Transect sampling strategies for semi-detailed characterization of inland valley systems

Abstract

For the semi-detailed characterization of inland valleys systems in the Ivory Coast, transect sampling was used to describe valley morphology, soils, and land use characteristics. Initially, for each valley one transect with a width of 200 to 400 m was proposed. More recently, use of a larger number of transects has been suggested. However, there is no statistical basis for either the number of transects to be described or their optimal width. Two sets of aerial photographs (1979, scale 1:50 000, and 1996, scale 1:20 000) of the Boundiali key area were processed and the results digitized. Using GIS, sets of transects with different width were made, and used for the determination of the sample size for estimating the land use ratio and the valley bottom ratio at valley system level with given confidence intervals. These results were validated with data from the field and with sets of pre-selected transects. The calculated sample size is based on simple random sampling and depends on both transect width and variability in the study area. For the characterization of inland valleys with reasonable confidence intervals, the sample size for random sampling is too large for the semi-detailed characterization method. Smaller samples of transects selected on the basis of sub-recent aerial photographs gave comparable results for land cover characteristics estimation as the larger samples for random sampling. There is scope for reducing sample size by formalizing criteria for transect placement on the basis of the earlier aerial photographs.