TY - JOUR
T1 - New geomorphometric variables for non-continuous hillslopes – Assessing the value for digital soil mapping
AU - Temme, Arnaud J.
AU - Schoorl, Jeroen M.
AU - van der Meij, W.M.
PY - 2022/7/15
Y1 - 2022/7/15
N2 - We present a set of new geomorphometric variables that express landscape position relative to breaklines in hillslopes, and test whether these variables are of value in explaining soil property variation in three study sites in the United States, the Netherlands, and Spain. Underlying this work is the recognition that slope breaks, such as cliff lines, lynchets, and large slumped blocks, are associated with processes that affect soil formation around them. For each study site, we digitized slope breaks and calculated vertical and horizontal distance to the nearest lower and higher slope breaks, as well as the relative position between multiple slope breaks, the slope increase at the next higher slope break, and the slope decrease at the next lower slope break. We then assessed the value of these geomorphometric variables in the prediction of slope properties by simple linear regression. At each study site, models were fitted to existing soil observations using a traditional set of geomorphometric variables, and the traditional set plus our newly developed variables. Model comparison indicated that the new variables substantially improved model fit and reduced model error for the site in the United States (Kansas, n = 100), improved model fit but did not reduce model error for the site in the Netherlands (Limburg, n = 192), and did not solve model overfitting issues for the small dataset in Spain (Malaga, n = 66).
AB - We present a set of new geomorphometric variables that express landscape position relative to breaklines in hillslopes, and test whether these variables are of value in explaining soil property variation in three study sites in the United States, the Netherlands, and Spain. Underlying this work is the recognition that slope breaks, such as cliff lines, lynchets, and large slumped blocks, are associated with processes that affect soil formation around them. For each study site, we digitized slope breaks and calculated vertical and horizontal distance to the nearest lower and higher slope breaks, as well as the relative position between multiple slope breaks, the slope increase at the next higher slope break, and the slope decrease at the next lower slope break. We then assessed the value of these geomorphometric variables in the prediction of slope properties by simple linear regression. At each study site, models were fitted to existing soil observations using a traditional set of geomorphometric variables, and the traditional set plus our newly developed variables. Model comparison indicated that the new variables substantially improved model fit and reduced model error for the site in the United States (Kansas, n = 100), improved model fit but did not reduce model error for the site in the Netherlands (Limburg, n = 192), and did not solve model overfitting issues for the small dataset in Spain (Malaga, n = 66).
KW - Cliff
KW - Digital soil mapping
KW - Geomorphometric variable
KW - Geomorphometry
KW - Lynchet
KW - Slope break
U2 - 10.1016/j.geoderma.2022.115848
DO - 10.1016/j.geoderma.2022.115848
M3 - Article
AN - SCOPUS:85127194448
SN - 0016-7061
VL - 418
JO - Geoderma
JF - Geoderma
M1 - 115848
ER -