Abstract
Estimation of aquatic plant infestation areas has been investigated on the basis of remote sensing data. In the studies, aquatic plants were separated from water due to CCT counts having larger than a certain value. Since it is a general case that only some portion of a pixel is covered by waterchestnut (Trapa natans var. japonica Nakai) leaves, areas found do not refer to the total area covered by waterchestnut leaves alone, but represent the total pixel areas where density (surface area covered by waterchestnut leaves divided by a pixel area) is larger than the lower limit, generally around 0.5. In order to evaluate the total surface area covered successively by waterchestnut leaves, the density of the surface leaves must be determined in each pixel. The results would be more useful for an accurate estimation of the total vegetation in a lake.
In order to estimate the density of waterchestnut leaves from Landsat TM data, a simple computational method is presented. The density in each pixel is calculated from radiance at bands 5 and 7 of LANDSAT TM after determining two different densities based on a ground survey. The method is applied to the evaluation of the total area of waterchestnut leaves in the North Lake Inbanuma. The resultant density varies with geographical location with an average density of about 0.5. It is stressed that the computational results of the total surface area covered by waterchestnuts are significantly affected by the lower limit of density or CCT count used in separating waterchestnut from water. The effect of radiance from waterchestnut present below the surface upon the density estimation is also discussed.
