712-7 Field Level Digital Soil Mapping of Cec.

See more from this Division: A08 Integrated Agricultural Systems
See more from this Session: Managing Spatial Variability/Div. A08 Business Meeting

Wednesday, 8 October 2008: 10:00 AM
George R. Brown Convention Center, 371C

John Triantafilis, Sam Mostyn Buchanan and Kevin La Lau, School of Biological, Earth and Environmental Sciences, The University of New South Wales, Sydney, Australia
Abstract:
At the field level the demand for spatial information of soil properties is rapidly increasing owing to its requirements in precision agriculture and soil management. One of the most important is the cation exchange capacity (CEC-cmol(+)/kg). This is because CEC is an index of the shrink-swell potential and is a rough measure of soil structural resilience to tillage. Owing to the time consuming nature and expense associated with measuring CEC, various ancillary data sets and statistical methods can be used to develop environmental correlations. However, there is little scientific literature which implements this approach or addresses the issue of the amount of ancillary data required to maximise precision and minimise bias of spatial prediction of CEC at the field level. In this paper we address this by illustrating a case study, whereby ancillary data (i.e. electromagnetic (EM) induction and spectral brightness) are coupled with soil CEC to develop a hierarchical spatial regression (HSR) model that is used to map the spatial distribution of average (0-2 m) CEC across an irrigated cotton field. In the first instance, EM signal data is collected in the vertical mode of operation using an EM38 (EM38-v) and EM31 (EM31-v) along 19 transects spaced 24 m apart. Spectral brightness data (Red, Green and Blue) is obtained from a digitised colour aerial photograph at each EM survey site. Calibration data is collected at 33 soil sampling locations which generally cover the areal extent of the field and the low, intermediate and high values of EM signal data. We statistically compare a standard least squares MLR model which includes all ancillary data, and a stepwise MLR model which only includes the statistically valid EM38-v signal data and the Easting trend surface vector. The reliability of the latter HSR model is analysed by comparing prediction precision (RMSE – root mean square error) and bias (ME –mean error) using the EM38-v transect data on various contrived transect spacing (i.e. 48, 96, 144, 192, 240 and 288 m) and comparing them with ordinary kriging (OK) prediction of the 33 calibration sites. In terms of RMSE, the use of the 48 (2.04), 96 (2.08) and 144 m (2.40) spacing were optimal. In terms of ME the use of the 96 (0.03) and 48 m (-0.11) spacing are optimal. These results were confirmed when considering the relative improvement (RI) in prediction, whereby RI is greatest when the 48 (20.15 %), 96 (18.57 %) and 144 m (5.97 %) spacing were employed. The mean rank (Ri) and standard deviation (Si) of rank also suggest these spacing were optimal. We recommend EM transects of either 96 or 144 m are appropriate at a reconnaissance level (i.e. broad scale farming) and 24 or 48 m spacing are used at smaller levels: especially where detailed information is required for the construction of earthern water reservoirs and supply channels in irrigated areas.

See more from this Division: A08 Integrated Agricultural Systems
See more from this Session: Managing Spatial Variability/Div. A08 Business Meeting