ORIGINAL RESEARCH
Comparison of Different Geostatistical Methods
for Soil Mapping Using Remote Sensing
and Environmental Variables
in Poshtkouh Rangelands, Iran
Seyed Zeynalabedin Hosseini1,2, Martin Kappas1, Mohsen Bagheri Bodaghabadi3,
Mohammad Ali Zare Chahouki4, Enayatollah Ranjineh Khojasteh5
More details
Hide details
1Department of Cartography, GIS and Remote Sensing, University of Goettingen, Germany
2Faculty of Natural Resources and Desert Studies, Yazd University, Iran
3Department of Soil Science, Shahrekord Univeristy, Iran
4Faculty of Natural Resources, Tehran University, Iran
5Faculty of Mining Engineering, Sahand University of Technology, Tabriz, Iran
Pol. J. Environ. Stud. 2014;23(3):737-751
KEYWORDS
ABSTRACT
The aims of this study were: 1) to map the different soil parameters using three geostatistical approaches,
including; ordinary kriging (OK), cokriging (CK), and regression kriging (RK), 2) to compare the accuracy
of maps created by the mentioned methods, and 3) to evaluate the efficiency of using ancillary data such
as satellite images, elevation, precipitation, and slope to improve the accuracy of estimations. In the rangelands
of the Poushtkouh area of central Iran, 112 soil samples were collected. The maps of different soil parameters
were created using the mentioned methods. To assess the accuracy of these maps, cross-validation analyses
were conducted. The cross-validation results were assessed by the root mean square error (RMSE) and normal
QQ-plot together with sum and average error to suggest the best estimation approach for mapping each
soil parameter. The results have shown that, in most cases, taking the ancillary data into account in estimations
has increased the accuracy of the created maps. Except for clay, the OK method was suggested as the best estimation
method, and the RK and CK were the best recommended estimation methods for the rest of the parameters.
The results suggest the application of the framework of this study for similar areas.