To create a vegetation map for Zambezi National Park and Matetsi Units 6 & 7 to inform effective management of these areas, and for the use of researchers working within these areas.
An understanding of vegetation types in any protected area is crucial for effective management due to the delicate link between vegetation types and the distribution of large mammals; herbivores in particular. The Zimbabwe Parks and Wildlife Management Authority and ALERT have embarked on a process to create a vegetation map for Zambezi National Park and Matetsi units 6 and 7, following an exercise which was done in Matetsi units 1-5 from the 17th to 22nd of April 2013.
This attempt is amongst the first aimed at producing a remote sensing based vegetation map, simplified enough to be used by personnel in the field. For this exercise, satellite data is used from Landsat TM 5 images, with a spatial resolution of 30m. Pre-processing of this image includes Geo-referencing and calibration to correct for atmospheric conditions at the time of acquisition. An unsupervised classification of the image was undertaken to classify 4 colour classes, representing broad vegetation cover types, using the K means classification algorithm. 6 replicates of each colour class have been selected for sampling purposes, with a total of 24 sampling points. At each point (and adjusting for edge effects from any nearby road), a modified step point method over 100 metres with 1m intervals was undertaken. At each point grass, shrub and tree layer are recorded and photos taken. Photos are used to verify vegetation classes obtained from analysis of the data collected. The data points are being used to train non probabilistic classifiers in a remote sensing environment. Algorithms used are the support vector machine and the neural network and since data collection did not follow a random sampling scheme, non-probabilistic classifiers were used. A testing dataset is used to test the accuracy of the classification output and these data points are collected at the time of sampling. The kappa statistic is used to test agreement.
A consultant, from Lupane State University, is then to produce two finalised maps of cover and dominant species.
2013: 8 of the proposed 24 sampling points have already been surveyed.