Initial Evaluation of the NOAA Unified Forecast System Coupled Model Using the NASA Land Verification Toolkit

NOAAs Unified Forecast System (UFS) is a global coupled Earth modeling environment with considerable spatial and temporal flexibility. The UFS is currently being used for development of the next generation operational Global Forecast System, GFS v.17. The NOAA Environmental Modeling Center lacks a systematic evaluation tool for land model development. A more extensive and modernized land model evaluation framework is desired both for determining the benefits of adding the Noah-MP land surface model in GFS v.17, and for subsequent evaluations of land surface effects on the coupled system. This work documents recent steps taken to configure the Land Verification Toolkit (LVT) from NASAs Land Information System Framework (LISF) for UFS model evaluation. LVT is a formal system for land surface model evaluation, providing numerous evaluation capabilities for in-situ observations, as well as gridded, remotely sensed, and reanalysis products in their native formats. LVT offers accuracy-based metrics, uncertainty and ensemble diagnostics, information theory, spatial similarity and scale decomposition metrics, as well as external masking and stratification options, lending it considerable evaluation breadth and flexibility. A custom LVT reader was created to process UFS single-month outputs in their native format, and was extended to allow for multi-year evaluations. Initial accuracy-based evaluations were performed on UFS soil moisture output against International Soil Moisture Network CONUS stations from July 2012. Results were found to be consistent with previously performed evaluations. Subsequently, 8 years of UFS soil moisture outputs (2011-2018) were evaluated using the ESA CCI SM Combined (active scatterometer and passive radiometer) product (0.25º daily), with a focus on land and soil classifications. Finally, albedo evaluations were performed using the MODIS GLASS albedo product (0.05 8-daily). This initial work aimed to incentivize the use of LVT by non-LIS groups. Future work would seek to increase interfacing between UFS outputs and LVT, so as to take full advantage of the frameworks evaluation capabilities.