热红外遥感图像中云覆盖像元地表温度估算初论

A Preliminary View on the Estimation of Land Surface Temperature Under Cloud Cover from Thermal Remote Sensing Data

云覆盖是热红外遥感应用和地表温度(Land Surface Temperature,LST)遥感定量反演的重要障碍。如何估算热红外遥感图像中云覆盖像元的地表温度,是热红外遥感的前沿研究难题。以地表热量平衡为基础,根据地表温度的空间分布连续性、植被对地表温度的影响,提出三种解决云覆盖像元地表温度估算方案:空间插值修正法、植被关系修正法和改进型地表热量平衡法,并探讨云覆盖区地表温度空间分布的洼地效应现象、洼地效应强度及计算方法。基于地表热量平衡方程的洼地效应强度因子和影像灰度值之间关系的数值模拟,是三种估算方案切实可行的关键。

Land surface temperature （LST） is a very important parameter controlling the energy and water bal-ance between atmosphere and land surface. Since it is difficult to obtain such information from ground-based measurements, it appears to be very attractive by using satellite thermal infrared measurements to estimate LST since it can be used for estimating surface temperature at global or local scale. Moreover, the estimation of LST by using satellite remote sensing data is feasible. Cloud cover is a major obstacle to thermal infrared re- mote sensing applications and remote sensing quantitative retrieval of land surface temperature. Furthermore, cloud frequently exists in most time and covers roughly half the surface of the Earth even if the sky is clear. This is the case especially in some regions of high latitudes in the north hemisphere, e.g. the tropics are cov-ered by cloud for about 60% of the time. Therefore, the influence of clouds on LST deserves more discussion and how to estimate LST of pixels covered by cloud on thermal remotely sensed imagery is one of the cut-ting-edge research problems. In this article, based on the theory of surface energy balance （SEB）, three meth- ods, which are spatial interpretation adjustment method, the adjustment method by correlations between LST and Vegetation Indices （VIs） and improved surface energy balance method, have been put forward for the esti- mation of LST when the sky is cloudy. Moreover, the lowland effect of LST spatial distribution under cloud cover and the method for the calculation of its intensity （denoted as SE） were also discussed. Generally speak-ing, when SE equals to 1, it means that SE reaches its maximum due to thick cloud cover .While SE equals to 0, it means that there is no lowland effect in clear sky. SE is strongly affected by the cloud and surface condi-tions. That is to say, SE is influenced greatly by cloud properties such as the time it appears and lasts, its shape, thickness and height and surface characteristics. In normal conditions, SE reaches the highest at the center of cloud cover, and the closer to the cloud cover margin, the less it is. Numerical simulation of the relationship be-tween the intensity factors of the lowland effect and the image gray value, which is based on the SEB, is the key to the practice of the three methods of LST estimation under cloudy conditions.