RETRIEVAL OF GEOPHYSICAL PARAMETERS FROM GOES - EVALUATION OF A SPLIT-WINDOW TECHNIQUE

The performance of a physical split-window retrieval algorithm used to retrieve skin temperature (ST) and precipitable water (PW) from Geost ationary Operational Environmental Satellites' (GOES) infrared measure ments is evaluated. The evaluation assesses the potential of using GOE S measurements to provide accurate retrieval products for climate rese arch studies. Several algorithm performance issues are addressed, incl uding the time of retrieval (diurnal effects), sensitivity to the firs t-guess field, and an evaluation of performance differences associated with the split-window channel characteristics of the GOES-7 VISSR (Vi sible and Infrared Spin Scan Radiometer) Atmospheric Sounder (VAS) and the GOES-8 imager and sounder. The investigation used a mesoscale mod el, initialized by radiosonde data, to generate a simulated atmosphere representative of a case study characterized by summertime conditions over the east-central United States. Synthetic GOES channel radiances were developed from the surrogate atmosphere using GOES channel respo nse functions and an appropriate radiative transfer code. The model fi elds also provided the necessary ground truth and first-guess field fo r the retrieval algorithm. Retrievals of ST and PW were made from the simulated channel radiances associated with the VAS and GOES-8 imager and sounder split-window channels. Retrieval methodologies were applie d to address issues of importance in climate research studies, such as long-term trends and diurnal variability of ST and PW. The performanc e was measured by comparing the retrieved values with the model values at each of the retrieval locations. The algorithm performance for bot h ST and PW was found to be sensitive to the quality of the first-gues s field and to the channel characteristics of the GOES sensors. An est imate of the lower bound on ST and PW retrieval errors was determined. The ST retrievals in all cases showed a significant improvement over the first-guess values. The GOES-8 imager ST retrieval errors, which w ere about half of the VAS values, ranged from about 0.2 to 0.6 K, exhi biting little diurnal effect. The PW retrieval errors ranged from abou t 2.0 to 7.0 mm with a modest sensitivity to the different sensor chan nels. However, a significant diurnal trend in the PW retrieval errors that correlated with the presence of surface- and low-level temperatur e inversions was observed. The algorithm performance results provide i nsight into the application of GOES split-window retrieval methodologi es for climate variability studies and may have implications for opera tional applications of similar retrieval techniques.