Statistical assessment of variations in soil properties within and among map units generated by GIS

The statistical analysis conducted on the results from a second order scale (1:25,000) semi-detailed soil survey of the Al-Wafra area in Kuwait demonstrates how GIS-based survey data can be reused to provide an alternative representation of soil property distributions. The results indicate that the mapping of consociations and associations has variable accuracy depending on the type of soils and geomorphic setting of the area. The results show the average map unit purity at the subgroup classification level to be 83% and 75% at the soil series level. This shows the high precision of the soil survey data. The relative variance of diagnostic horizons was very high, reflecting the absence and presence of these features in contrasting soil material. The relative variance results for selected soil properties show that the map units have separated the landscape so that useful statements can be made in relation to soil property distribution that would assist with interpretations for land use. The statistical analysis procedures show that the information in the database can be reinterpreted and analyzed to provide additional quantifiable information. It also shows that impure map units are due to the classification as well as the placement of map boundaries and the scale of mapping used. Interpretation could be improved by using statistical data with a map