Weed Mapping Based on Integrated Remote Sensing Methods

Abstract

It is well known that weeds compete with the cultivated plant for the nutrient and water, thus weed coverage could a great influence on the profitability of production. Accordingly, to define the distribution of weed patches is a very important task in precision agriculture. Some traditional and modern methods are available to scout weeds. To survey weed coverage on larger fields by traditional methods is often time consuming. Remote sensing instruments are effective tools to detect weeds in larger area and it is relatively cheaper than traditional way. Two new different techniques of the remote sensing technologies become known: the airborne laser scanning (ALS) and the hyperspectral imaging. From the hyperspectral data, vegetation indices can be calculated, which can help to segment weeds form the soil. In this paper, the writers present a method, where the airborne LIDAR and hyperspectral technique were integrated into a common geoinformatics environment. The test site was a one hectare arable field. Different surface coverage categories were defined by the hyperspectral image. Based on these classes, an n-dimensional classification algorithms were used to define the combination of weed coverage and the extent of biomass. Based on the laser scanning data set a 3D surface was created with runoff conditions. Determination of weed coverage is very important, while it could provide basic information for managing to pass the herbicides out in precision agriculture.