Remote sensing methods for wetlands: report to Water Authority of Western Australia and Department of Conservation and Land Management
Kobryn, H.T., Strehlow, K. and Froend, R.H. (1991) Remote sensing methods for wetlands: report to Water Authority of Western Australia and Department of Conservation and Land Management. Report to Water Authority of Western Australia and Department of Conservation and Land Management
Landsat MSS imagery was evaluated for its accuracy in delineating, mapping and classifying wetlands, catchment areas and land cover types. A comparison was made between results from Landsat MSS imagery, 1:25000 scale colour aerial photography and ground truth data, for the same area in the Coobidge Creek Catchment, Western Australia.
Landsat MSS was found to be useful in identifying lakes as small as 0.4 ha with differing water chemistry.
A comparison was made between the aerial photography based inventory (1988) and satellite image derived inventory (1984-1986). Analysis of the satellite based inventory helped to confirm high fluctuations in the water table of many wetlands in the catchment.
Change detection images and bivariate plots of the image data indicate that the physical size of the wetland and a rainfall pattern are only some of the factors influencing wetlands appearance in the satellite image. Lakes of similar size, depth, of close proximity (eg. Lake Carbul, Kubitch and Gidong), varied significantly in the water volume between seasons. Very large lakes, such as Lake Gore fluctuate in their boundaries by as much as 400 m.
Wetland water quality parameters could only be established in qualitative terms. For quantitative assessment, field checks coinciding with the satellite overpass would have to be arranged. Hence, it was not possible to establish water salinity in this study. Relative water turbidity, water depth and fringing vegetation were possible to compare using bivariate scatter plots. Changes within the same water body between the seasons were observable on the same plots overlayed for different dates. Limiting factors in water quality detection were water depth and size of the water body relative to the resolution of the sensor.
Due to the limited time, a detailed comparison of the wetland classification scheme based on a unitemporal and multitemporal data set were not completed. Preliminary assessment indicates that the classification accuracy will improve if a separate training (image sampling) procedure is followed for each of the dates. This is primarily due to large radiance changes between the seasons of the same surfaces. It was hoped that some time would be saved if a common training data set could be used for classifying each image.
|Murdoch Affiliation:||School of Environmental Science|
|Series Name:||Environmental science report; 92/6|
|Publisher:||Report to Water Authority of Western Australia and Department of Conservation and Land Management|
|Item Control Page|
Downloads per month over past year