Premium
Integrating properties of soil map delineations into ordinary kriging
Author(s) -
Boucneau,
Van Meirvenne,
Olivier Thas,
; Hofman
Publication year - 1998
Publication title -
european journal of soil science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.244
H-Index - 111
eISSN - 1365-2389
pISSN - 1351-0754
DOI - 10.1046/j.1365-2389.1998.00157.x
Subject(s) - kriging , interpolation (computer graphics) , mathematics , variance (accounting) , classification of discontinuities , statistics , geology , computer science , mathematical analysis , artificial intelligence , motion (physics) , accounting , business
Stratification of a region based on soil map delineations followed by within‐stratum interpolation is sometimes used to combine soil maps and spatial interpolation. However, not all delineations are equally suitable to subdivide an area into precisely located mutually exclusive strata. This paper proposes a flow‐path to characterize the nature of soil map delineations and a methodology to integrate the properties of map delineations into ordinary kriging.Four types of delineations were distinguished based on three criteria: the nature of transition (discontinuous or gradual), the mapping accuracy, and the structure of the within‐unit spatial variation. For each type of delineation the ordinary kriging algorithm was modified to integrate its properties in the interpolation.As a test case, the sand content of the topsoil in the province of West‐Flanders (Belgium) was mapped, using independent test data for validation. Inside the mapping units and at delineations identified as gradual transitions, our procedure, termed ordinary kriging integrating properties of map delineations (OKPD) , performed similarly to stratified ordinary kriging (SOK). However, close to the delineations identified as inaccurately mapped discontinuities the mean square prediction error of OKPD was 0.64 times that of SOK. Moreover, near these delineations, the prediction variance was largely underestimated by SOK (relative variance = 5.1), whereas OKPD produced a more realistic value (relative variance = 1.5).