CREST / TERRAHydro: An Earth System Digital Twin Framework
Theory & Modeling
The development of Earth System Digital Twins (ESDTs) represents an ongoing journey towards more accurate and integrated simulations of Earth processes. Inherently interdisciplinary, the endeavor grapples with the challenge of melding subsystems developed by experts in different fields and organizations, requiring communication between different science domains, technology stacks, and data modalities. The Coupled Reusable Earth System Tensor (CREST) framework is a key aspect of our efforts to address these difficulties: by implementing a generic abstraction layer over existing tensor libraries (e.g. TensorFlow, PyTorch, JAX), CREST provides the software foundation for building, operating, and deploying community developed ESDTs. This framework is designed to allow scientists to easily couple together process-based and data-driven models into broader digital twin workflows, while taking advantage of significant efficiency improvements from hardware accelerators.
CREST aims to be a step forward in combining traditional modeling techniques with emerging computational methods, particularly in the context of machine learning. Machine learning plays a foundational role in our approach, both contributing to the development of new data-driven models and aiding in efficient coupling with existing models. Through CREST, we aim to enhance model integration and foster more dynamic interactions within the modeling pipeline – primarily addressing the issues of limited support in current frameworks for gradient propagation, hardware acceleration, and federation with external models. In addition, CREST operational capabilities will include data assimilation, end-to-end distributed model training, black-box model coupling, what-if scenario analysis, and an easy-to-use GUI interface for end users.
Biophysical variable estimation
Remote Sensing
Researching and applying machine learning models to remotely sensed imagery in order to accurately estimate water quality parameters (e.g. Chl-a, TSS, etc.), absorption and scattering spectra, and other inherent optical properties. These imagery are sourced from multispectral satellite missions such as Landsat and Sentinel, as well as from a number of hyperspectral sources in preparation for PACE. The goal is to provide consistent global products across any period of time with satellite coverage, primarily targeting ecosystems in near-shore / inland bodies of water with highly turbid and eutrophic compositions.
Atmospheric Correction Intercomparison Exercise
ACIX is an international collaborative initiative to inter-compare a set of atmospheric correction (AC) algorithms for moderate-resolution optical sensors. The exercises will focus on Sentinel-2 and Landsat-8 data over a set of test areas. The end goal is to identify strengths and weaknesses of various atmopsheric correction methods and propose a hybrid approach to generate operational Level-2 reflectance products.
Visualization of products to support country reporting of Sustainable Development Goals (SDG) 6.3.2 and 6.6.1 related to water quality
Development of tools which aid in monitoring and reporting of water quality across a number of sites in Africa. Water authorities and Ministry of Environment in these countries are working with us to evaluate satellite-derived products at the pilot sites.