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

A document by Christine Olson. Click on the document to view its contents.

A document by Sarah C Kavassalis. Click on the document to view its contents.

Methodology Conclusions and Future work Results This study investigates the impacts of climate change and increasing water demand on water stress levels in the Apalachicola-Chattahoochee-Flint (ACF) River Basin, a critical transboundary water system shared by Georgia, Alabama, and Florida. Recognized as a compound natural hazard, drought in this region is exacerbated by climatic shifts and human-driven policy changes. By analyzing compound drought indices such as SPI, MSSRI, and PDSI, the study explores the interplay between drought events and water management policies, highlighting the influence of state and federal decisions. Results emphasize the need for integrated policy frameworks to address compounded water deficits and mitigate transboundary conflicts.

A document by Kevin Raeder. Click on the document to view its contents.

The Relative Difference (%) between Solar Maximum and Minimum is highest in the UV range. We calculated the shortwave spectral files used in our models with the SOCRATES radiative transfer code. To account for stellar spectral features sensitive to activity, we designed a new band scheme (see the tables) (https://www.met.reading.ac.uk/~yt910424/home/socrates.php) We used SSI data obtained by the LASP TSIS-1 Spectral Irradiance Monitor (SIM) onboard the International Space Station. The data includes measurements taken during at Solar Minimum (averaged from December 1-7/2019, proposed as the reference spectrum for the upcoming CMIP7, as suggested by Funke et al. (2024)) and the Solar Maximum (averaged June 3-20/2024). For wavelengths beyond 2399 nm, both spectra incorporate the same data as the reference spectrum constructed by Coddington et al. (2023).

A document by MILUSKA ANTHUANNET ROSAS BARTUREN. Click on the document to view its contents.

Accurate weather forecasts are crucial to planning disaster management, protecting existing infrastructure, and sustaining federal mission areas. Geospatial foundation models (GFMs) have recently been developed as impactful tools for various stakeholders as vision transformers increasingly outperform numerical weather prediction (NWP) models. The state-of-the-art for machine learning-based weather foundation models has been recently fueled by the vision transformer architecture and may be finetuned for regional or local weather predictions. The vision transformer architecture uses patches representing regions of space and time over the input data. However, weather reanalysis datasets, such as ERA5, provide weather data as projections of the Earth’s spheroidal topology onto a rectangle. This introduces both position-dependent warping and an arbitrary seam at the 180° line of longitude. This may be eliminated by considering the topology of the ERA5 dataset as a cylinder, padding certain longitudes to effectively achieve a wrap-around patch embedding that is not affected by a seam, analogous to previous work developing cylindrical convolutional layers. We demonstrate promising results on smaller vision transformers and propose that this method improves performance on regional finetuning tasks in the vicinity of the 180° line of longitude. To mitigate the warping effect of projection and reduce overreliance on positional embedding, we employ an inverse projection of the source data back to spherical coordinates. We then rotate the sphere so that data are projected back onto the source projection, where the patches are centered and minimally warped. This ensures consistent patch representation across latitudes. By selecting patches in spherical coordinates, the model weights geospatial weather data equally across the globe. We propose these techniques to improve the model’s generalization and are likely to scale to larger models. Further work is required to fully evaluate the impact on state-of-the-art GFMs. We implement these techniques on Microsoft’s ClimaX foundation model, finetuning to make forecasts local to government installations in the Pacific, working towards improved mitigation of weather uncertainty for operational assets near the 180° line of longitude.

Tree height estimation is crucial for accurate biomass estimation and forest monitoring. Synthetic Aperture Radar (SAR) offers two products useful for tree height estimation: SLC (Single Look Complex) images and derived tomographic cubes. In this work, we present machine learning methods to predict forest height in two ways, directly from SLC images and from tomographic cubes, providing valuable context to the upcoming European Space Agency’s Biomass Satellite mission.

Belharra is a wave energy hotspot located on the French Basque coast. Here, unusually large waves occasionally break over a 15-meter deep seamount, producing world-class conditions for big wave surfing. This does not occur often, even though the region frequently encounters highly energetic swells, especially during winter. This work attempts to better understand the underlying processes leading to extreme waves at this location. A combination of spectral and phase-resolving wave models enables analysis of how the incident swell conditions and tide level affect the wave field at both regional and local scales. We carry out a detailed wave-by-wave investigation of the local conditions for a wide range of plausible scenarios. Qualitative comparisons are drawn with data from the trajectory of a surfer riding waves at Belharra during a big swell event. Regional and local features in the bathymetry appear to govern the concentration of swell energy at this spot. Long-period swells from directions around 300° - 315° generate the most energetic local conditions. Lower tide levels lead to greater breaking intensities, albeit with weaker wave amplifications.

