Comparing management zone maps to address infertility and sodicity in sugarcane fields

Abstract

In the sugarcane growing area of the Herbert Valley the soil is inherently infertile with low cation exchange capacity (CEC – cmol(+)/kg). It is also characteristically high in exchangeable sodium percentage (ESP%); moderately sodic (6–10%). To manage these issues, the industry developed the Six-Easy-Steps nutrient management guidelines to assist farmers determine suitable rates of fertilisers (e.g. lime) and ameliorants (e.g. gypsum). In this research, we explore the use of proximal sensed data from a digital elevation model, γ-ray (RS-700 spectrometer) and electromagnetic induction (DUALEM-421) instruments to delineate management zones (2–4) by numerical clustering (k-means). We test these digital soil map (DSM) derived zones by calculating mean square prediction error (MSPE) relative to topsoil (0–0.3 m) CEC and ESP using restricted maximum likelihood (REML). These maps were compared with zones based on a traditional soil texture map (k = 3) and field-based delineation (k = 3). The DSM approach delineated maps of zones were more precise given they minimised the within field MSPE for predicting CEC and ESP as compared to either the traditional soil texture map or field-based delineations (k = 3). This was especially the case for the DSM of k = 2 and for CEC (MSPE = 2.20) and the DSM of k = 3 for ESP (5.60). Results also showed that although various proximal sensed data could be used independently, DSM of management zones generated using all sources of proximal sensed data were most accurate. Differential application rates for lime and gypsum could also be recommended using the Six-Easy-Steps nutrient management guidelines with implications with respect to cost discussed.