Land Surface Parameter And Heat Fluxes Retrieved From Remote Sensing Over The Heterogeneous Underlying Surface

As an interface about exchange of matter and energy between air and land, the land surface plays an important role in the earth science system. The exchange of matter and energy between air and land is determined by the underlying surface, affecting the regional/global atmospheric circulation and climate change. In the study of actual land surface process, it is necessary to get the characteristics of land surface and turbulent energy in a wide range. Fore the conventional meteorological observation is confined to a certain point or a small pat of land. Therefore, using remote sensing technology to get land surface characteristics data occupies an absolute advantage. Remote sensing data has the characters of time dynamic changes and space continuity, so that the conversion from point to area can be achieved, and the key problem is how to get more accurate surface temperature and surface characteritics from remote sensing data.Gulang Heterogeneous Underlying Surface Layer Experiment (GHUSLE) was set up in a small-scale heterogeneous underlying surface, and the distance of two adjacent observation sites was about1.0km during the intensive observations period. To ensure the observed data quality, the instrument parallel contrast observation was designed over the homogeneous underlying surface. The experiment can provide a valuable data for researches about the oasis-desert interaction, and also can provide a reliable data for retrieval method over heterogeneous underlying surface. Using the TERRA/MODIS and FY-3A/VIRR data during the GHUSLE period, the normalized difference vegetation index (NDVI) and surface albedo was estimated in experiment area, Firstly. Then, the algorithms of land surface temperature retrieval using remote sensing data were researched. Two modified algorithms were provided to improve the retrieval accuracy according the generalized local split-window algorithms. In the first method, the error of the observed surface temperature and retrieved value was employed to amend the key parameters in the local split-window algorithms to improve retrieval accuracy. However, this method was limited to have the observed surface temperature, and correction coefficients were not the same in different regiones. In the second method, the new calculations of split-window algorithm for bare soil and complete vegetation coverage conditions were given using the Sobrino’s surface emissivity algorithm at first, and then the calculation for mixed condition was given using vegetation coverage. The second method was more universal and did not require the observed surface temperature. Finally, the energy fluxes were estimated using the satellite data. The Thom’s aerodynamic resistance algorithm was employed in regional flux retrieval. To let Thom’s aerodynamic resistance algorithm apply to the heterogeneous underling surface, the momentum and heat roughness were separated. There are some main conclusions here:(1) The retrieval vegetation coverage was coincided with observational one, assuming it was desert when NDVI was less than0.2, and it was oasis when NDVI greater than0.5. The retrieved albedo using the surface reflectances of MOD09bands were close to observational values, and the root mean square error was0.027.(2) The retrieval accuracy of land surface temperature was improved by the two modified algorithms. The retrieval accuracy by the first method was higher than the second. However, the first method was limited to actual observation data, and the applicability was not universial. The land surface temperatures retrieved using TERRA/MODIS data comparing with the observational values, the correlation coefficients were0.93and0.92, respectively, and the root mean square errors were2.38Ks2.55K, respectively.(3) The response function sensitivity and geographical calibration accuracy were important factors for satellite retrieving. Comparing to FY-3A/VIRR, the TERRA/MODIS sensors on the spectral changes were more sensitive, and the geographical calibration accuracy was higher. Therefore, the land surface temperatures retrieved using TERRA/MODIS data were closer to the observed values. At all the main part of the deviations of retrieval surface temperatures using TERRA/MODIS and FY-3A/VIRR were small than3.0K.(4) The features of the transition zone among oasis-desert heterogeneous underlying surface can be distinguished accurately with satellite data. Using the NDVI and land surface temperature regional distribution diagrams of the experiment area, the boundaries of the oasis, transition zone and the desert can be clearly displayed. The gradients of NDVI and land surface temperature in the oasis and desert regions are smaller than that in transition zone, and are greater in the transition zones.(5) Using Thom’s algorithm with separation of the momentum and heat roughnesses, the retrieval accuracy of aerodynamic resistance can be improved over the heterogeneous underling surface. The root mean square error between the retrieval sensible heat flux and the observed one was23.49W·m-2.