This code contains deep learning code used to model hydrologic systems from soil moisture to streamflow or from projection to forecast.
There are two different methods for hydroDL installation:
conda create -n mhpihydrodl python=3.7
conda activate mhpihydrodlInstall our hydroDL stable package from pip (Python version>=3.0)
pip install hydroDL
Install our latest hydroDL package from github
pip install git+https://github.com/mhpi/hydroDL
Note: If you want to run our examples directly, please download the example folder (It contains the code and data for these examples).
There exists a small compatibility issue with our code when using the latest pyTorch version. Feel free to contact us if you find any issues or code bugs that you cannot resolve.
The detailed code for quick start can be found in tutorial_quick_start.py
See below for a brief explanation of the major components you need to run a hydroDL model:
# imports
from hydroDL.model.crit import RmseLoss
from hydroDL.model.rnn import CudnnLstmModel as LSTM
from hydroDL.model.train import trainModel
from hydroDL.model.test import testModel
# load your training and testing data
# x: forcing data (pixels, time, features)
# c: attribute data (pixels, features)
# y: observed values (pixels, time, 1)
x_train, c_train, y_train, x_val, c_val, y_val = load_data(...)
# define your model and loss function
model = LSTM(nx=num_variables, ny=1)
loss_fn = RmseLoss()
# train your model
model = trainModel(model,
x_train,
y_train,
c_train,
loss_fn,
)
# validate your model
pred = testModel(model,
x_val,
c_val,
)
Several examples related to the above papers are presented here. Click the title link to see each example.
The dataset used is NCAR CAMELS dataset. Download CAMELS following this link.
Please download both forcing, observation data CAMELS time series meteorology, observed flow, meta data (.zip) and basin attributes CAMELS Attributes (.zip).
Put two unzipped folders under the same directory, like your/path/to/Camels/basin_timeseries_v1p2_metForcing_obsFlow, and your/path/to/Camels/camels_attributes_v2.0. Set the directory path your/path/to/Camels
as the variable rootDatabase inside the code later.
Computational benchmark: training of CAMELS data (w/ or w/o data integration) with 671 basins, 10 years, 300 epochs, in ~1 hour with GPU.
Related papers:
Feng et al. (2020). Enhancing streamflow forecast and extracting insights using long‐short term memory networks with data integration at continental scales. Water Resources Research.
The dataset used is also NCAR CAMELS. Follow the instructions in the first example above to download and unzip the dataset. Use this code to test your saved models after training finished.
Related papers:
Feng et al. (2021). Mitigating prediction error of deep learning streamflow models in large data-sparse regions with ensemble modeling and soft data. Geophysical Research Letters.
Feng et al. (2020). Enhancing streamflow forecast and extracting insights using long‐short term memory networks with data integration at continental scales. Water Resources Research.
The example dataset is embedded in this repo and can be found here. You can also use this script to train model if you don't want to work with Jupyter Notebook.
Related papers:
Fang et al. (2017), Prolongation of SMAP to Spatio-temporally Seamless Coverage of Continental US Using a Deep Learning Neural Network, Geophysical Research Letters.
Related papers:
Fang et al. (2020). Evaluating the potential and challenges of an uncertainty quantification method for long short-term memory models for soil moisture predictions, Water Resources Research.
How to use: click here
Related papers:
Liu et al. (2022). A multiscale deep learning model for soil moisture integrating satellite and in-situ data, Geophysical Research Letters.
If you find our code to be useful, please cite the following papers:
Feng, DP., Lawson, K., and CP. Shen, Mitigating prediction error of deep learning streamflow models in large data-sparse regions with ensemble modeling and soft data, Geophysical Research Letters (2021), https://doi.org/10.1029/2021GL092999
Feng, DP, K. Fang and CP. Shen, Enhancing streamflow forecast and extracting insights using continental-scale long-short term memory networks with data integration, Water Resources Research (2020), https://doi.org/10.1029/2019WR026793
Shen, CP., A trans-disciplinary review of deep learning research and its relevance for water resources scientists, Water Resources Research. 54(11), 8558-8593, doi: 10.1029/2018WR022643 (2018) https://doi.org/10.1029/2018WR022643
Liu, J., Rahmani, F., Lawson, K., & Shen, C. A multiscale deep learning model for soil moisture integrating satellite and in-situ data. Geophysical Research Letters, e2021GL096847 (2022). https://doi.org/10.1029/2021GL096847
Major code contributor: Dapeng Feng (PhD Student, Penn State), Jiangtao Liu(PhD Student., Penn State), Tadd Bindas (PhD Student., Penn State), and Kuai Fang (PhD., Penn State).
hydroDL has a Non-Commercial license, as found in the LICENSE file.