Benchmarking and evaluating the NASA Land Information System (version 7.5.2) coupled with the refactored Noah-MP land surface model (version 5.0)

Abstract. We integrate the refactored community Noah-MP version 5.0 model with the NASA Land Information System (LIS) version 7.5.2 to streamline the synchronization, development, and maintenance of Noah-MP within LIS and to enhance their interoperability and applicability. We evaluate and compare 5-year (2018–2022) global and regional benchmark simulations of LIS/Noah-MPv5.0 and LIS/Noah-MPv4.0.1 for a set of key land surface variables. Both models capture the spatial and seasonal distributions of observed soil moisture, latent heat (LH), snow water equivalent (SWE), snow depth, snow cover, and surface albedo, with similar bias patterns. Both models tend to have negative soil moisture bias over wet soil regimes and positive bias over dry soil regimes, with slightly higher (≤ ~0.01 m³/m³ for global mean) soil moisture in LIS/Noah-MPv5.0 than LIS/Noah-MPv4.0.1 across most regions. The model bias patterns of LH overall follow those of soil moisture, while LIS/Noah-MPv5.0 has a lower LH across most non-polar regions than LIS/Noah-MPv4.0.1, which reduces the global mean LH bias from 0.99 W/m² to -0.39 W/m². The model SWE bias patterns are dominated by the precipitation and temperature forcing uncertainties, with slightly lower SWE values in LIS/Noah-MPv5.0 (global mean bias of -13.2 mm) than LIS/Noah-MPv4.0.1 (global mean bias of -10.1 mm). The model bias patterns of snow depth generally follow those of SWE. LIS/Noah-MPv4.0.1 consistently overestimates snow cover globally with a mean bias of 0.11, while LIS/Noah-MPv5.0 effectively reduces the overestimates across the global snowpacks with a mean bias of 0.07 because of updated snow cover parameters. Both models show widespread overestimates of surface albedo over mid-latitude and high-latitude regions but significant underestimates in the Sahara Desert and Antarctica. This study reveals possible model deficiencies, motivates future improvements in coupled canopy-snowpack-soil processes and input soil data, and points to the importance of considering observational and forcing data uncertainties in model evaluation.