Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Soil Nutrient Mapping Using Aerial Hyperspectral Image and Soil Sampling Data – A Geostatistical Approach

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  Paper number  031046,  2003 ASAE Annual Meeting . (doi: 10.13031/2013.14925) @2003
Authors:   Haibo Yao, Lei Tian, Guangxing Wang, Ignacio A Colonna
Keywords:   Hyperspectral image, soil nutrient, colocated cokriging, geostatistics, Markov model

This paper focused on soil nutrient mapping by colocated cokriging estimator using soil sampling data and aerial hyperspectral image. The soil nutrient factors are K, OM, P1, and pH. A soil nutrient map is a key component in precision farming practices using variable rate technologies. A commonly used approach for soil nutrient mapping is univariate spatial interpolation based on soil sampling data. Another approach is the reflectance-based method using remote sensing images. While both methods use soil sampling and image data independently, cokriging estimator provides a way to combine the two data sets together for better soil nutrient estimation. Furthermore, the colocated cokriging estimator ease the demands in calculating the large and unstable cross-covariance between the primary soil nutrient data and the secondary image data using a Markov approximation. Aerial hyperspectral image can provide soil surface reflectance information with fine spectral details. This makes hyperspectral image a good source of extensively sampled secondary data when applying the colocated cokriging method for soil nutrient mapping. In this study, a single hyperspectral image band was selected for each soil nutrient factor based on the correlation between the nutrient factor and the image band. When comparing with regression analysis, results from colocated cokriging have better correlations and smaller RMSEs. Specifically, correlation of colocated cokriging is 0.84, 0.74, 0.67, 0.76 comparing with the regression correlation of 0.41, 0.70, 0.49, 0.49 for K, OM, P1, and pH respectively. The results show that the colocated cokriging approach is promising for soil nutrient mapping.

(Download PDF)    (Export to EndNotes)