Integration of remote sensing and crop growth modeling for nitrogen management decision support in corn

Date
2006-01-01
Authors
Thorp, Kelly
Major Professor
Advisor
William D. Batchelor
Brian L. Steward
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Journal Issue
Series
Department
Agricultural and Biosystems Engineering
Abstract

This dissertation describes efforts to move toward a completely integrated remote sensing and crop growth modeling tool for developing precision nitrogen management recommendations for corn. Aerial hyperspectral remote sensing imagery collected throughout the 2004 growing season was used to estimate corn plant stand density, and a machine vision system was used to map corn population on the ground. Multiple linear regression analysis was used to assess the ability of all combinations of three reflectance bands to estimate corn plant population at resolutions of 2 m, 6 m, and 10 m. Coefficients of multiple determination of up to 0.82 were achieved in this endeavor. Although some limitations apply, remote sensing can be used as a tool to provide corn plant population inputs for crop growth simulations. A cross validation technique and bivariate confidence ellipses were used to evaluate CERES Maize simulations of spatial corn yield variability across an Iowa cornfield. Results indicated that the model performed most poorly when using the wettest or driest growing seasons to validate the model, because the model parameters fitted under the conditions of moderate growing seasons were less flexible for simulating yield in growing seasons with more extreme weather. Results also indicated that topography affects the model performance spatially. CERES-Maize was also used to simulate yield and unused nitrogen remaining in the soil at harvest for a sequence of historical weather data. Simulations were run for 13 spring-applied nitrogen rates over a cornfield divided into 100 0.2 ha grid cells. A methodology based on cumulative probability distributions was then developed to use model output for assessing the link between yield and nitrogen left behind for various nitrogen rates in each grid cell. This methodology can be used to develop precision nitrogen management strategies that address both the economic and environmental concerns of nitrogen management practices. Although the three projects in this dissertation furthered the development of remote sensing, crop growth modeling, and decision support technologies, more work is required to obtain a completely integrated tool for development of precision nitrogen management strategies in midwestern cornfields.

Comments
Description
Keywords
Citation
Source