Combining SMMR and SSM/I Data for Time Series Analysis of Central North American Snow Water Equivalent

Abstract

When Special Sensor Microwave Imager (SSM/I) and Scanning Multichannel Microwave Radiometer (SMMR) data are combined, the time series of dual-polarized, multichannel, spaceborne passive microwave brightness temperatures extends from 1978 to the present. The Meteorological Service of Canada (MSC) has developed operational snow water equivalent (SWE) retrieval algorithms for western Canada that have been applied to both SMMR and SSM/I data. Climatological research questions that demand a time series of significant length can now be addressed with passive microwave?derived datasets of this nature. Attention must be given, however, to the impact of the slightly different spatial, temporal, and radiometric characteristics between the SMMR and SSM/I data on SWE algorithm performance and, therefore, time series continuity and consistency. In this study, potential bias on SWE retrieval with the MSC algorithms caused by differences between the SMMR and SSM/I sensors is assessed with a series of comparative tests. First, nine winter seasons of SMMR (1978/79?1986/87) and nine winter seasons of SSM/I (1987/88?1995/96) derived SWE estimates are compared to a distributed network of in situ measurements taken from the MSC digital archive of Canadian station data. No mean error bias between the two sensors is found?in some locations the microwave SWE estimates are more accurate during the SMMR years, while for other locations the SSM/I estimates are more accurate. Second, a pentad (5 day) resolution SWE anomaly time series for central North America has been produced for the winter seasons spanning 1978 through 1999. A general anomaly shift between sensors is apparent, with the SMMR seasons containing largely negative SWE anomalies, and SSM/I years containing largely positive anomalies. This trend is confirmed by a difference of means test applied to each winter season pentad, which isolates statistically significant SWE differences between the two time series, with SWE during SSM/I seasons always exceeding SWE retrieved during SMMR seasons. Finally, a principal components analysis (PCA) of the SMMR and SSM/I time series is used to identify any shift in the dominant winter season spatial modes of SWE for central North America. The PCA results also illustrate that the SSM/I time series is characterized by distributions of higher SWE values compared to the SMMR years. The important issue is whether the difference in derived SWE between sensors is induced by climatological factors or algorithm performance.