Dynamic prescription maps for site-specific variable rate irrigation of cotton

A prescription map is a set of instructions that controls a variable rate irrigation (VRI) system. These maps, which may be based on prior yield, soil texture, topography, or soil electrical conductivity data, are often manually applied at the beginning of an irrigation season and remain static. The problem with static prescription maps is that they ignore spatiotemporal changes in crop water status. In a two-year study (2012 and 2013), a plant feedback system, including a wireless sensor network of infrared thermometers (IRTs), was used to develop dynamic prescription maps to accomplish adaptive irrigation scheduling for cotton (Gossypium hirsutum L.). One-half of a center pivot field was divided into manually and plant feedback-controlled irrigation treatment plots. Irrigation treatments were at three levels, 75, 50 and 25 percent of full as defined by either replenishment of crop water use to field capacity or by the equivalent threshold of the IRT sensed crop water stress. The system accepted user input to control irrigation for the manual treatment plots (I-75M, I-50M, and I-25M), and calculated and compared a thermal stress index for each plant feedback-controlled treatment plot (I-75C, I-55C and I-25C) with a pre-determined threshold for automated irrigation scheduling. The effectiveness of the plant feedback irrigation scheduling system was evaluated by comparing measured lint yield, crop water use (ETc), and water use efficiency (WUE) with the manually scheduled treatment plots. Results for both years indicated that average lint yields were similar between the manual and plant feedback-control plots at the I-75 level (181 and 182 gm(-2), respectively, in 2012; 115 and 103 g m(-2), respectively, in 2013) and I-50 level (146 and 164 gm(-2), respectively, in 2012; 95 and 117 g m(-2), respectively, in 2013). At the I-25 level, average lint yield was significantly greater for the plant feedback-compared with the manual-control treatment plots (142 gm(-2) and 92 gm(-2), respectively), but the mean amount of irrigation was twice that of the manual-control plots. Mean water use efficiencies (WUE) within the same irrigation treatment levels were similar between methods. Importantly, the automatic plant feedback system did not require the time consuming and expensive manual reading of neutron probe access tubes that was required to schedule the manual treatments. These results demonstrate that the integration of a plant feedback system with a commercial VRI system could be used to control site-specific irrigation management for cotton at higher irrigation treatment levels, i.e., I-75 percent and I-50 percent of full. Such a system can facilitate the use of a VRI system by automating prescription map coding and providing dynamic irrigation control instructions to meet variable crop water needs throughout the irrigation season. As of yet, further research is required to maintain automatic deficit irrigation at a level equivalent to 25 percent replenishment of crop water use relative to field capacity. Published by Elsevier B.V.