Sensitivity of Near-Surface Temperature Forecasts to Soil Properties over a Dryland Region in Complex Terrain

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 6 February 2014: 4:15 PM
Room C206 (The Georgia World Congress Center )
Jeffrey D. Massey, University of Utah, Salt Lake City, UT; and W. J. Steenburgh, S. W. Hoch, and J. C. Knievel

Dugway Proving Grounds in Northwest Utah has a silt loam desert land surface and adjacent playa land surface that have very different diurnal temperature ranges. The playa has a reduced diurnal temperature range compared to the silt loam desert and the resulting temperature differences drive thermally forced circulations during quiescent large-scale conditions. Unfortunately, 4 months of operational Weather Research and Forecasting Model (WRF) forecasts over this region erroneously underpredict nocturnal cooling over the silt loam desert with a mean positive bias error in temperature at 2 m (AGL) of 3.4C in the early morning [1200 UTC (0500 LST)]. Over the playa, there is a mean early morning cool bias of -0.7C. The forecasted diurnal temperature ranges are similar over both land surfaces, which prevents the WRF from accurately developing thermally forced flows.

The silt loam desert warm bias is related to the improper initialization of soil moisture and parameterization of the soil thermal conductivity. 2-m temperature forecasts were improved over silt loam and sandy loam soil textures by initializing with observed soil moisture and by replacing the Johansen (1975) parameterization of soil thermal conductivity in the Noah land-surface model with that proposed by McCumber and Pielke (1981). A case study demonstrates how these changes can reduce a nighttime 2-m temperature warm bias of 4.9C over silt loam soil textures to 0.8C. Near-surface temperature improvement is very sensitive to the initialized soil moisture and the greatest improvement occurred during low soil-moisture periods. Predicted ground heat flux and soil thermal conductivity for silt loam soils also more closely matched observations made during the Mountain Terrain Atmospheric Modeling and Observations (MATERHORN) field campaigns when the McCumber and Pielke (1981) method is used along with observed soil moisture. We anticipate similar results in other dryland regions with analogous soil types, sparse vegetation, and low soil moisture.