Search

Highlights: [•] Land surface diurnal temperature differences (DTD) were negatively correlated with soil organic matter (SOM) in low-relief plains. [•] The soil-adjusted vegetation index (SAVI) was positively correlated with SOM in these areas. [•] Regression kriging with a mixed model using DTD, SAVI and soil type predicted SOM. [•] The generic mixed model RK method must be calibrated for each area separately. [Copyright &y& Elsevier]

Abstract: In low relief areas such as plains, easily obtained soil forming factors generally do not co-vary with soil conditions over space to the level that they can be used effectively in digital soil mapping. Mapping variation of soil properties over such areas remains a challenge. This paper presents an approach to mapping soil texture using environmental covariates derived from temporal responses of the land surface to a rainfall event (dynamic feedbacks) collected through remote sensing techniques. The approach consists of four steps: (1) construction of a set of environmental covariates from dynamic feedbacks of the land surface, captured daily from MODIS (Moderate Resolution Imaging Spectroradiometer) images over a short period (6–7days) after a major rain event; (2) derivation of environmental classes based on the set of environmental covariates using a fuzzy c-means clustering; (3) Identification of typical soil texture value for each of the environmental classes from a dataset of field soil samples; (4) mapping of spatial variation of soil texture through a linearly weighted averaging function. The approach was applied to produce soil texture maps in a low relief area situated in south-central Manitoba, Canada. Its performance was assessed through comparison with soil texture maps generated from 1:20,000 traditional soil survey. The assessment was based on 34 field sample sites, independent of the samples used for prediction. The error values (9.42 for MAE and 12.56 for RMSE) of A-horizon percentage of sand from the proposed approach are less than these from the detailed soil survey (10.59 for MAE and 15.12 for RMSE). Similar results were obtained for A-horizon percentage of clay. In addition, the difference between the results of multiple linear regression analysis without and with the MODIS derived variables further demonstrated the effectiveness of the variables at differentiating patterns of soil texture. These indicated that the proposed approach is effective for mapping the variation of soil texture over the low relief area and it could be used to map other soil property variation over similar areas. [Copyright &y& Elsevier]