On the accuracy of soil survey in Queensland, Australia

Abstract

The accuracy of soil survey is not well described in a consistent manner for either conventional or digital soil mapping in Queensland or more generally in Australia. Concepts of accuracy are often poorly understood and the rise of digital soil mapping has led to further terminology confusion for clients. Despite long-standing recommendations for derivation of accuracy statistics of soil surveys via statistically-based external validation, accuracy assessment by this method has been limited. Concepts for accuracy description (overall, producers and users accuracy) from the remote sensing discipline are applicable to soil survey and their use should be encouraged. An analysis of 12 published 1:50000 and 1:100000 soil surveys in Queensland revealed a 73% to 97% match between mapped polygonal and site data. This, in conjunction with accuracy standards for similar mapping disciplines and published soil survey accuracy assessments, leads us to recommend that a benchmark of 80% accuracy is realistic for all types of soil surveys. The adoption of a benchmark is however dependent upon further development and evaluation of accuracy assessment methods and standards, particularly in relation to minimum sample size and acceptance criteria. These outcomes will only be achieved if all surveys include accuracy assessment within the survey design.