Improving Satellite Remote Sensing Estimates of the Global Terrestrial Water Cycle Via Neural Network Modeling

48 Pages Posted: 24 Feb 2024

See all articles by Matthew Heberger

Matthew Heberger

affiliation not provided to SSRN

Filipe Aires

affiliation not provided to SSRN

Victor Pellet

affiliation not provided to SSRN

Abstract

Satellite remote sensing is commonly used to observe the water cycle at spatial scales ranging from river basins to the globe. Yet it remains difficult to obtain a balanced water budget using earth observation (EO) data, which highlights their errors and uncertainties. Various methods have been proposed to correct EO datasets to make them more coherent, with a more balanced water budget. This study aimed to improve estimates of water budget components (precipitation, evapotranspiration, runoff, and total water storage change) at the global scale using the methods of optimal interpolation (OI) and neural network (NN) modeling. We trained a set of NNs on a set of 1,358 river basins and validated them on an independent set of 340 basins. We extended the NN models to make pixel-scale estimates of calibrated water cycle components for near-global coverage at 0.5° resolution. Calibrated datasets result in lower water budget residuals in validation basins: the mean of the imbalance was reduced from 11 mm/mo (uncorrected EOs) to 0.03 mm/mo (after calibration by the NN models). Further, the variance of the imbalance was reduced from 44 mm/mo to 24 mm/month after calibration. Results were also evaluated by comparison to in situ observations of precipitation, evapotranspiration, and river discharge. This study suggests to data producers where corrections could be made to the EO datasets, and demonstrates the benefits of machine learning models for studying the water cycle at the global scale.

Keywords: Remote Sensing, Water cycle, Large-sample hydrology, Machine Learning, Neural Networks

Suggested Citation

Heberger, Matthew and Aires, Filipe and Pellet, Victor, Improving Satellite Remote Sensing Estimates of the Global Terrestrial Water Cycle Via Neural Network Modeling. Available at SSRN: https://ssrn.com/abstract=4737861 or http://dx.doi.org/10.2139/ssrn.4737861

Matthew Heberger (Contact Author)

affiliation not provided to SSRN ( email )

No Address Available

Filipe Aires

affiliation not provided to SSRN ( email )

No Address Available

Victor Pellet

affiliation not provided to SSRN ( email )

No Address Available

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
20
Abstract Views
77
PlumX Metrics