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.

Abstract: Globally, more people are impacted by extreme hydrologic events such as flooding than all other types of natural disasters combined, and the effects can be devastating. Two examples are the 2012 and 2022 floods along the Niger and Benue Rivers within the Lower Niger River Basin (LNRB) in Nigeria. Flooding within the LNRB typically occurs annually during the rainy season, however, the 2012 and 2022 flood events were of similar magnitude, had catastrophic socioenvironmental impacts, and occurred one decade apart. Limited historical gage data along the Niger and Benue Rivers precludes traditional flood frequency analysis in the LNRB. Hence, this study seeks to utilize globally available observations from satellite remote sensing to compute flood depths using the Floodwater Depth Estimation Tool (FwDETv2.1 version) implemented in Google Earth Engine. Other hydrological and physiographic characteristics of LNRB in 2012 and 2022 are also evaluated using remote sensing observations. Since the FwDET requires only globally available input data (flood inundation map and Digital Elevation Model) which favors data-sparse regions such as Nigeria, the potential for the FwDET tool to automatically quantify flood water depths, an important variable in flood frequency estimation and damage assessment, can be analyzed even when historical observations are lacking. The utility of the FwDETv2.1 for flood management and mitigation studies along global rivers with limited historical data is discussed. ReferenceIdowu, Dorcas, and Wendy Zhou. "Performance evaluation of a potential component of an early flood warning system—A case study of the 2012 flood, Lower Niger River Basin, Nigeria." Remote Sensing 11.17 (2019): 1970.Brakenridge, G. R., Kettner, A. J., Paris, S., Cohen, S., Nghiem, S. V. , River and Reservoir Watch Version 4.5, Satellite-based river discharge and reservoir area measurements, DFO Flood Observatory, University of Colorado, USA. http://floodobservatory.colorado.edu/ SiteDisplays/ 20.htm (Accessed 6 December 2023).Cohen, S.; Peter, B.G.; Haag, A.; Munasinghe, D.; Moragoda, N.; Narayanan, A.; May, S. Sensitivity of Remote Sensing Floodwater Depth Calculation to Boundary Filtering and Digital Elevation Model Selections. Remote Sens. 2022, 14, 5313. https://doi.org/10.3390/rs14215313.B. G. Peter, S. Cohen, R. Lucey, D. Munasinghe, A. Raney and G. R. Brakenridge, "Google Earth Engine Implementation of the Floodwater Depth Estimation Tool (FwDET-GEE) for Rapid and Large Scale Flood Analysis," in IEEE Geoscience and Remote Sensing Letters, vol. 19, pp. 1-5, 2022, Art no. 1501005, doi: 10.1109/LGRS.2020.3031190.Brakenridge, G. Robert, Son V. Nghiem, and Zsofia Kugler. "Passive microwave radiometry at different frequency bands for river discharge retrievals." Earth and Space Science 10.8 (2023): e2023EA002859.Idowu, Dorcas. Assessing the Utilization of Remote Sensing and GIS Techniques for Flood Studies and Land Use/Land Cover Analysis Through Case Studies in Nigeria and the USA. Diss. Colorado School of Mines, 2021.Idowu, Dorcas, and Wendy Zhou. "Global Megacities and Frequent Floods: Correlation between Urban Expansion Patterns and Urban Flood Hazards." Sustainability 15.3 (2023): 2514.

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.

A document by Nicholas Lutsko. Click on the document to view its contents.

A document by Poulomi Ganguli. Click on the document to view its contents.

A document by Poulomi Ganguli. Click on the document to view its contents.

A document by Collin Bode. Click on the document to view its contents.

During marine cold-air outbreaks (MCAOs), when cold polar air moves over warmer ocean, a well-recognized cloud pattern develops, with open or closed mesoscale cellular convection (MCC) at larger fetch over open water. The Cold-Air Outbreaks in the Marine Boundary Layer Experiment (COMBLE) provided a comprehensive set of ground-based in-situ and remote sensing observations of MCAOs at a coastal location in northern Norway. We determine MCAO periods that unambiguously exhibit open or closed MCC. Individual cells observed with a profiling Ka-band radar are identified using a water segmentation method. Using self-organizing maps (SOMs), these cells are then objectively classified based on the variability in their vertical structure. The SOM-based classification shows that comparatively intense convection occurs only in open MCC. This convection undergoes an apparent lifecycle. Developing cells are associated with stronger updrafts, large spectral width, larger amounts of liquid water, lower precipitation rates, and lower cloud tops than mature and weakening cells. The weakening of these cells is associated with the development of precipitation-induced cold pools. The SOM classification also reveals less intense convection, with a similar lifecycle. Such convection, when weakening, becomes virtually indistinguishable from the more intense stratiform precipitation cores in closed MCC. Non-precipitating stratiform cores have weak vertical drafts and are almost exclusively found during closed MCC periods. Convection is observed only occasionally in the closed MCC environment.

