There is a newer version of the record available.

Published November 15, 2023 | Version 0.1
Journal article Open

Never train an LSTM on a single basin

Description

About

This repository contains accompanying data to the "Never train an LSTM on a single basin" paper by Kratzert et al. (in review).

  • basin_lists/ contain several text files of basin groups that were used for different experiments. All basin ids are from the CAMELS US dataset. Text files are required to run models with the NeuralHydrology Python package.
  • literature_study/ contains data that was generated during a literature research for this paper.
  • misc/ contains the shapefile of the CONUS level 2 HUCs, which is used to create some of the figures.
  • run_dirs/ contain all model runs and model simulations + metrics. All models were trained with NeuralHydrology.

 

NOTE: This is a pre-release while the paper is under review.

Files

Files (5.8 GB)

Name Size Download all
md5:2d6f9692e9c701ee97bafb19c6a15d54
5.8 GB Download