linear regression, multivariate statistics, nitrogen, NPS, soil moisture, spatial analysis, visualization
Although soil nitrate nitrogen (N) is a nutrient source for crop, it could be a potential nonpoint pollution source to the environment when its content remains high with an inappropriate management. Soil nitrate N content is affected by various factors, such as cultivation practices, N fertilizer application rate, soil properties, and climatic conditions. Understanding the effects of these factors on soil nitrate N content is necessary for nitrogen management and nonpoint source pollution control. Taking the data measured from 1996 to 1998 in a 25 ha row crop field located in Central Iowa, this paper intended to study the interwoven effects of these factors on soil nitrate N content using multivariate statistical analysis techniques of sample mean plots, a multivariate analysis of variance (MANOVA) model, and a multivariate linear regression model. The inferences made by the sample mean plots and MANOVA model indicate that the effects of these factors are additive, i.e., their main or direct effects are statistically significant but the interaction effects between and among them are insignificant at a 5% significance level. Incorporating these additive effects, a multivariate linear regression model was fitted to the dataset. The residual plots show that the dataset follows an approximate bivariate normal distribution, which is assumed by the MANOVA and multivariate linear regression models. The validation using the field data collected in 1999 indicated that the model explained more than 93% variations exhibited by the measured sublayed-averaged data on soil nitrate N content and soil moisture. However, this model is unable to account for the within-sublayer variations.
BYU ScholarsArchive Citation
"Development of a Multivariate Regression Model for Soil Nitrate Nitrogen Content Prediction*,"
Journal of Spatial Hydrology: Vol. 6
, Article 2.
Available at: https://scholarsarchive.byu.edu/josh/vol6/iss2/2