Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Spatial Validation of Cotton Simulation Model in Relation to Soils and Multispectral Imagery

Authors
item Igbal, Javed - MISSISSIPPI STATE UNIV
item Whisler, Frank - MISSISSIPPI STATE UNIV
item Read, John
item Thomasson, Alex - MISSISSIPPI STATE UNIV
item Jenkins, Johnie

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: September 1, 2003
Publication Date: October 1, 2003
Citation: Igbal, J., Whisler, F.D., Read, J.J., Thomasson, A., Jenkins, J.N. 2003. Spatial validation of cotton simulation model in relation to soils and multispectral imagery [abstract]. Proceedings of Ecosystems Dynamics, Agricultural Remote Sensing and Modeling, and Site Specific Agriculture. 5153:200-207.

Technical Abstract: Studies were conducted in 1998 and 1999 at Perthshire Farm, MS in a 162-ha field with a 2-m elevation range. The dominant soil series are Commerce silt loam, Robinsonville sandy loam and Souva loam. Objectives were to 1) compare GOSSYM simulated lint yield with actual yield, 2) predict yield from airborne multispectral data, and 3) determine sampling date with best correlation between field yield and remote sensing data. To monitor biophysical condition of the cotton vegetation, airborne multispectral images were collected on 10 dates in 1998 and 17 dates in 1999 from April to September. Plants in 1-m of row were sampled in 12 plots in each of two transects. GOSSYM simulation runs were made for each sampling location (profiles, n=24) and compared to actual crop parameters using the root mean square error (RMSE) statistic. Within a soil-map unit, GOSSYM predicted yield varied from 0.45 to 0.61 bales per acre. A significant correlation (P<0.01) was obtained between yield and normalized difference vegetation index (NDVI) on 5 and 17 July, with drainage areas having lowest NDVI. The NDVI curves of different sites in 1999 showed least distinctiveness due to somewhat greater rainfall as compared to drier weather in 1998. Results suggest multispectral images acquired between 800 and 1500 growing degree days (GDD) 60 are useful tools to monitor crop health and predict yield under normal weather conditions.

Last Modified: 7/28/2014
Footer Content Back to Top of Page