Development and Applications of a Soil Moisture Validation Tool for the UFS Prototype Experiments

Soil moisture is an important variable for improving the NWP and GCM forecast skills because of land-atmosphere interactions and soil moisture memory. Evaluating model soil moisture by using in-situ and satellite observations may identify model structure and/or model parameter deficiencies and therefore help to improve model physical processes and parameters. Therefore, development of an automatic validation/assessment tool is the first step to achieve such a purpose. To develop the UFS development, NCEP EMC performs many prototype experiments including updates of atmospheric boundary and convection schemes, land surface models, as well as some minor upgrades and bug fixes. In support of this effort, the EMC land team initiated a Prototype/Regional Land Evaluation project in 2021 to evaluate/validate the variables closely associated with land surface processes and parameters from these experiments. EMC is also transitioning to using METplus as the primary UFS evaluation tool. The work in this study also will contribute to the expansion of METplus capabilities for evaluation of land-focused model outputs. Such an automatic evaluation tool was developed based on the soil moisture as the first example and it will be easily extended to AmeriFlux, FLUXNET, GHCN and SNOTEL snow depth, snow depth and SWE, CCI and SMOPS soil Moisture, GLEAM ET, etc.

The tool includes the four steps: (1) processing station-based csv files for a given period, (2) extracting a variable from UFS model grib2 files to save the data into a netCDF file, (3) extracting station-point csv data from the netCDF file, (4) making histogram, spatial map, and time series plots. The python codes and wgrib2 command are used in the validation tool. Besides such a validation tool, we also developed a Fortran code to re-grid regular latitude-longitude soil moisture into the UFS grid. We use this tool to analyze histograms of bias and RMSE, spatial distributions of bias and RMSE, and time series of regionally-averaged soil moisture for various UFS prototype experiments and networks. This presentation shows the preliminary plots and analysis. In the future, many other station-based observations will be used to evaluate/validate UFS model outputs and some model evaluation metrics and scores will be plotted to support the UFS model, METplus, and other related tasks.