Visuele interpretasie van ASTER-satellietdata Deel II : grondgebruikkartering in Mpumalanga, Suid-Afrika

Visual interpretation of ASTER satellite data, Part II : land use mapping in Mpumalanga, South Africa Since the initiation in 1960 of the era of satellite remote sensing to detect the different characteristics of the earth, a powerful tool was created to aid researchers. Many land-use studies were undertaken using Landsat MSS, Landsat TM and ETM, as well as SPOT satellite data. The application of these data to the mapping of land use and land cover at smaller scales was constrained by the limited spectral and/or spatial resolution of the data provided by these satellite sensors. In view of the relatively high cost of SPOT data, and uncertainty regarding the future continuation of the Landsat series, alternative data sources need to be investigated. In the absence of published previous research on this issue in South Africa, the purpose of this article is to investigate the value of visual interpretation of ASTER satellite images for the identification and mapping of land-use in an area in South Africa. The study area is situated in Mpumalanga, in the area of Witbank, around the Witbank and Doorndraai dams. This area is characterised by a variety of urban, rural and industrial land uses. Digital image processing of one Landsat 5 TM, one Landsat 7 ETM and one ASTER satellite image was undertaken, including atmospheric correction and georeferencing, natural colour composites, photo infrared colour composites (or false colour satellite images), band ratios, Normalised Difference Indices, as well as the Brightness, Greenness and Wetness Indices. The efficacy with which land use could be identified through the visual interpretation of the processed Landsat 5 TM, Landsat 7 TM and ASTER satellite images was compared. The published 1:50 000 topographical maps of the area were used for the purpose of initial verification. Findings of the visual interpretation process were verified by field visits to the study area. The study found that the ASTER satellite data produced clearer results and therefore have a higher mapping ability and capacity than the Landsat satellite data. Hence, it is anticipated that the use of the full range of the spectral resolution of the ASTER satellite data - which were not available for this study - in statistical pattern recognition and classification methods will enhance the value of the process. Statistical methods are often used to produce visual information which could be applied to prepare land-use change inventories. This should be a