Abstract Archive Select a year below to view:
Downscaling GRACE Equivalent Water Thickness Data for Mississippi Using Neural Nets
Proceedings of the 2022 Mississippi Water Resources Conference
Year: 2022 Authors: Awawdeh A.R., Yasarer H., Pulla S., Kumar M.
The importance of having high-resolution and effective hydrological data has increased with the recent climate change and continuous relying on underground water. Having such data has been made possible after launching the Gravity Recovery and Climate Experiment (GRACE) in 2002. GRACE enabled researchers to extract data about terrestrial water storage, ice loss, and sea-level change in a temporal resolution of one month. Although this was a great achievement, it is still can't be relied on for small regions because the GRACE data grids are very coarse, i.e. 25 km by 25 km. The goal of this paper is to address the efficiency of using Feedforward Artificial Neural Networks (ANNs) with backpropagation error algorithm to scale down GRACE precipitation data for the State of Mississippi to smaller grids to be used on smaller regions. Both Climate Hazards group Infrared Precipitation with Stations (CHIRPS) and TerraClimate will play an important role in the downscaling process. CHIRPS provides a high-resolution rainfall dataset while TerraClimate provides a dataset of monthly climate and climatic water balance for global terrestrial surfaces. A script in Python programming language and executed via Jupyter Notebook was developed to download all the needed data as well as for the ANN model development. Preliminary results showed that the ANN approach performed well with a significant accuracy and the developed models can be utilized to predict Equivalent Water Thickness with a high accuracy in the Mississippi region.