Techniques for Using MODIS Data to Remotely Sense Lake Water Surface Temperatures

Abstract

his study describes a stepwise methodology used to provide daily high-spatial-resolution water surface temperatures from Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data for use nearly in real time for the Great Salt Lake (GSL). Land surface temperature (LST) is obtained each day and then goes through the following series of steps: land masking, quality control based on other concurrent datasets, bias correction, quality control based on LSTs from recent overpasses, temporal compositing, spatial hole filling, and spatial smoothing. Although the techniques described herein were calibrated for use on the GSL, they can also be applied to any other inland body of water large enough to be resolved by MODIS, as long as several months of in situ water temperature observations are available for calibration. For each of the buoy verification datasets, these techniques resulted in mean absolute errors for the final MODIS product that were at least 62% more accurate than those from the operational Real-Time Global analysis. The MODIS product provides realistic cross-lake temperature gradients that are representative of those directly observed from individual MODIS overpasses and is also able to replicate the temporal oscillations seen in the buoy datasets over periods of a few days or more. The increased accuracy, representative temperature gradients, and ability to resolve temperature changes over periods down to a few days can be vital for providing proper surface boundary conditions for input into numerical weather models.