Water Management Decision Support System Of Cotton Under Drip Irrigation With Plastic Mulching Based On Soil Moisture Monitoring Model

[Objective] To improve the accuracy of monitoring and forecasting irrigation and acceleratethe process of the automatic cotroling of water saving irrigation, we researched and developedWater Management Decision Support System of Cotton under Drip Irrigation with PlasticMulching (WMDSSCDIPM) based on soil moisture monitoring model.[Method]In this study, the time domain Reflectometry (TDR) and frequency domainReflectometry (FDR) sensors were used to monitor soil water content, incorporating withsystems engineering idea, being application multi-disciplinary theory and method, and usingof cotton cultivation theory and technical knowledge. New contents and functions were addedin the WMDSSCDIPM.[Results]The results presented as following:(1) The linear regression models based on the two sensors between voltage values ofsensors and soil water content were established from4species of soil texture. The resultsshowed: For the observed and simulated values of TDR sensor, the relative errors (%) in clay,heavy loam, light loam, and sandy soil were3.39、2.42、5.09、3.94, respectively, and rootmean square error (RMSE) was1.0080、0.5912、1.1470、0.6938in clay, heavy loam, lightloam, and sandy soil, respectively. Both the decision coefficient and consistency coefficientwere close to1, confidence alpha were less than0.05. For the observed and simulated valuesof FDR sensor, the relative errors (%) in clay, heavy loam, light loam, and sandy soil were3.92、4.41、4.77、5.68, respectively, and the RMSE was1.1920、1.0738、1.0741、1.0014inclay, heavy loam, light loam, and sandy soil, respectively. Both the decision coefficient andconsistency coefficient were close to1, confidence alpha were less than0.05. Thus, soil watercontent can be stimulated by FDR and TDR sensors.(2) The model function base on FDR sensor was y=2.1654x2-5.4401x+2（R2=0.9713）for clay soil, y=5.3512x2-33.928x+71.046（R2=0.9683） for heavy loamsoil, y=1.509x2-0.8723x+5.7308（R2=0.9870）for light loam soil, and y=-1.404x2+16.817x-10.619（R2=0.9861）for sandy soil (Note: where y was soil water content, x wasvoltage value of TDR/FDR sensors, the same as following). Meanwhile, The model functionsbase on TDR sensor in clay, heavy loam, light loam, and sandy soils were y=3.7021x2+11.505x+2.8002（R2=0.9874）、y=14.478x2-16.086x+15.83（R2=0.9960）、y=12.038x2-11.449x+14.672（R2=0.9954）、y=-6.2349x2+36.268x-11.842（R2=0.9932）,respectively. The results on precision and sensitivity of FDR and TDR sensors indicated thatTDR sensor had a higher precision than FDR sensor in clay, heavy loam, and sandy soils, butFDR sensor was more precision than TDR sensor in light loam soil.(3) Especially, the configuration mode of current main cultivation mode, which oneplastic (width:205cm) with two drip tapes were applied and planted six lines cotton, were10-66-10-66-10-66cm, drip tapes were in the middle of wide rows. Our results suggested thesite vertical with drip tapes and being about25.3cm away drip emitter in horizontal directioncould be considered as the best monitoring location for TDR sensor.(4) The available soil water content can be calculated by comparing soil water contentbetween the total soil water content value and irrigation threshold. Our data showed thatirrigation threshold of root zone (0-65cm) in clay, loam, and sandy soil was169mm、127.4mm、80.6mm, respectively. If total soil content was lower than irrigation threshold, indicating cotton being in water stress and no need irrigation. If total soil content was higherthan irrigation threshold, we can predict the irrigated time by available soil water contentwithin24hours,48hours, or72hours or even then after72hours future.(5) We designed and developed WMDSSCDIPM based on meteorological data, soilmoisture content, information of crop growth and farmland water balance model. Also, usingsystem decision results and production technicians or cultivation compared with experiencejudgment, decision results along with the change of real time water content changed, andforecast the result is consistent with the basic irrigation date or floating before and after thedate of actual water to computer decision-making of irrigation forecast is more accurate.[Conclusion] Optimized the dynamic monitoring model and irrigation forecast model throughthe acquisition of farmland moisture real-time monitoring data and meteorological data aswell as a variety of related basic data can calculate the decision results. It can provide serviceson network, intelligent, intuitive image cotton monitoring soil moisture and watermanagement for agricultural producers, managers and technical personnel.