Boreal, Temperate, and Tropical Forests as Vicarious Calibration Sites for Spaceborne Microwave Radiometry

Abstract

We develop methods of using boreal, temperate, and tropical forests as vicarious calibration sites for spaceborne microwave radiometers. The extended sites and improved calibration techniques enable examining warm-scene performances as a complement to cold scenes over the ocean, providing information about scan-dependent biases, detecting and correcting possible calibration errors, and intercalibrating different radiometers. Specifically, this paper shows results as follows. Warm-scene calibration is expanded beyond previously utilized sites in the Amazon rainforest to include inland boreal and temperate forests, as well as forests in coastal and island regions. These additional sites increase the available warm-scene sample size by a factor of 30 compared with using Amazon rainforests alone, allowing more accurate and statistically robust calibration. In addition, their widespread near-global distribution enables near-continuous monitoring of radiometer performance, e.g., with a temporal resolution of almost once per day. Accurate warm-end vicarious calibration enables or enhances several new capabilities for radiometer characterization. Scan-dependent calibration biases that are present at the high end of a radiometer's dynamic range can, for the first time, be reliably detected and characterized. Using this method, we are able to detect the along-scan magnetic interference and edge-of-scan biases that are present in Global Precipitation Measurement Microwave Imager (GMI) observations at warm brightness temperatures (TBs). Our techniques are able to detect and characterize very small calibration anomalies, e.g., a scan-dependent error in GMI with a peak-to-peak amplitude of 0.1 K. In general, scan-dependent performance and antenna pattern correction can be improved with better constraints at both the high and low ends of on-Earth observed TBs. Regional dependence and seasonal changes in the brightness and predictability of the forested calibration sites are detected and characterized. Calibration exhibits a dependence on latitude, most significantly at water vapor channels, where tropical regions are different from middle and high latitudinal regions. Although the root cause is still under investigation, its presence suggests that using limited sites could result in undesirable latitude-dependent calibration biases. The distribution of calibration sites shifts seasonally, which results from green-up of the vegetation canopy in summer and defoliation in winter, and from snow contamination.