基于MODIS数据的地表组分温度反演研究

组分温度的反演是当前定量遥感的一个难点。与众多的陆面温度遥感模型相比，成熟的组分温度遥感模型相对较少。通过对近地面热红外辐射能量的线性展开，建立了MODIS多波段植被、土壤组分温度反演模型。该模型将组分温度与地表温度定量联系起来，模型中所涉及到的参数充分考虑到研究区湿地环境的特点。在计算组分温度时，首先利用单窗算法得到的MODIS陆面温度反演模型计算地表温度，然后由所建立的组分温度模型计算植被、土壤温度。选用2005年10月31日的鄱阳湖地区MODIS数据试验，与实测的植被、土壤温度相比，反演的组分温度的误差在15~21℃；通过对模型参数的敏感性分析，表明该模型具有一定的稳定性。实践证明，该方法能够有效实现组分温度的分离，是一种组分温度反演的较好方法。

RESEARCH ON THE INVERSION OF SURFACE COMPONENT TEMPERATURE USING MODIS DATA

The inversion of component temperature is one of many problems in the domain of quantified remote sensing.In comparison with land surface temperature models，the models in component temperature inversion seem relatively immature.In this paper，a component temperature model，using MODIS multi bands data was developed.The quantitative relationship between component temperature and land surface temperature was built in the model.In determining the parameters of the model，the character of marsh environment was considered sufficiently.In computing component temperature，the average temperature was acquired using the MODIS land surface temperature inversion model based on the LST model；and then the component temperatures model was constructed to obtain plant temperature and soil temperature.MODIS data (acquired on October 31st，2005) of Poyang Lake district was used to compute component temperature.The validation in the Poyang Lake experiment shows that the error of component temperatures ranges from 15℃ to 21℃.Through analyzing the flexibility of model parameters，it is found that the model developed is relatively steady.It proves that the model built in this paper can effectively separate plant component temperature and soil component temperature.It is a suitable method to compute component temperatures.