Satellite radar, coherence radar and the difficulty to measure glacier melting

Climate science requires quantitative data about the melting of glaciers. A major tool is satellite microwave radar. However, the elevation data are highly compromised by the penetration of microwaves (up to 100 m for ice). There have been efforts to calibrate satellite data via airborne laser radar. Following this idea, we found one more serious problem: the satellite data do not only display systematic errors but locally varying noise, up to tens of meters. It appears that this "noise" is equivalent to what we find in white light interferometry at rough surfaces (coherence radar): there is no local elevation averaging over the lateral resolution cell (speckle size). Instead, there is averaging about the random local complex amplitude. Hence, the precision is limited by the surface roughness, which is large at crivassed areas. To overcome this fundamental limit we developed "a fix": Via the precise laser data, areas with small roughness could be identified. Indeed, at these locations, the satellite data display a noise of less than one meter, sufficiently low to determine the penetration error of the microwave measurements.