Modeling entire nested river trees by integrating the river hierachy into the neural network architecture. This manuscripts proposes HydroNets, an architecture designed for modeling multiple nested gauge stations.
Abstract
Accurate and scalable hydrologic models are essential building blocks of several important applications, from water resource management to timely flood warnings. However, as the climate changes, precipitation and rainfall-runoff pattern variations become more extreme, and accurate training data that can account for the resulting distributional shifts become more scarce. In this work we present a novel family of hydrologic models, called HydroNets, which leverages river network structure. HydroNets are deep neural network models designed to exploit both basin specific rainfall-runoff signals, and upstream network dynamics, which can lead to improved predictions at longer horizons. The injection of the river structure prior knowledge reduces sample complexity and allows for scalable and more accurate hydrologic modeling even with only a few years of data. We present an empirical study over two large basins in India that convincingly support the proposed model and its advantages.
Paper
Kratzert, F. and Klotz, D. and Brandstetter, J. and Hoedt, P.-J. and Nearing, G. and Hochreiter, S.: “Using LSTMs for climate change assessment studies on droughts and floods”. Workshop on Tackling Climate Change with Machine Learning 33rd Conference on Neural Information Process-ing Systems (NeurIPS 2019), Vancouver, Canada.
ArXiv: https://arxiv.org/abs/2007.00595
Slides: PDF-Link
Citation
@inproceedings{moshe2020hydronets,
title={HydroNets: Leveraging River Structure for Hydrologic Modeling},
author={Zach Moshe and Asher Metzger and Gal Elidan and Frederik Kratzert and Sella Nevo and Ran El-Yaniv},
booktitle={Workshop on AI for Earth Sciences 8th International Conference on Learning Representations (ICLR 2020)},
venue={Addis Ababa, Ethopia},
date={26 Apr - 1 May},
year={2020}
}