Ku-band dual-polarization radar backscatter measurements from the SeaWinds on QuikScat scatterometer and microwave radiometer measurements from the Special Sensor Microwave/Imager (SSM/I) are used to determine periods of surface melt and freeze in the Antarctic ice-shelves. The normalized radar backscatter (sigma-0) and backscatter polarization ratio (PR) are used in the maximum likelihood estimation of the ice-state. This method is used to infer the daily ice-surface conditions for 25 selected study points located on the Ronne, Ross, Larsen, Fimbul, Amery, and Shackleton Ice-shelves. The temporal and spatial variations of the radar response are also observed for various neighborhood sizes surrounding each given point during the study period. Criteria for determining the dates of melt-onset and freeze-up for each Austral summer are also presented. Validation of the ice-state and melt-onset date estimates is performed by analyzing corresponding brightness temperature (Tb) measurements from radiometers. QuikScat sigma-0 measurements from 1999 through 2003 are analyzed and it is shown that Ku-band scatterometers are very useful for determining periods of melt in Antarctic ice-sheets and provide high temporal and spatial resolution ice-state estimates. These estimates can be important for long-term studies of the climatic effects of the seasonal and inter-annual melting of the Antarctic ice-sheets. The SeaWinds on QuikScat (QuikScat) and SeaWinds on ADEOS-2 (SeaWinds) scatterometers are identical radar sensors on different spaceborne platforms traversing similar orbits. QuikSCAT and SeaWinds data are used to infer near-surface wind vectors, polar sea-ice extent, polar-ice melt events, among others. In order to verify the relative calibration of these two sensors a simple cross-calibration method is implemented based on land measurements. A first-order polynomial model for the incidence angle dependence of sigma-0 is used to account for biases in the sigma-0 measurements. This model is applied to selected regions of the Amazon rainforest and the Sahara desert. It is shown that the two sensors are well calibrated. Additionally, evidence of a previously presumed diurnal cycle in the Amazon rainforest backscatter is given.

Ku-band dual-polarization radar backscatter measurements from the SeaWinds on QuikScat (QuikScat) and SeaWinds on ADEOS-2 (SeaWinds) scatterometers are used to classify the melt state and estimate melt severity in Greenland. Backscatter measurements are organized into high temporal and high spatial resolution images created using the Scatterometer Image Reconstruction (SIR) algorithm and a new temporal data segmentation technique. Melt detection is performed using a layered electromagnetic model combined with a Markov chain model. The new melt detection method allows classification of the snow-pack into three states: melt, refreeze, and frozen. Melt severity and refreeze severity indexes are also developed. The melt detection methods developed in this thesis are verified by using a one-dimensional geophysical/electromagnetic model simulation of the snow-pack under melting conditions and by comparison with in situ weather station data at the ETH Camp in western Greenland. The diurnal cycle of backscatter measurements is also analyzed at this location. The melt detection and estimation method is applied to the entire Greenland ice-sheet. The resulting melt classifications and melt severity indexes are used to generate a number of maps outlining the features of the 2003 melt season. Good agreement of the melt severity and a 1978 SASS Greenland ice facies map is observed.