Future cotton lint yield in such a large field is linked to topographic features. The predicted cotton yield varies with distance across the field (Fig. 5), depending on soil water supply and elevation. High cotton lint yield is predicted on the central part of the field, where low elevation could favor accumulation of water and nutrients and reduce loss of water through evaporation. Low cotton lint yield will be on the slope and the plateau, where the soil is susceptible to loss of water, organic matter and nutrients through erosion. The nature of the deterministic (effect of irrigation levels), stochastic (effect of topographic features), and the random component of the cotton lint yield was described. The state-space analysis reflects weight of contribution of underlying variable to explain crop yield variability. The state-space models helps us interpret and identify variables that are correlated in space to understand the rate and extent occurring within the landscape and the complex relationship between cotton lint yield and natural field heterogeneity, soil physical and chemical properties, and irrigation and fertilization practices.

Fig. 5. Estimated and measured cotton lint yield along the 50% ET water level (A) and the 76% ET water level (B). Also shown are the state-space models with the 95% confidence interval.

Objective 2:   Develop and evaluate instrumentation and software to measure and analyze variability in crop production and plant response to that variability.

Multispectral Remote Sensing Related to Water and Nitrogen Use in Irrigated Cotton. Assessment of real-time crop and soil conditions using remotely sensed data promises to realize site-specific water and N application in large fields under semiarid conditions. Plant/soil reflectance and spectral vegetation index have been used in characterizing soil, water, nutrient, and plant development conditions, in forecasting crop yield, and in making day-to-day farm management decisions in irrigation and fertilization. The objectives of this study were (i) measurement of cotton/soil reflectance related to specific irrigation and N fertilization, (ii) determination of cotton/soil spectral and agronomic characteristics, (iii) assessment of variability in N status and lint yield across large cotton field, and (iv) to forecast irrigation and N fertilization using spectral vegetation index.

Soil and Cotton Plant Reflectance. The spectral curves (Fig. 1) show the soil, plant, and water reflectance as a function of wavelength (l) and time (cotton development). Sensor outputs are typically in the green, red, and near infrared portions. In June, red (0.648-0.674 mm) reflectance was higher (Fig. 1), which corresponded to high percentage of exposed soil surface at the early vegetative stage. As cotton grew, the red radiance (soil) decreased and the reflected NIR (0.797-0.829 mm) quickly increased from June to August (Fig. 1). The highest NIR plant reflectance was measured in mid-august at plant maturity. As a result of leaf nutrients being transferred to boll, the red increased and the reflected NIR decreased in September (Fig. 1).

Fig. 1. Multispectral reflectance of a cotton crop in Lamesa, TX during the 1998 growing season.

The spectral curves also map the soil, plant, and water reflectance as a function of irrigation level. Dry soil (Fig. 2) reflected highly in the red, then increased to the mid-infrared (1.523-1.752 mm). The plant absorbs blue and red energy and reflects green and NIR energy. As compared to 50% ET, the 75% ET plots showed the relative higher absorption of blue (0.447-0.476 mm), red and mid infrared (MIR) energy, and higher reflection of green (0.546-0.571 mm) and NIR energy (Fig. 2). In MIR, the reflectance is attributable to plant water content. Therefore, the 75% ET input was more adequate to plant growth.

Fig. 2. Multispectral reflectance for dry soil, 50 and 75% ET at Lamesa, TX during the 1998 growing season.