The Second Order Method For AMVs Derivation In Clear-sky Region From Geostationary Satellite And Its Quantitative Assessment

In this paper, a new method, named as the Second Order Difference method (the SD method), is raised to extract the faint tracing signal in clear-sky regions from thermal infrared split window channels of geostationary satellite, which is effective in deriving the clear-sky region Atmospheric Motion Vectors (cAMVs) in the lower troposphere and it is a good supplement for the traditional Cloud Motions Vectors (CMVs) and Water Vapor Wind Vectors (WVWV) method.First, the Brightness Temperature (TB) of thermal infrared split window channels from FY-2E geostationary satellite under the tropical atmosphere, the mid-latitude summer atmosphere, the mid-latitude winter atmosphere and the US standard (1976) atmosphere is calculated with the atmospheric radiation transfer model MODTRAN4. Further, a multiple linear regression is carried out on TB and surface temperature ST, water vapor content WV and aerosol optical depth AOD. Based on the multiple linear regression equation achieved, the Signal Noise Ratio (SNR) of the Second Order Difference (SD), the Split Window Difference (BTD) and the Temporal Difference (TD) is calculated.Then, we simulated the ST, WV, AOD and wind field that can represent two typical clear-sky regions, the tropical oceanic clear-sky region and the mid-latitude continental clear-sky region, respectively. Based on the simulation of the two typical clear-sky regions above, MODTRAN is used to calculate the TB of FY-2E thermal infrared split window channels, and then, the SD method, the BTD method and the TD method are used in deriving the cAMVs in these two clear-sky regions. BIAS, Mean Vector Difference (MVD), Standard Derivation (SD) and Root Mean Square Error (RMSE) of derived cAMVs are provided so as to assess the accuracy of these three methods quantitatively.Case studies are presented at last. In case studies, we derived the cAMVs around typhoon with the SD method, the BTD method and the TD method using four successive times’satellite images from the thermal infrared split window channels of FY-2E geostationary satellite. And the BLAS, MVD, SD and RMSE of these cAMVs are also calculated as compared with NCEP wind field. The results of simulation experiments and case studies indicate that, the SD method is superior to the BTD method and the TD method, which can further decrease the negative impact brought about by both the spatial variation and the temporal variation of surface temperature on the texture of trace module, and it can largely increase the accuracy of cAMVs and provides more detailed wind filed information in the lower troposphere, with accuracy comparable to CMVs.