Verification of new cloud discrimination algorithm using GOSAT TANSO-CAI in the Amazon

Abstract:

Greenhouse gases Observing SATellite (GOSAT) was launched in 2009 to measure the global atmospheric CO₂ and CH₄ concentrations. GOSAT is equipped with two sensors: the Thermal And Near-infrared Sensor for carbon Observation-Fourier Transform Spectrometer (TANSO-FTS) and the Cloud and Aerosol Imager (TANSO-CAI). The presence of clouds in the instantaneous field-of-view (IFOV) of the FTS leads to incorrect estimates of the concentrations. Thus, the FTS data which are suspected to be cloud-contaminated must be identified using a CAI cloud discrimination algorithm and rejected. Conversely, overestimation of clouds leads to reduce the amount of the FTS data which can be used to estimate the greenhouse gases concentrations. It becomes a serious problem in the region of tropical rainforest such as the Amazon, where there are very few remaining FTS data by cloud cover.

The preparation for the launch of the GOSAT-2 in fiscal 2017 has been progressing. To improve the accuracy of estimates of the greenhouse gases concentrations, we need to refine the existing CAI cloud discrimination algorithm. For the reason, a new cloud discrimination algorithm using support vector machines (SVM) was developed. Visual inspections can use the locally optimized thresholds, though the existing CAI cloud discrimination algorithm uses the common thresholds all over the world. Thus, it is certain that the accuracy of visual inspections is better than these algorithms in the limited region without areas such as ice and snow, where it is difficult to discriminate between clouds and ground surfaces. In this study we evaluated the accuracy of the new cloud discrimination algorithm by comparing with the existing CAI cloud discrimination algorithm and visual inspections of the same CAI images in the Amazon. We will present our latest results.