TEMPORAL PATTERNS OF COTTON REFLECTANCE AND NDVI-DAYS LINT YIELD MODELING

Authors: LI, HONG - TEXAS A&M; LASCANO, R - TEXAS A&M; BOOKER, J - TEXAS A&M; BRONSON, K - TEXAS A&M; BARNES, EDWARD; WILSON, L - TEXAS A&M; SEGARRA, E - TEXAS A&M

Submitted to: National Cotton Council Beltwide Cotton Conference
Publication Type: Proceedings
Publication Acceptance Date: 10/1/2001
Publication Date: N/A
Citation: N/A

Interpretive Summary: During a crop-growing season, agricultural managers must balance the cost of crop inputs (such as irrigation water and pesticides) with the expected income from the yield at the end of the season. Forecasting yield during the season must incorporate many factors such as soil type, weather conditions and crop variety. Various models have been developed to predict yield; however, the inputs for these models can be quite extensive, and they all require some level of adjustment for particular crop varieties. A less complicated approach was developed by researchers in California to predict cotton lint yield using remotely-sensed data that can be obtained from airborne cameras or satellite sensors. The results of the study indicated the approach could provide acceptable predictions of lint yield if irrigation levels were accounted for. The need to account for irrigation levels in this approach indicates that more research is needed to define limitations for its application to a broad range of conditions. However, the results of this study do show the approach is promising and could ultimately provide farm mangers with a new source of information for crop management decision making.

Technical Abstract: To establish new and improve existing relationships between crop condition and remotely sensed data, multispectral data were collected during a water and N precision agriculture study of cotton in a center pivot irrigated field on the southern High Plains of Texas. In this two-year study(1998-1999) the treatments consisted of irrigation at 50% and 75% calculated cotton potential evapotranspiration (ET) and N application at rates of 0,90 and 135 kg/ha arranged in an incomplete block design. Reflectance data were collected in the field with a hand- held radiometer in eight discrete wavelengths that ranged between 447 and 1752 nm. Cotton reflectance in the near infrared band and normalized difference vegetative index (NDVI)measured during the growing season were positively correlated with soil water content, total N uptake, plant biomass, and lint yield. Based on the significant correlation between NDVI and lint yield, we further attempt to predict lint yield growth using a "NDVI-days" model, where NDVI-days represent the integral of NDVI versus time during the growing season. The NDVI- days model provided good estimates of lint yield each year when the model was calibrated for both the 50% and 75% irrigation treatments.