A method for the determination of the hydraulic properties of soil from MODIS surface temperature for use in land-surface models

Soil hydraulic properties (SHPs) play an important role in land-surface models, but their spatial distribution is poorly known, and it is not feasible to make field measurements of SHPs everywhere they are needed. In addition, the scale SHPs are measured on (10 cm) is substantially smaller than the scale at which land-surface models are run (>1 km). As a result, land-surface models need landscape hydraulic properties (LHPs), not SHPs. We present a method for identifying LHPs from MODIS surface temperatures. We calibrated LHPs in the Noah land-surface model using MODIS surface temperatures in 2005 at 14 sites from the Atmospheric Radiation Measurement Program (ARM) using locally observed forcing data from 2005. We then used observed flux data during this same time period for model verification. Next, we determined LHPs from MODIS surface temperature at five sites using High Resolution Land Data Assimilation forcing data from 2002. We then used these LHPS to run Noah with 2005 ARM forcing data and compared the output to the same observed 2005 fluxes. Fitting LHPs to MODIS data decreases the error in modeled latent heat flux from 98 W/m² to 67 W/m². Fitting LHPs to these same latent heat flux measurements decreases the error to 50 W/m². Therefore, two thirds of the parameter estimation improvement from calibration to in situ flux data can be achieved using remotely sensed surface temperature. These results are insensitive to errors in other parameters. For example, changing albedo by 0.1 changes the saturated conductivity (Ks) by 10% and the van Genuchten "m" parameter by 1%. However, changing minimum canopy resistance by 40 s/m produced a significant but mutually compensating change in both Ks and "m."