Transfer Learning Approaches for Knowledge Discovery in Grid-based Geo-Spatiotemporal Data

TL;DR

This paper introduces HydroDeep, a transfer learning model for geo-spatiotemporal data that enhances regional water discharge predictions efficiently, reducing training time and improving accuracy across different regions.

Contribution

The paper presents HydroDeep, a novel pretrained transfer learning model specifically designed for geo-spatiotemporal data, enabling effective knowledge transfer between regions.

Findings

HydroDeep improves Nash-Sutcliffe efficiency by up to 108%.

Transfer learning reduces training time by 95%.

Four transfer approaches enhance interpretability and accuracy.

Abstract

Extracting and meticulously analyzing geo-spatiotemporal features is crucial to recognize intricate underlying causes of natural events, such as floods. Limited evidence about hidden factors leading to climate change makes it challenging to predict regional water discharge accurately. In addition, the explosive growth in complex geo-spatiotemporal environment data that requires repeated learning by the state-of-the-art neural networks for every new region emphasizes the need for new computationally efficient methods, advanced computational resources, and extensive training on a massive amount of available monitored data. We, therefore, propose HydroDeep, an effectively reusable pretrained model to address this problem of transferring knowledge from one region to another by effectively capturing their intrinsic geo-spatiotemporal variance. Further, we present four transfer learning approaches on HydroDeep for spatiotemporal interpretability that improve Nash-Sutcliffe efficiency by 9% to 108% in new regions with a 95% reduction in time.