A COMPARISON OF THREE LEVELS OF AGGREGATION OF AVAILABLE SOIL WATER INFORMATION FOR USE IN A LARGE AREA CROP YIELD PREDICTION MODEL

CHRISTOPHER EDWARD SEUBERT, Purdue University

Abstract

Crop yield prediction models are an important component in accurate crop yield prediction. The objective of this research was to assess the contribution of soils information to the prediction accuracy of a large area crop yield prediction model. The available water holding capacity, defined as the soil water held between 0.3 and 15 bars tension, was used as the major component to study the interaction of soil water holding capacity and weather variability. The available soil water holding capacity was estimated from soil texture for 902 soil profiles representative of Indiana soils. A multivariate cluster analysis of the available soil water holding capacity in ten 15-cm layers was used to classify the soil profiles in 4, 8, and 12 cluster classes. These cluster classes, which represent three levels of aggregation of available soil water holding capacity, were compared to each other in terms of available water holding capacity, areal extent in the nine Indiana crop reporting districts, and their contributions to improving the accuracy of a large area soybean yield prediction model. These analyses indicate: (1) distinctly different available soil water holding capacity groups can be distinguished in a population of soils with a wide range, (2) most of the soybean area in a crop reporting district can be represented by five soil water holding capacity groups, although these groups vary from one district to another, (3) the soybean yield prediction model weather indices are not sensitive to the water holding capacity differences between all of the soil water holding capacity groups at all levels of aggregation, but important differences between poorly and well and high and low water holding capacity soils do exist.

Degree

Ph.D.

Subject Area

Agronomy

COinS