CAMELS benchmark models

This is a collection of hydrology model outputs that have all been calibrated using the same forcing data (Maurer) and the same period (1 Oct 1999 until 30 Sep 2008) for the CAMELS data set. The data set contains the model outputs of the validation period (1 Oct 1989 until 30 Sep 1999) only.

About

The data was generated by various different modeling groups (see references below), were each group calibrated their own model to avoid bias in the model calibration. The data was collected by me (Frederik Kratzert) from these groups and converted into a uniform format. The data set was first used in our manuscript "Benchmarking a Catchment-Aware Long Short-Term Memory Network (LSTM) for Large-Scale Hydrological Modeling" https://arxiv.org/abs/1907.08456 that is currently under review in HESS. A separate brief note describing the data set in more detail is in preparation.

The models

Runoff simulations from the following models are included in the data set: - SAC-SMA coupled with Snow-17 snow routine - VIC - FUSE - HBV - mHM

For all of these, outputs of basin-wise calibrated models exist. Additionally, for VIC and mHM, regionally calibrated model outputs exist. For the FUSE model, three different model realizations exist, that are named FUSE_$seed$, where $seed$ is a 3-digit number, specifying the random seed. For the HBV model, two different simulations exist per basin. One, called lower bound (lb) contains the ensemble mean of 1000 uncalibrated HBV models. The other, called upper bound (ub) contains the ensemble mean of 100 basin calibrated HBV models. For more details see the particular references mentioned at the bottom.

Note: Not all models were calibrated for all of CAMELS basins. The basin-specific netCDF file only contain entries for those models that were calibrated for the specific basin.

Structure of netCDF files

For each basin a single netCDF file exist. The files are named as follows: $USGS_ID$_benchmark_models.nc, where $USGS_ID$ is the 8-digit gauge id, as specified in the CAMELS data set.

For each model that was calibrated for a particular basin, the netCDF contains one Variable. Additionally the netCDF file contains the discharge observations under QObs and are stored with date time indices.

Usage

Using Python, the netCDF files can be loaded using e.g. the xarray library

```python import xarray

file_path = 'path/to/netcdf/file.nc' xr = xarray.open_dataset(file_path) ```

The models contained for a particular basin file can be retrieved as follows:

```python

Print list of models for a particular basin

for key in xr.keys(): if key != "QObs": print(key) ```

To plot the data use e.g.

```python import matplotlib.pyplot as plt

fig, ax = plt.subplots() for key in xr.keys(): ax.plot(xr[key], label=key) ax.legend() ```

References

The following papers describe the calibration of the included models: - SAC-SMA & VIC (basin-wise calibrated): Newman, A. J., Mizukami, N., Clark, M. P., Wood, A. W., Nijssen, B., and Nearing, G.: Benchmarking of a physically based hydrologic model, Journal of Hydrometeorology, 18, 2215–2225, 2017. - FUSE: These runs were generated by Nans Addor (n.addor@uea.ac.uk) and passed to me by personal communication. The runs are part of on-going development on FUSE itself and might not reflect the final FUSE performance. - HBV: Seibert, J., Vis, M. J. P., Lewis, E., and van Meerveld, H. J.: Upper and lower benchmarks in hydrological modelling, Hydrological Processes, 32, 1120–1125, 2018. - mHM (basin-wise calibrated): Mizukami, N., Rakovec, O., Newman, A. J., Clark, M. P., Wood, A. W., Gupta, H. V., and Kumar, R.: On the choice of calibration metrics for “high-flow” estimation using hydrologic models, Hydrology and Earth System Sciences, 23, 2601–2614, 2019 - VIC (regionally calibrated): Mizukami, N., Clark, M. P., Newman, A. J., Wood, A. W., Gutmann, E. D., Nijssen, B., Rakovec, O., and Samaniego, L.: Towards seamless large-domain parameter estimation for hydrologic models, Water Resources Research, 53, 8020–8040, 2017. - mHM (regionally calibrated): Rakovec, O., Mizukami, N., Kumar, R., Newman, A. J., Thober, S., Wood, A. W., Clark, M. P., and Samaniego, L.: Diagnostic Evaluation of Large-domain Hydrologic Models calibrated across the Contiguous United States, J. Geophysical Research – Atmospheres., in review, 2019.