Soil remediation decision making in presence of uncertainty in crop yield response

Abstract

Compacted soil has a characteristic “hard pan” layer that impedes root penetration resulting in reduced crop yield. Soil remediation measures such as deep ripping would improve yield; however, such remediation is relatively expensive. We show how the economic benefit of remediation that is, the expected increase in crop yield can be assessed in the presence of sparse sampling. Geostatistical tools are used for uncertainty characterization. Remediation is recommended when the discounted expected profit due to increased yield exceeds the cost of remediation. A 320×200 m field near Leduc, Alberta, Canada was extensively studied. A penetrometer was used to measure the depth of the compacted soil layer. Elevation and soil characteristics were recorded in 1997 while yield was measured for 3 consecutive years. The goals of data collection were to (1) predict the yield response of remediation of compact soil, and (2) develop an economic model for cost/benefit analysis of remediation in the presence of sparse data. Multivariate statistics combined with geostatistical simulation techniques are used to build maps of all needed variables: elevation, depth to compacted soil, and the local soil landform class. The critical relationship between yield response and remediation is established from the historical yield data. This relationship depends on elevation and local soil landform class. Since deep ripping is expensive, determining the precise areas to remediate can result in substantial savings. An economic model is used with the geostatistical maps to quantify risk and the expected benefit of remediation.