajectories. Sea ice motion occurs on a wide range of scales, from the interaction of individual pieces of ice at sub-meter scales, the formation of linear kinematic features, and ice transport via basin-wide gyres. Most existing methods for tracking ice motion from remote sensing imagery rely on cross-correlation and are optimized for the winter season in the central Arctic. Cross-correlation-derived motion vectors estimate area-averaged motion and thus are well-suited for close-packed central Arctic ice; however, such estimates have high uncertainties in the dynamic, strongly deforming sea ice cover of the marginal ice zone. Our tool aims to fill this gap by using shape detection and feature tracking to observe floe-scale ice motion.The Ice Floe Tracker algorithm consists of a series of customizable modules. The code is structured as a modular package written in open-source languages. It includes parallel processing, unit testing, a command line interface, and thorough documentation (available on Github). Routines are provided to download imagery from the NASA Moderate Resolution Imaging Spectroradiometer. The satellite imagery is processed to enhance the contrast between liquid water and sea ice, sharpen floe boundaries, and remove atmospheric noise. The image is then segmented, and geometric features of ice floes are extracted. Finally, ice floe geometry and locations are compared to those in subsequent images and linked to form trajectories.  By making this tool open-source, we aim to encourage cross-disciplinary collaboration. Recent results from collaborations between observational oceanography and discrete-element sea ice model development will be highlighted. 

Volcanic aerosols reduce global mean precipitation in the years after major eruptions, yet the mechanisms that produce this response have not been rigorously identified. Volcanic aerosols alter the atmosphere's energy balance, with precipitation changes being one pathway by which the atmosphere acts to return towards equilibrium. By assessing the atmosphere's energy budget in climate model simulations, we here show that global precipitation reduction is largely a consequence of Earth's surface cooling in response to volcanic aerosols reflecting incoming sunlight. In addition, these aerosols also directly add energy to the atmosphere by absorbing outgoing longwave radiation, and this is a major cause of precipitation decline in the first post-eruption year. We also identify mechanisms that oppose the post-eruption precipitation decline, and provide evidence that our results are robust across climate models.

The escalating impact of climate change underscores the need for precise and timely forecasts of meteorological phenomena, particularly droughts, due to their extensive effects on agriculture, water resources, and ecosystems. Addressing this, we introduce a deep learning framework that merges Computer Vision with modified Transformer networks, tailored to predict future drought conditions leveraging historical global climate data. Our model inputs are stacked monthly global maps of Sea Surface Temperature, Temperature 2m above ground, and Total Precipitation, each spanning a year, thus creating a 36-channel input to capture seasonal variability.This study extends conventional Vision Transformers (ViT) by adapting them for sequence processing, enabling the model to learn the intricate temporal dynamics and spatial interdependencies inherent in climate data. By employing a sliding window approach, the model assimilates a sequence length of 12 months for each variable, and the target variables are stacks of Standardized Precipitation & Evapotranspiration Index (SPEI). Our modified ViT architecture successfully integrates the temporal sequencing by adjusting convolutional patch embeddings and positional embeddings, rendering the model sensitive to both the chronological progression and spatial distribution of climatic factors. Preliminary evaluations indicate the model's robust capability in forecasting drought conditions on a global scale. We substantiate these findings with performance metrics that illustrate the model's efficacy in interpreting and predicting the complex, non-linear, and non-stationary patterns of drought phenomena.

We used computer vision (U-Net) model to leverage Standardized Precipitation Evapotranspiration Index (SPEI), Google Trends Search Interest, and Twitter data to understand patterns with which people in Continental United States (CONUS) indicate awareness of and interest in droughts. We found significant statistical relationships between the occurrence of meteorological droughts, as measured by SPEI, and search interest on drought topics over CONUS. SI tends to lag meteorological drought by a period of 2-3 months, however relationships between meteorological drought and corresponding search interest varies significantly over CONUS in both space and time. People in states with increasingly drier conditions generally have become increasingly interested in drought topics. However, with worsening drought conditions in California, public search interest on drought topics in the state has not increased significantly between 2016 and 2020, despite the overall search interest being high. We additionally applied sentiment analysis on 5 million tweets related to droughts and found that public emotions towards drought have become more polarized over time.

A document by Fabio Ciulla. Click on the document to view its contents.

A document by HUANG Yuanyuan. Click on the document to view its contents.

In the recent past, the Gravity Recovery and Climate Experiment (GRACE) satellite mission and its successor GRACE Follow-On (GRACE-FO), have become invaluable tools for characterizing drought through measurements of Total Water Storage Anomaly (TWSA). However, the existing approaches have often overlooked the uncertainties in TWSA that stem from GRACE orbit configuration, background models, and intrinsic data errors. Here we introduce a fresh view on this problem which incorporates the uncertainties in the data: the Probabilistic Storage-based Drought Index (PSDI). Our method leverages Monte Carlo simulations to yield realistic realizations for the stochastic process of the TWSA time series. These realizations depict a range of plausible drought scenarios that later on are used to characterize drought. This approach provides probability for each drought category instead of selecting a single final category at each epoch. We have compared PSDI with the deterministic approach (SDI) over major global basins. Our results show that the deterministic approach often leans towards an overestimation of storage-based drought severity. Furthermore, we scrutinize the performance of PSDI across diverse hydrologic events, spanning continents from the United States to Europe, the Middle East, Southern Africa, South America, and Australia. In each case, PSDI emerges as a reliable indicator for characterizing drought conditions, providing a more comprehensive perspective than traditional deterministic indices. In contrast to the common deterministic view, our probabilistic approach provides a more realistic characterization of the TWS drought, making it more suited for adaptive strategies and realistic risk management.