Remotely sensed data is generally influenced by temporal factors such as Sun-target-satellite geometry and atmospheric conditions, which are major concerns in quantitative monitoring of long-term changes of the Earth's surface. The inconsistencies are usually eliminated with radiometric normalization or cross-sensor calibration. In most cases, these strategies use statistical relationships among multi-temporal images, which do not meet the rigorous requirements of quantitative remote sensing. While MODIS Level-1B (L1B) products have wide applications, a mathematical relationship was derived among the pixel values of pseudo-invariant features (PIFs) of the multi-temporal images. The quantitative relationship was validated using visible and reflective infrared bands of MODIS LIB products. The results showed that the quantitative relationship consists of additive and multiplicative parts relying on sun-target-satellite geometry, atmospheric conditions and sensor parameters. The derived relationship had a good agreement with the multi-temporal relationship of PIF pixels obtained from individual images. It offers a quantitative basis for statistically based radiometric normalization or cross-sensor calibration.%由不同时相和不同传感器所导致的遥感数据非一致性是定量监测地表长期变化所必须面对的一个重要问题.用于消减遥感数据非一致性的归一化辐射校正和传感器交叉定标,多采用统计方法,难以满足定量遥感发展的需求.本文针对MODIS L1B光学波段产品数据,推导了不同时相非变像元灰度值(SO之间的数学表达式,并利用MODIS的可见光-短波红外波段L1B数据对该定量关系进行了验证.研究表明,在可见光-短波红外波段,L1B非变像元在不同时相间的定量关系分为乘性和加性两部分,主要与太阳光照条件、大气状态以及传感器自身等变化因素有关.虽然基于MODIS非变像元时相关系的检验结果与理想状态下的结果存在偏差,但总体趋势仍然较好.
展开▼