A document by Javed Ali. Click on the document to view its contents.

Forecasting hydroclimatic extremes holds significant importance considering the increasing trends in natural cascading climate-induced hazards such as wildfires, floods, and droughts. This study evaluates the performance of five Copernicus Climate Change Service (C3S) seasonal forecast models (i.e., CMCC, DWD, ECCC, UK-Met, and Météo-France) in predicting extreme precipitation events from 1993 to 2016 using 28 extreme precipitation indices reflecting timing and intensity of precipitation in a seasonal timescale. We design indices using various precipitation thresholds to reflect model skill in capturing the distribution and intensity of precipitation over a season. We use percentage bias, the Kendall Tau rank correlation, and ROC scores for skill evaluation. We introduce an impact-based framework to evaluate model skill in capturing extreme events over regions prone to natural disasters such as floods and wildfires. The performance of models varies across regions and seasons. The model skill is highlighted primarily in the tropical and inter-tropical regions, while skill in extra-tropical regions is markedly lower. Elevated precipitation thresholds correlate with heightened model bias, revealing deficiencies in modelling severe precipitation events. The impact-based framework analysis highlights the superior predictive capabilities of the UK-Met and Météo-France models for extreme event forecasting across many regions and seasons. In contrast, other models exhibit strong performance in specific regions and seasons. These results advance our understanding of an impact-based framework in capturing a broad spectrum of extreme climatic events through the strategic amalgamation of diverse models across different regions and seasons, offering valuable insights for disaster management and risk analysis.

This study evaluates the performance of a global storm-resolving model (GSRM), the Simple Cloud-Resolving E3SM Atmosphere Model (SCREAM). We analyze marine boundary layer clouds in a cold air outbreak over the Norwegian Sea in a 40-day simulation, and compare them to observations from satellite and a field campaign of the Atmospheric Radiation Measurement program (ARM). SCREAM qualitatively captures the cold air outbreak cloud transition in terms of the boundary layer growth, cloud mesoscale structure, and phase partitioning. SCREAM also correctly locates the greatest ice and liquid in the mesoscale updraft. However, the study finds that SCREAM might underestimate cloud supercooled liquid water in the cumulus cloud regime. This study showcases the promise of employing high-resolution and high-frequency observations under similar large-scale conditions for evaluating GSRMs. This approach can help identify model features for future process-level studies before allocating extra resources for a time-matched model intercomparison of a specific case.

(Revised) The streamer to leader transition defines much of the physics of long sparks near atmospheric pressures. Streamer length is an important parameter in understanding lightning protection because of its link to step length and striking distance. While streamers are routinely observed in the lab, there have been only a few observations in the field. Fewer still are of natural flashes, and almost none have been observed much above sea-level.

One of the outstanding questions in lightning research is how dart leaders (also called recoil leaders or K-leaders) initiate and develop during a lightning flash. Dart leaders travel quickly (106-107 m/s) along previously ionized channels and occur intermittently in the later stage of a flash. We have recently reported some insights into dart leader initiation and development based on our BIMAP-3D observations. In this presentation we will expand on that work by combining observations and modeling to try to understand the observed dart leader behaviors. BIMAP-3D consists of two broadband interferometric mapping and polarization (BIMAP) systems that are separated by 11.5km at Los Alamos National Laboratory. Each station maps the lightning VHF sources in a 2D space, and the combination of the 2-station measurements provides a detailed 3D source map. A fast antenna is also included at each station for electric field change measurements. Our previously reported observations suggest dart leaders commonly exhibit an initial acceleration, followed by a more gradual deceleration to a stop. We also modeled the dart leader electric field change with a simple configuration of two point-charges, finding that the modeled tip charge increased in magnitude during the initial acceleration in some simple cases. We now employ a more sophisticated model to better understand the distribution of charge along the dart leader channel, and the background electric field in which the dart leader develops.Presented at the AGU 2023 Fall Meeting

Since 2007, the National Academy for Sciences Engineering and Medicine (NASEM) has recommended priorities for Earth Science research and investment every ten years. The Decadal Survey balances the continuation of essential climate variable time series against unmet measurement needs and new Earth Observations made possible by technological breakthroughs. The next survey (2027-2028, DS28) must anticipate the observational needs of the 2030s-2040s, a world increasingly dominated by climate extremes and a rapidly changing Earth system. Here, we identify the critical Earth Observation needs for a hotter, more extreme world where expect challenges in maintaining a safe operating space.

