This paper reviews methods for validation of digital soil maps, and describes how a sampling scheme for validation by additional sampling can be designed. It is argued that validation by selecting additional locations by probability sampling is to be preferred over validation by data-spliiting or by cross-validation. The paper describes the elements of a sampling scheme for collecting an additional sample for validation. This is illustrated by two case studies. The first map validated is a categorical soil map of a province in the Netherlands. User's and producers's accuarcy are estimated by stratified simple random sampling and wih ratio-estimators. The second map depicts topsoil organic matter content for a region in Senegal. As the accessibility to the area is poor, the spatial means of the error and of the squared error are estimated by random transect sampling.
|Title of host publication||Pedometrics 2009 Conference. Biennial Meeting of Commission 1.5 Pedometrics - Division 1 of the International Union of Soil Science (IUSS), Beijing, China, 26 - 28 August, 2009|
|Place of Publication||Beijing, China|
|Publisher||China Agricultural University|
|Publication status||Published - 2009|
|Event||Pedometrics 2009 - Beijing, China|
Duration: 26 Aug 2009 → 28 Aug 2009
|Period||26/08/09 → 28/08/09|