Implementation and Validation of AOT Retrieval Algorithm for GOSAT/TANSO-CAI and Comparison of Them with AOT Used in TANSO-FTS Algorithm

An aerosol retrieval algorithm for Greenhouse gases Observing SATellite (GOSAT) / Thermal And Near infrared Sensor for carbon Observation – Cloud and Aerosol Imager (TANSO-CAI) was developed. This one implements modified Kaufman method when it estimates the ground albedo at 380 nm. In this method, ground albedo at 380 nm is parameterized as a function of NDVI and ground albedo at 680 nm. We also retrieve AOT by minimum reflectance method. We compared these results with those by AERONET (AErosol RObotic NETwork) in 14 sites: Halifax, GSFC, Laegeren, Bratt's Lake, Egbert, Fresno, Yekaterinburg, Yakutsk, Bonanza Creek, Paris, Sevastopol, Tomsk, Sede Boker, and Skukuza. The period of the comparison is April through December, 2009. As aerosol models, we used bimodal log-normal distribution function. As SSAs, we have used two assumptions: non-absorbing aerosol model and statistical averaged value based on AERONET. We have calculated correlation coefficients, biases, and root mean square differences. The results showed that modified Kaufman method is generally better than minimum reflectance method. Correlation coefficient of modified Kaufman method is larger than that of minimum reflectance method in 10 cases. The exceptions are three sites: Egbert, Fresno, and Paris. Correlation coefficient for all sites calculated by modified Kaufman method is 0.59, and that by minimum reflectance method is 0.47 in non-absorbing aerosol model case. Correlation coefficient for all sites by modified Kaufman method is 0.60, and that by minimum reflectance method is 0.48 in AERONET's SSA case. The correlation coefficients by modified Kaufman method are larger than those by minimum reflectance method in both cases. The average bias for all sites calculated by modified Kaufman method was 0.032 and that value of minimum reflectance method was -0.094 in non-absorbing aerosol case. The average bias for all sites by modified Kaufman method was 0.061 and that by minimum reflectance method was -0.080 in AERONET's SSA case. The biases by modified method are closer to zero than those by minimum reflectance method in both cases. The root mean square difference for all sites calculated by modified Kaufman method was 0.118 and that value of minimum reflectance method was 0.132 in non-absorbing aerosol case. The root mean square difference for all sites by modified Kaufman method was 0.147 and that by minimum reflectance method was 0.135 in AERONET's SSA case. The root mean square differences by modified method are smaller than those by minimum reflectance method in both cases. We also compared AOT from CAI with that used in TANSO-Fourier Transform Spectrometer (FTS) algorithm. In TANSO-FTS, the prior value of AOT is made by aerosol transport model, called SPRINTARS. The AOT that is estimated in TANSO-FTS simultaneously with XCO2 is smaller than that by CAI. The AOT that is used in TANSO-FTS as prior value is also compared with ATO from CAI. This AOT is also smaller than that by CAI.