Existing stochastic rainfall generators (SRGs) are typically limited to relatively small domains due to spatial stationarity assumptions, hindering their usefulness for flood studies in large basins. This study proposes StormLab, an SRG that simulates precipitation events at 6-hour and 0.03° resolution in the Mississippi River Basin (MRB). The model focuses on winter and spring storms caused by strong water vapor transport from the Gulf of Mexico—the key flood-generating storm type in the basin. The model generates anisotropic spatiotemporal noise fields that replicate local precipitation structures from observed data. The noise is transformed into precipitation through parametric distributions conditioned on large-scale atmospheric fields from a climate model, reflecting both spatial and temporal nonstationarity. StormLab can produce multiple realizations that reflect the uncertainty in fine-scale precipitation arising from a specific large-scale atmospheric environment. Model parameters were fitted for each month from December-May, based on storms identified from 1979-2021 ERA5 reanalysis data and AORC precipitation. Validation showed good consistency in key storm characteristics between StormLab simulations and AORC data. StormLab then generated 1,000 synthetic years of precipitation events based on 10 CESM2 ensemble simulations. Empirical return levels of simulated annual maxima agreed well with AORC data and displayed bounded tail behavior. To our knowledge, this is the first SRG simulating nonstationary, anisotropic high-resolution precipitation over continental-scale river basins, demonstrating the value of conditioning such stochastic models on large-scale atmospheric variables. The simulated events provide a wide range of extreme precipitation scenarios that can be further used for design floods in the MRB.

A document by Raj Pandya. Click on the document to view its contents.

Extreme weather events are reshaping hydrological cycles across the globe, yet our understanding of the groundwater response to these extremes remains limited. Here we analyze groundwater levels across the South Coast of British Columbia (BC) in the Pacific Northwest with the objective of determining groundwater responses to atmospheric rivers (ARs) and drought. An AR catalogue was derived and associated to local rainfall defining extreme precipitation. Droughts were quantified using dry day metrics, in conjunction with the standardized precipitation index (SPI). From September to January, approximately 40% of total precipitation is contributed by ARs. From April to September, more than 50% of days receive no precipitation, with typically 26 consecutive dry days. We used the autocorrelation structure of groundwater levels to quantify aquifer memory characteristics and identified two distinct clusters. Cluster 1 wells respond to recharge from local precipitation, primarily rainfall, and respond rapidly to both ARs during winter recharge and significant rainfall deficits during summer. Cluster 2 wells are also driven by local precipitation, and are additionally influenced by the Fraser River’s large summer freshet, briefly providing a secondary recharge mechanism to South Coast aquifers. Accordingly, groundwater recessions are offset to later in the summer, contingent on the Fraser River, mediating drought. The results suggest that groundwater memory encapsulates multiple hydrogeological factors, including boundary conditions, influencing the response outcome to extreme events.

The Arctic is experiencing unprecedented moistening, which is expected to have far-reaching impact on global climate and weather patterns. However, it remains unclear whether this newly-sourced moisture originates locally from ice-free ocean regions or is advected from lower latitudes. In this study, we use water vapour isotope measurements in combination with trajectory-based diagnostics and an isotope-enabled AGCM, to assess seasonal shifts in moisture sources and transport pathways in the Arctic. Continuous measurements of near-surface vapour, δ18O, and δD were performed onboard RV Polarstern during the MOSAiC expedition from October 2019 to September 2020. Combining this isotope dataset with meteorological observations reveals that the spatiotemporal evolution of δ18O mimics changes in local temperature and humidity at synoptic to seasonal time scales, while corresponding d-excess changes suggest a seasonal shift in the origin of moisture. Simulation results from the particle dispersion model FLEXPART support these findings, indicating that summer moisture originates from nearby open ocean, while winter moisture comes from more remote sources with longer residence time over sea-ice. Results from a nudged ECHAM6-wiso simulation also indicate that evaporative processes from the ocean surface reproduce summer sotope values, but are insufficient to explain measured winter isotope values. Our study provides the first isotopic characterization of Central Arctic moisture over the course of an entire year, helping to differentiate the influence of local processes versus large-scale vapour transport on Arctic moistening. Future process-based investigations should focus on assessing the non-equilibrium isotopic fractionation during airmass transformation over sea-ice.