In agricultural science, researchers face difficulties publishing and sharing their data-driven crop yield forecast models. The lack of direct and public access to the resulting models impedes collaboration among stakeholders, obscures relevant context and hinders continuous use of the contributions. In addition, researchers have to spend a significant amount of time on accessing, aggregating and pre-processing feature data from various sources. To address both issues, we present a platform that enables (i) public access to standardized and high-quality data by providing (ii) a uniform interface and SDK for data retrieval, as well as (iii) seamless hosting of data-driven crop yield prediction models. Our platform standardizes datasets, sourced from canonical providers such as Copernicus and ISRIC in the areas of observed and forecast weather, soil, and vegetation indices. We facilitate access to this data at different spatial and temporal resolutions through a uniform interface. Thereby, researchers can immediately focus on model development without having to manually gather data from heterogeneous sources. We further allow researchers to directly upload and deploy their crop yield prediction models by submitting their own code written in the Python programming language. The upload process is flexible, minimizing the integration effort on the researcher’s side. The uploaded models are deployed on the platform; historical as well as future prediction values are computed automatically and continuously. The predictions are visualized in an intuitive way, enabling direct comparison to other models and actual historical yield values as well as a direct assessment of models’ forecast capabilities. By publishing their models on our platform, researchers can increase their work’s visibility to actors from academia, media, politics, and business. This enables direct communication between the scientific community and the public, cultivating engagement, feedback and knowledge sharing. In summary, our platform revolutionizes the development and deployment of yield forecast models by providing a comprehensive suite of standardized data sets and the functionality for direct deployment of researchers’ models. This makes yield modeling seamless, collaborative and visible.

In the ongoing effort to improve Earth orientation parameter (EOP) prediction accuracy, the effect of various atmospheric models on EOP forecasts is explored. The first derivative of UT1-UTC, or length of day (LOD), is proportional to the axial component of the dimensionless effective atmospheric angular momentum (AAM) functions, given conservation of angular momentum in the Earth-atmosphere system. Previous work showed that using a robust multiple regression algorithm to pre-process AAM data improves 0 and 7-day predictions of UT1-UTC by 47% and 20%, respectively, even when significant outliers are present in other UT1 or LOD measurement sources during combination. AAM data are produced by various atmospheric modeling centers, and previous work used inputs from multiple AAM models weighted by their regression fit alone after applying effects from estimated systematic errors. This present work compares different AAM models to the derived geodetic AAM excitation functions to better characterize the systematic effects of the individual models. Various techniques to compare model accuracies are explored, and the effects of different AAM models on UT1-UTC forecasts are estimated

We present a novel approach to motion estimation for determining accurate horizontal deformation of Arctic sea ice across all pixels of a SAR image set, developed in support of the Sea Ice Dynamics Experiment (SIDEx) in the Beaufort Sea, northern Alaska, during March and April 2021.

Estimating evapotranspiration (ET) is crucial for understanding changes in the water cycle, yet it remains a challenge due to its dynamic nature. Numerous gridded global ET products exist, but no single product serves as a definitive benchmark. Validation of these large-scale ET products against point data is difficult since ET can vary significantly over space unless atmospheric conditions are stable and land-surface cover is uniform. In this study, six widely used global gridded ET products were first cross-compared and then validated using selected Eddy-covariance towers. The products included: ERA5 (Reanalysis based), GLEAM (Remote sensing based), WGHM (Hydrological model based), Fluxcom (Machine learning based) and KF-ET, WA-ET (Water balance based). The comparison is conducted at a 1˚ spatial and monthly temporal resolution from 2003-2016. Significant differences were observed among the ET datasets, with products performing similarly in some regions but varying considerably in others. Specifically, the Amazon, Southern Africa, Eastern India, and Southern China exhibited large discrepancies, characterized by low correlation (R) and high RMSE values. KF-ET closely resembled Fluxcom in R and RMSE, except in the Amazon and Eastern India, regions where the Fluxcom network is sparse. A dedicated network of Eddy-covariance towers is essential for optimal validation, covering at least a few 1˚ grid cells. To ensure fair comparison, 24 Fluxnet sites were selected from 267 global sites of Fluxnet 2015 dataset based on similarity between the site’s land-use/land-cover (LULC) class and the dominant LULC class within the corresponding 1˚ grid. Further the sites with atleast 30% usable data from 2003-2016 were chosen. Analysis revealed that Fluxcom, ERA5, and KF-ET performed similarly, with Fluxcom outperforming at some sites. However, as Fluxcom is directly derived from Fluxnet sites, its superior performance is expected. WGHM, incorporating various irrigation scenarios, proved reliable among hydrological models. The dominant LULC class percentage within a grid influenced the NSE value between Fluxnet sites and gridded products. The linear relationship between NSE and the dominant LULC class percentage underscores the challenge of comparing site-scale and 1˚ grid-scale data.

The Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) user facility deploys a mix of long-term fixed atmospheric observatories and short to mid-term mobile facilities with a complex array of instrumentation from surface meteorology and flux stations to lidars, radars, and more. Over 30 years and 4 petabytes of data are publically available in ARM’s data discovery tool. ARM has long developed code to support quality control, data product development, and various other tools for working with the data but these have mostly been developed internally in silos across the various groups of ARM infrastructure. Twenty-one years into existence, ARM released its first open-source community software and has been engaging with the open-science community more and more as the years pass.

The Gridded Surface Subsurface Hydrologic Analysis (GSSHA) model is an integrated system of physics-based hydrologic process models. This approach of using a system of physics-based models allows GSSHA to be a good fit for many watershed management decision support applications, such as understanding changing flood dynamics from urban growth as well as using natural and nature-based solutions, such as wetlands or artificial beaver dams, to ameliorate flooding potential. One of the key roles GSSHA has played is modeling compound coastal storm flooding – extreme rainfall combined with storm surge from tropical systems. Simulating such events requires coupling of overland, riverine, coastal, and weather modeling systems. A challenge facing the USACE modeling community is enabling the programmatic connections between models to reduce manual input errors and increase the throughput in time-sensitive situations. In this talk we will discuss how GSSHA has been modernized to enable programmatic-level access to GSSHA. This allows GSSHA to be used in systems of computational models as well as opens up GSSHA to a wide level of data science integration applications, particularly as a part of the NextGen Framework.

Classically, routing in Land Surface Models (LSMs) has been achieved using simple schemes with relatively coarse spatial resolution, such the 0.5-degree TRIP model and its derivatives. However, moving to higher resolution requires us to re-think this approach, but at the expense of increasing complexity and computational cost. Some clues to what these new methodologies might be can potentially be found in the field of global flood inundation modelling. Here, over the last ten years, a set of methods have been developed to map areas at risk of flood inundation at high (now 1 arcsecond) resolution over the entire land surface using full 2D hydrodynamic methods. Such approaches are likely still too costly to implement directly in Land Surface Models, but a consideration of findings from these studies can indicate the essential elements that would need to be included in more a lightweight approach.

Prado Dam, located near Los Angeles California, is used to control flows for water supply and flood control in the Sana Ana River with a basin area of 6,216 km2. Existing operations at Prado Dam are supported by streamflow forecasts issued by the California Nevada River Forecast Center (CNRFC). The CNRFC uses a well-tested set of semi-lumped models developed in the 1970s to simulate and predict flows upstream of Prado as well as Prado reservoir inflows. The models are calibrated to current watershed conditions and executed in a modern forecasting framework. Nonetheless, there remains a potential to improve streamflow forecasts using a contemporary physics-based, gridded hydrologic model that can be coupled with a mesoscale weather forecast model run on a similar scale (West-WRF), which may perform better in an arid region like the Santa Ana River Watershed where there are a limited number of events for model tuning. Additional benefits for reservoir operations may be derived from integrated simulation of the watershed, streams, and reservoir. To test this hypothesis, the Santa Ana River watershed and Prado Dam were simulated with the Gridded Surface Subsurface Hydrologic Analyses (GSSHA) model, as an alternative to other watershed models such as those used by the CNRFC. The model is calibrated/verified to observed streamflows and changes in reservoir volume using observed rainfall data, then tested against observed reservoir inflows for an extended simulation period. Utilizing changes in reservoir volume in the calibration allowed for better estimates of reservoir volume without significantly deteriorating flow estimates. The GSSHA model, developed for larger flood events, outperformed the current system for larger events, while performing comparable for other events, indicating it might provide additional information in support of reservoir operations An experimental operational model was developed to run on the Center for Western Weather and Water Extremes (CW3E) computer resources using a West-WRF forecast; output is downloaded to and displayed on the US Army Corps of Engineers UMIP system, which shows all types of measured and simulated hydrologic data online. Pending funding, this system could be operationalized to provide additional reservoir operational guidance, especially during extreme events.