Development Of Crop Water Production Function For Summer Maize Based On CERES-Maize Model

The conventional approaches to get crop water production function(CWPF) through specialized drought experiment and statistical regression method are complicated and time-consuming and have hindered the further research and application of CWPFs. In addition, the static CWPF only gives water sensitive index values specified at each growth stages of a given crop, nonetheless such water sensitivity index usually suffered large temporal and spatial variations. This study used the CERES-Maize model to develop water production function of summer maize. First, field experiment data were used to calibrate the CERES-Maize model; then the calibrated model was used to simulate the different scenarios of water stresses at different development stages of summer maize; finally, a new Jesen function of crop water production of summer maize was developed with the simulated results with historical weather data of 58 years(1955-2012).A water control experiment was carried out under rainout shelter in the growth seasons of 2013 and 2014 to calibrate the model and investigate the influences of water stresses imposed at different development stages on development and grain yields of summer maize in Yanling, Shaanxi Province. The whole growth season of summer maize was divided into four main stages: seeding, jointing, tasseling, and grain filling. Irrigations were applied on three stages while keeping one stage dry, i.e. four scenarios of water stress(D1-D4) were designed during each growth season. There were two irrigation levels(I1=70 mm, I2 =110 mm). Thus, there were a total of eight treatments that had three replicates and followed a split-plot experiment design. An extra treatment with irrigations at all four stages was arranged nearby. With the experiment data, the CERES-Maize model was calibrated and validated. Finally a Jesen function was developed for summer maize with the simulation results with historical weather data. Some main conclusions have been drawn as follows:(1) Water stress at early vegetative stage(before jointing) could seriously affect the growth of summer maize since plant height, LAI, aboveground biomass were all remarkably lower than other treatments. It could also seriously affect the grain yields of summer maize since ear length and diameter were lower and bald length was longer. Water stress at early vegetative stage(before jointing) could extend the growing season of summer maize. The maximum delay of tasseling date could be six days when water stress occurred at seeding or jointing stage. Water stress at reproductive stage(after tasseling) could shorten the growing season of summer maize. Water stress at tasseling or grain filling stage could advance maturation date by two days. In addition, different irrigation levels could also influence the length of growing season of summer maize. The number of kernel rows of summer maize was not sensitive to water stress since drought periods and irrigation levels had no significant influence on it. Different irrigation levels on water stress stages had strong interactions on grain yield of summer maize. Grain yield was relatively low when there was water stress before jointing. Water stress at tasseling remarkably reduced grain number per ear, but the grain yield was the highest among the treatments with the same irrigation level. Water stress at grain filling stage not only obviously reduced hundred-kernel weight and grain yield, but also decreased water use efficiency of summer maize since much water was wasted. In general, water stress should be avoided at the vegetative growth stages(seedling and jointing) when under a lower irrigation level to prevent large loss of grain yield. However, when under a higher irrigation level, limited irrigation could be applied at the seedling stage but water stress should be avoided at the grain filling stage to guarantee a higher yield.(2) The CERES-Maize model can precisely simulate the growth and yield of summer maize under the condition of sufficient irrigation. The average absolute relative error(ARE) and relative root mean square error(RRMSE) between simulations and observations of phenology dates, yield, grain weight, and aboveground biomass were between 6% and 8%. The simulated dynamic trends of the aboveground biomass, LAI and plant height were the same as measured ones, with determination coefficient(R2) and Consistency index(d) of 0.929-0.982 and 0.88-0.98, respectively. The simulations of soil moisture contents and cumulative evaporation transpiration were also accurate. There were still some shortcomings in the CERES-Maize model to simulate the growth and yield of summer maize under water stress. This model cannot simulate the difference of phonological caused by water stresses imposed at different development stages. The simulation of plant growth was not accurate enough with drought at vegetative stages. It underestimated the influence of water stress on grain weight, biomass, and yield when drought occurred at vegetative stage(D1and D2). The simulation of soil moisture also had certain discrepancy. In short, the model can simulate summer maize growth and yield under condition of water stresses imposed at different development stages, but the simulation discrepancy for scenarios of water stress at early stages(D1and D2) could be big and the simulation results should be carefully analyzed.(3) The Jensen water production function of summer maize was derived with the simulation results of CERES-Maize model with historical weather data of 58 years(1955-2012). The result of multivariate linear regression was significant according to F test and multiple correlation coefficient test. The crop water sensitive indices were small at early stage, big at middle stage, and then small again at late stage. The crop water sensitive indices based on the measured data of two years were big at early and late stage,and small at middle stage. This showed that crop water sensitive indices were greatly influenced by climatic conditions, it had a larger interannual variability. The crop water sensitive indices with historical weather data of 58 years(1955-2012) can eliminate effects of different climate conditions, and it had a more general applicability. The new Jensen function can be used to estimate the yield of summer maize in water deficit condition and to evaluate and guide irrigation water management. At the same time, it was proved to be feasible to derive crop water production function with the simulation data of CERES-Maize model. This method can then be promoted to use in other different areas. With a limited amount of field data to calibrate model, the CERES-Maize model can be used to simulate different scenarios of summer maize growth to develop a crop water production function that takes into account climatic variations.