A document by Sushel Unninayar. Click on the document to view its contents.

A document by Peter Chu. Click on the document to view its contents.

The paper presents a new method for the decomposition of the horizontal wind divergence among the linear wave solutions on the sphere: inertia-gravity (IG), mixed Rossby-gravity (MRG), Kelvin and Rossby waves. The work is motivated by the need to quantify the vertical velocity and momentum fluxes in the tropics where the distinction between the Rossby and gravity regime, present in the extratropics, becomes obliterated. The new method decomposes divergence and its power spectra as a function of latitude and pressure level. Its application on ERA5 data in August 2018 reveals that the Kelvin and MRG waves made about 6% of the total divergence power in the upper troposphere within 10S-10N, that is about 25% of divergence. Their contribution at individual zonal wavenumbers k can be much larger; for example, Kelvin waves made up to 24% of divergence power at synoptic k in August 2018. The relatively small roles of the Kelvin and MRG waves in tropical divergence power are explained by decomposing their kinetic energies into rotational and divergent parts. The Rossby wave divergence power is 0.3-0.4% at most, implying up to 6% of global divergence due to the beta effect. The remaining divergence is about equipartitioned between the eastward- and westward-propagating IG modes in the upper troposphere, whereas the stratospheric partitioning depends on the background zonal flow. This work is a step towards a unified decomposition of the momentum fluxes that supports the coexistence of different wave species in the tropics in the same frequency and wavenumber bands. 

Canada’s first and only F5/EF5 tornado associated with a supercell touched down near Elie, Manitoba in the late afternoon of 22 June 2007. An observational and numerical simulation analysis with the Weather Research and Forecasting (WRF) model was undertaken to characterize the pre-storm environment and processes leading to storm initiation. WRF sufficiently reproduced the synoptic and mesoscale features, including a supercell-like storm in the region of interest, and supplemented available observations. Synthesis of observational and simulation data suggests that the environment near Elie immediately before storm initiation was primed for tornadic supercells, with large most-unstable and mixed-layer convective available potential energy (4000 J kg^-1) and sufficient vertical shear (effective bulk wind shear 40 kt; effective storm-relative helicity >200 m2 s^-2). Despite enhancement owing to a cold pool left behind by passing early-afternoon convection, shear remained weaker than those typically found in other North American significant tornadic supercell events. The interaction between a surface trough and convective boundary-layer thermals was the primary triggering mechanism of the Elie supercell. The former appeared to be associated with a low pressure arising from the juxtaposition of lower-troposphere cyclonic differential vorticity advection and lee troughing over the western Red River Valley. More observational analysis and numerical sensitivity experiments are required to better diagnose Manitoba terrain’s contribution to the Elie supercell initiation.

Comprehending the intricate interplay between atmospheric aerosols and water vapour in subsaturated regions is vital for accurate modelling of aerosol–cloud–radiation–climate dynamics. But the microphysical mechanisms governing these interactions with ambient aerosols remain inadequately understood. Here we report results from high-altitude, relatively pristine site in Western-Ghats of India during monsoon, serving as a baseline for climate processes in one of the world’s most polluted regions. Utilizing a novel quartz crystal microbalance (QCM) approach, we conducted size-resolved sampling to analyse humidity-dependent growth factors, hygroscopicity, deliquescence behaviour, and aerosol liquid water content (ALWC). Fine-mode aerosols (≤2.5 μm) exhibited size-dependent interactions with water vapour, contributing significantly to ALWC. Deliquescence was observed in larger aerosols (>180 nm), influenced by organic species, with deliquescence relative humidity (DRH) lower than that of pure inorganic salts. This research highlights the significance of understanding ambient aerosol-water interactions and hygroscopicity for refining climate models in subsaturated conditions.

Sap flow observations provide a basis for estimating transpiration and understanding forest water use dynamics and plant-climate interactions. This study developed a continental modeling approach using Long Short-Term Memory networks (LSTMs) to predict hourly tree-level sap flow across Europe based on the SAPFLUXNET database. We developed models with varying levels of training sets to evaluate performance in unseen conditions. The average Kling-Gupta Efficiency was 0.77 for gauged models trained on 50 % of time series across all forest stands and was 0.52 for ungauged models trained on 50 % of the forest stands. Continental models matched or exceeded performance of specialized and baseline models for all genera and forest stands, demonstrating the potential of LSTMs to generalize hourly sap flow across tree, climate, and forest types. This work highlights hence the potential of deep learning models to generalize sap flow for enhancing tree to continental ecohydrological investigations.