Heat flux from leads in pack ice from a Landsat TM image

A scale analysis of the spatial variance and covariance of surface temperature and albedo, based on the techniques of the one-dimensional discrete-wavelet transform, is applied to a springtime Landsat TM image of sea ice from the Beaufort Sea. The instrument resolution is 30 m for the visible channels and 120 m for the infrared channel. There is a high level of variability seen at scales from 100 m to 3200 m, with the largest amount seen in a broad peak between 200 m and 800 m. The high correlation between albedo and surface temperature at large scales allows the authors to enhance the temperature signal for scales smaller than the resolution of the thermal sensor. The sensible heat flux from leads is estimated using the difference between the enhanced surface temperature and the smoothed surface temperature, a width-dependent heat-transfer coefficient, and a wind speed. The mean flux from leads and the dependence of the estimate on sensor resolution are then determined. The regional heat flux from leads estimated with the widely available AVHRR instrument may be only 20% of the true value