论文部分内容阅读
以二向反射辐射信息为基础的植物（作物）冠层反射（ＣＲ）模型已经能够以其反演算法推算植物结构参数，例如叶面积指数、叶角分布、叶绿素含量等。随着计算方法效率的提高，计算机功能的发展，促进人们开发利用卫星影象图大面积估算植物结构参数的计算机程序，推动理论冠层反射模型反演算法的迅速发展。该文在１９９５年由Ｋｕｕｓｋ提出的马尔可夫链冠层反射模型的基础上，发展了一个直接利用ＮＯＡＡＡＶＨＲＲ卫星和陆地卫星影象图数据进行逐个象元反演的计算程序。在ＳＵＮ４工作站上，反演５１２５１２个象元的叶面积指数及叶绿素含量，仅需要几个小时的时间。另外利用地理信息系统中的集群分析法，可使反演更加有效。每个集群点只要运转一次。因此计算机需要的计算时间取决于集群点的数目而与影象图的尺寸无关。一幅影象图如有１００－２００个集群点，就有足够的植被冠层的识辨力。该文利用爱沙尼亚塔图观象台附近的一幅陆地卫星影象图６个波段的数据，再现了农田的叶面积指数。由于马尔可夫链模型对于森林效果不好，对森林的叶面积的估算值均偏低。陆地卫星有６个波段可利用、不像ＮＯＡＡＡＶＨＲＲ只能采用两个波段。大气纠正是必须要做的，该实验采用了Ｖｅｒｍｏｔｅ（１９９４）?
Plant (crop) canopy reflectance (CR) models based on doubly-reflected radiation information have been able to derive plant structure parameters such as leaf area index, leaf angle distribution, chlorophyll content, etc. using their inversion algorithms. With the improvement of computational efficiency and the development of computer functions, people are encouraged to develop computer programs that make use of large-scale satellite imagery estimation of plant structure parameters to promote the rapid development of the canopy reflection algorithm. Based on the Markov chain canopy reflection model proposed by Kuusk in 1995, this paper develops a calculation program that directly uses NOAAAVHRR satellite and terrestrial satellite image data for pixel-by-pixel inversion. In the SUN 4 workstation, the inversion of 512 * 512 pixel leaf area index and chlorophyll content, only a few hours time. In addition, the use of cluster analysis in geographic information systems can make the inversion more efficient. Just run once for each cluster point. Therefore, the calculation time required by the computer depends on the number of cluster points and has nothing to do with the size of the image. An image of 100-200 cluster points, there is enough vegetation canopy discrimination. The paper reproduces the leaf area index of farmland using the data of 6 bands in a terrestrial satellite image near the Observatory in Tartu, Estonia. Since the Markov chain model does not work well for the forest, estimates of the leaf area of the forest are low. Land satellites have six bands available, unlike the NOAA-AVHRR can only use two bands. Atmospheric correction must be done, the experiment used Vermote (1994)?