| The current techniques for deriving atmospheric motion vector with satellite images are mainly concentrated on cloud motion and water vapor motion which use cloud in IR imagery and water vapor in water vapor imagery as tracer, respectively. To the area where cannot be recognized as cloud-covered, only some middle-high winds derived from water vapor imagery can be achieved. Since the clear-sky region in the IR imagery contain not only surface temperature but also the water vapor and aerosol information, this paper concentrate on calculating wind field in clear-sky region using split window difference method. To elucidate the mechanism of this method, formula derivation, parameters sensitivity analysis, simulation image test, system design and instance analysis are carried out.By analyzing the expression of the split window brightness temperature difference (BTD), we point out the texture features on the difference imagery are caused by the non-uniform distribution of the temperature terms (surface temperature and atmospheric temperature profile) and atmosphere terms (water vapor content and aerosol content). By making a subtraction, temperature terms in the expression can be treated as the same order of magnitudes as the atmosphere terms. In the meantime, while in the correlation method procedure using difference imagery, the impact of local steady atmospheric composition (e.g. O3and CO2) can be ignored.In order to quantitative analysis the impact of the two terms (i.e. temperature terms and atmospheric terms), MODTRAN radiation transfer model is used to perform a sensitivity analysis on brightness temperature difference in thermal IR split window channels with respect to water and aerosol contents in the observation pixel, which might be used as tracers for deriving atmospheric motion wind vectors in clear atmosphere regions under middle latitude summer and winter atmosphere condition, by combination of atmospheric parameters and satellite-borne radiometer sensitivity performance. The results show that the impact of temperature terms in BTD is relatively small in both atmospheric conditions. As a result, in the case of the changed surface temperature, tracking the atmospheric weakness signals in the clear-sky region using BTD imagery is essentially tracking the movement of the water vapor and aerosol in the clear-sky area. On this basis, a water vapor mass under the middle latitude summer condition and an aerosol mass under the middle latitude winter condition are simulated and tracked both in IR imagery and BTD imagery, the result shows that by using the BTD imagery, the impact of changed surface temperature has been effectively suppressed, which makes a more precisely matching result.Based on the sensitivity analysis, the experimental system of clear-sky region wind retrieval is written on the Visual Studio2010platform. In the part of height assignment, using MODTRAN simulated the maximum BTD response layer of the water vapor and aerosol under the corresponding atmospheric clear-sky region, and adopting them as the tracer height of water vapor (800hPa) and aerosol (850hPa), respectively. To verify the effectiveness of this method in dealing with the real satellite imagery, two cases (one for water vapor tracking and the other for aerosol tracking) are chose and analyzed. The results of water vapor case show a good consistency with the NCEP atmospheric wind of800hPa while the results of aerosol show a relatively good consistency with the NCEP atmospheric wind of850hPa. |
