Assessment of MTI Water Temperature Retrievals with Ground Truth from the Comanche Peak Steam Electric Station Cooling Lake

Abstract

Surface water temperatures calculated from Multispectral Thermal Imager (MTI) brightness temperatures and the robust retrieval algorithm, developed by the Los Alamos National Laboratory (LANL), are compared with ground truth measurements at the Squaw Creek reservoir at the Comanche Peak Steam Electric Station near Granbury Texas. Temperatures calculated for thirty-four images covering the period May 2000 to March 2002 are compared with water temperatures measured at 10 instrumented buoy locations supplied by the Savannah River Technology Center. The data set was used to examine the effect of image quality on temperature retrieval as well as to document any bias between the sensor chip arrays (SCA's). A portion of the data set was used to evaluate the influence of proximity to shoreline on the water temperature retrievals. This study found errors in daytime water temperature retrievals of 1.8 C for SCA 2 and 4.0 C for SCA 1. The errors in nighttime water temperature retrievals were 3.8 C for SCA 1. Water temperature retrievals for nighttime appear to be related to image quality with the largest positive bias for the highest quality images and the largest negative bias for the lowest quality images. The daytime data show no apparent relationship between water temperature retrieval error and image quality. The average temperature retrieval error near open water buoys was less than corresponding values for the near-shore buoys. After subtraction of the estimated error in the ground truth data, the water temperature retrieval error was 1.2 C for the open-water buoys compared to 1.8 C for the near-shore buoys. The open-water error is comparable to that found at Nauru.