Fuzzy mapping of soil fertility — a case study on irrigated riceland in the Philippines

Abstract

Intensive irrigated rice production technologies have changed the input-output balance of many soil nutrients in tropical Asia, but their impact on soil fertility has not been studied adequately at regional scales. To assess the current soil fertility status in irrigated riceland of Nueva Ecija, Philippines, soil samples were collected from 384 farmers' fields in a 19,176 ha area. Block-kriged maps of topsoil properties (0–20 cm depth) were produced. Yields and nutrient uptake were measured for a subset of 63 farmers during the 1994 dry and wet seasons. Logical Boolean and fuzzy membership models for land classification with or without Monte Carlo simulation were compared. Crisp Boolean classification underestimated the potential area with severe soil fertility constraints to high rice yields by 0.2–5.2%. A combination of fuzzy membership functions with Monte Carlo simulation was used to produce maps of membership values for three soil fertility classes and two multivariate soil fertility qualities. Most of the area investigated had a high inherent fertility potential and was rated suitable for intensive rice production, but 97% of the land had limitations due to low soil status of one or more nutrients. Low available K, P, and Zn occurred on 54%, 64% and 63% of the land, respectively, whereas low S status was only found in minor areas. The average P balance in farmers' fields was positive, but both under- and overfertilization with P occurred. Long-term irrigated rice-rice monoculture has led to a significant depletion of soil K reserves in soils of Nueva Ecija. Although K uptake was larger than N uptake in both seasons, most farmers applied little or no K fertilizer. Average net K loss was 38 kg K ha−1 crop−1 and K is now one of the most important constraints to sustaining and increasing rice yields.