Premium
An approach to extrapolate categorical agricultural soil data to nonmapped areas using majority vote
Author(s) -
Böhm Christian,
Kiesel Joachim,
Deumlich Detlef,
Thiere Jürgen
Publication year - 2009
Publication title -
journal of plant nutrition and soil science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.644
H-Index - 87
eISSN - 1522-2624
pISSN - 1436-8730
DOI - 10.1002/jpln.200700057
Subject(s) - extrapolation , categorical variable , soil map , environmental science , scale (ratio) , soil classification , soil science , remote sensing , cartography , geography , mathematics , statistics , soil water
Soil maps are indispensable base material for planning processes and risk analyses. For many studies, however, area‐wide uniformly scaled soil maps are not available. Our objective was to develop a GIS‐based method for providing information for NoData areas of the Medium‐Scale Agricultural Site Mapping (MMK). NoData areas represent nonagricultural zones, in this case especially forests. A grid‐based methodology for extrapolating soil mapping units of agricultural areas to the entire landscape was developed and tested in Mecklenburg–Western Pomerania (total area 23,170 km 2 ). The principle of the extrapolation method is based on a statistical neighborhood analysis implemented by the moving‐window technology considering selected site features as geology, altitude, and relief energy. Results were illustrated exemplarily for soil‐moisture conditions, which are expressed as “hydromorphy association types”. Four different indicators were derived to ensure a spatially differentiated evaluation of the extrapolation quality. The precision with which it is possible to extrapolate already mapped MMK units in terms of the hydromorphy association type was used to validate the extrapolation method. Evaluation of the results was based on the percentage of Correctly Extrapolated Pixels (CEP), which were derived from a confusion matrix. The percentage of CEP was 26% when the 15 single classes of the hydromorphy association type were considered separately. The percentage of CEP rose up to 65% after grouping the hydromorphy association types into three practically combined categories. It could be demonstrated that the extrapolation technique applied represented an objective and relatively fast and cost‐effective method to obtain an area‐wide MMK mapping.