HYPER model based on bias correction.
Run the main_HYPER_BC.py
Alternative HYPER model based on hybrid reservoir computing (Pathak et al., 2018).
Run the main_RCH.py
HYPER-BC with regression based methods to estimate model weights for ungauged basins.
main_BcReg_kfold.py: (data-rich scenario) for kfold analysis
main_BcReg_random.py: (data-scarce scenario) for random selection of gauged basins including cases with limited number of gauged basins
main_BcReg_region.py: (remote scenario) for prediction of ungauged basins using regional gauged basins
HYPER-BC with spatial proximity based methods to estimate model weights for ungauged basins.
main_BcProx_kfold.py: (data-rich scenario) for kfold analysis
main_BcProx_random.py: (data-scarce scenario) for random selection of gauged basins including cases with limited number of gauged basins
main_BcProx_region.py: (remote scenario) for prediction of ungauged basins using regional gauged basins
RC model without the use of BMA.
Run the main_RC.py
LSTM code for gauged basins, based on Kratzert et al., 2018
- Kratzert, F., Klotz, D., Brenner, C., Schulz, K., and Herrnegger, M.: Rainfall–runoff modelling using Long Short-Term Memory (LSTM) networks, Hydrol. Earth Syst. Sci., 22, 6005-6022, https://doi.org/10.5194/hess-22-6005-2018, 2018.
LSTM code for ungauged basins, based on Kratzert et al., 2019
- Kratzert, F., Klotz, D., Herrnegger, M., Sampson, A. K., Hochreiter, S., & Nearing, G. S. ( 2019). Toward improved predictions in ungauged basins: Exploiting the power of machine learning.Water Resources Research, 55. https://doi.org/10.1029/2019WR026065
train_pub_kfold.sh: (data-rich scenario) for kfold analysis
train_pub_random.sh: (data-scarce scenario) for random selection of gauged basins including cases with limited number of gauged basins
train_pub_region.sh: (remote scenario) for prediction of ungauged basins using regional gauged basins
main_BMA.py: for prediction using BMA only
MARRMoT_model.py: for outputting the results of the uncalibrated MARRMoT models.
Exp1_fig.py ~ Exp4_fig.py corresponds to the result analysis conducted within the paper.
The meteorological daily dataset and observed streamflow from Sawada & Okugawa, 2023 was used to run the uncalibrated version of the 43 MARRMoT models (Knoben, 2019) specified in the paper.
- Sawada, Y. and Okugawa, S.: Multi-model Ensemble for Robust Verification of hydrological modeling in Japan (MERV-Jp) (2.0), https://doi.org/10.5281/zenodo.8176305, 2023.
- Knoben, W. J. M.: wknoben/MARRMoT: MARRMoT_v1.3, , https://doi.org/10.5281/zenodo.3552961, 2019.
basin_data*.csv: the basin characteristics such as the topological, climatical, land form, land use, soil, and geological data, collected from MLIT data portal.
- MLIT: Geospatial Information, https://www.mlit.go.jp/tochi_fudousan_kensetsugyo/chirikukannjoho/tochi_fudousan_kensetsugyo_tk17_000001_00028.html, last access: 22 January 2025.
pub_region_list_ver*.csv: The region column is used to classify the basins into regions.
distance_matrix_v2*.csv: Created by mapping basins' location to QGIS and then using the "Distance Matrix" tool