Predicting paddy soil productivity

Abstract

Paddy soils are naturally heterogeneous in terms of their physico-chemical properties which affect rice productivity. Currently, uniform application of agricultural inputs on the entire field is not efficient and could result in either insufficient or excess nutrient supply. Good agricultural practices can be achieved if soil and nutrient variations within a farm are considered, and a soil-yield interrelationship is established. Thus, simple and rapid methods to characterize soil properties differences are needed. This study was conducted on two different plots in MARDI research station, Malaysia. The soils were sampled at two depths and analyzed for the physico-chemical properties. Crop cutting test yields were taken at the same soil sampling locations. Georeferenced ECa measurements were obtained by using Veris 3100 cart equipped with a differential global positioning system. The results of correlation analysis showed that the coefficients between yield and soil measurements were generally low and inconsistent for both 9-ha and subsurface drainage plots. On the contrary, when the two plots data were pooled and correlated, the coefficients between yield and soil measurements were high, consistent, and significant. A boundary line using a log-normal function was fitted to the upper edge of data in the scatter plots. Significant relationships between potential grain yield (Ypo) and ECa were detected with r2 > 0.8 in four out of six cropping seasons. Comparison of Ypo and observed yield (Yob) delineated farm areas into different management zone and allows for discriminate fertilizer application, thus avoiding under or overfertilisation.