The objective of this research is to derive land surface temperature from GOES data at hourly intervals for atmospheric model assimilation to improve short range weather forecasts and nowcasting applications.
The rate of change of LST is sensitive to the characteristics of the land surface such as soil moisture, land use and vegetation. Regions of high soil moisture content or dense vegetation which has access to a a source of moisture exhibit cooler LST than dry soil or vegetation which is stressed because a lack of available soil moisture. This is illustrated in the figures to the right.
The retrieval of land surface temperature (LST) from GOES measurements is accomplished with a physical split window algorithm and the 11 and 12 micrometer channels of either the imager or the sounder. The technique is derived from a perturbation form of the radiative transfer equation that is simplified through parameterization to retrieve the surface parmater corrected for atmospheric water vapor effects. The physical approach requires a priori information, which includes estimates of temperature and mixing ration profiles, precipitable water, and skin temperature. The guess information is used with forward radiative transfer code and GOES spectral response information to calculate channel transmittances and brightness temperatures reuired for the solution equations. LST retrievals are only weakly dependant on the guess profile information. The quality of the LST degrades slightly under inversion conditions (either in the first guess or retrieval environment). Under optimal observing conditions (known surface thermal emissivity), LST retrieval errors are as small as 0.2 K. Variations in surface thermal emissivity unaccounted for in the retrieval process will increase the magnitude of the errors. However, in this particular application the time rate of change of the LST is used rather than its absolute value. As a result, the effects of varying thermal emissivity are negligble. The Geophysical Parameter Retrieval page provides more details into the retrieval process.
A technique has been developed for assimilating GOES-IR skin temperature tendencies into the surface energy budget equation of a mesoscale model so that the simulated rate of temperature change closely agrees with the satellite observations. The simulated latent heat flux, which is a function of surface moisture availability, is adjusted based upon differences between the modeled and satellite-observed skin temperature tendencies. For more information on the satellite data assimilation see the MM5 modeling page or the attached chart.
Several forms of validation are now under way. First, data from the ARM/CART network is being used to assess the accuracy of the LST retrievals. Secondly, the MM5 forecast of temperature and moisture is being compared to local ground truth data and regional surface observations as in the attached comparison. Obvious improvements in low level temperature and mixing ratio fields (not shown) are evident in these limited examples.
Last updated on: November 2, 1999