| China is a populous country,with great food demand.Irrigation and nitrogen（N）fertilization are important assurances for high and stable grain yield.Unreasonable extensive application of irrigation and N fertilizers lead to significant waste of resources and economic losses.Though the traditional experimental trials have been proved as well-established means to evaluate different irrigation and N fertilization scheduling,they are usually costly and time consuming.Crop growth simulation models could simulate the collective effects of meteorological conditions,soil properties,cultivar characteristics,and management practices on crop growth,development,and yield.They have been widely used on crop water and N management.Based on the sensitivity analysis of crop cultivar and ecotype parameters,the field data of an irrigation and N fertilization experiment of winter wheat which was conducted at waater-saving irrigation station of Northwest A&F University in 2011-2014 was used to calibrate and validate the DSSAT-CERES-Wheat model,which was used then to analysize the uncertainty of weather input on yield prediction,to explore water-saving irrigation system and N-saving fertilization mode matching with the local soil and weather conditions.The irrigation scheduling based on both growth-stage（GS）and available soil water（ASW）while the N application optimization aimed at high yield or net profit.Finally,the uncertainty of irrigation and N optimization using long-term historical weather data was analysed,in order to provide scientific basis and technical support for high yield and efficiency and sustainable development of wheat production in Guanzhong Plain.The main results were as follows:（1）The water and N stresses affected the sensitivity of the various model outputs to crop parametersThe most influencing parameters for dry matter accumulation,leaf area index（LAI）,aboveground N uptake,grain yield and evaportranspiration were PARUE（photosynthesis available radiation use efficiency during vegetative stage）,P1（accumulated temperature in the duration of emergence stage to terminal spike differentiation stage）,P1D（photoperiod response）,and LSPHS（timing of the final leaf start to senesce）whose importance increased much during grain-filling stage.Additionally,the various model outputs were sensitive to some specific parameters,e.g.dry matter accumulation to PARU2（photosynthesis available radiation use efficiency during productative stage）,LAI and aboveground N uptake to PHINT（interval between successive leaf appearances）,evaportranspiration to RDGS（root depth growth rate during early growth stage）,grain yield to G2（standard kernel size）and G1（kernel number per unit canopy weight at anthesis）.The water and N stresses increase the sensitivity of RDGS and GNMN（minimum grain N）.（2）CERES-Wheat model could simulate crop yield and soil water danamics under various water and N treatments wellThe PEST software was used to aotumatically calibrate the sensitive parameters and the DSSAT-CERES-Wheat was then validated by independent data in 2012-2013.The results showed that:the model could basically reflect the phenological stages in different years but not accumulately simulate the effect of different water and N on the developmental stages of winter wheat.The yield and seed number were simulated accurately with nRMSE less than 20%.The absolute simulation bias of grain weight was small,while the difference between treatments could not be reflected reasonably.The biomass accumulation and LAI expansion with higher N application rate were simulated much better than that under N stress and the simulated bias was much larger for LAI than biamass,indicated that the growth process could be improved under water and N stress,especially for LAI.The model could simulate soil water content in different layer under various treatments with nRMSE smaller than 20%.Although the soil water content in 80-100 cm was simulated with slight higher error,the total soil water storage within 1 m was accurately simulated.（3）Yield prediction was influenced by both the weather variability and crop production potential levelThe effect of certain weather conditions on crop production is different in various production levels.Under potential yield level,the 1℃increase of air temperature promoted the anthesis and maturity date by 4 and 5 days,respectively,improved yield by 7%with excess temperature reducing yield.The 10%increase of solar radiation had no effect on phenological stages but could increase yield by 10.7%.By contrast,under water-limited yield level,the radiation and rainfall were the main factors restricting the wheat yield.The 10%bias of solar radiation and precipitation could result in 9%and10%yield change,respectively.As the yield improved,the response of yield to solar radiation and precipitation enlarged,indicating that in low-yielding year,the uneven precipitation distribution was the main reason causing water stress and wheat yield loss.Long-term simulation demonstrated that nitrogen input enlarged the annual variability of wheat yield generally.Within-season yield prediction showed that,regardless of nitrogen inputs,yield forecasts in the later growing season improved the accuracy and reduced the uncertainty of yield prediction.In a low-yielding year（2011–2012）and a high-yielding year（1991–1992）,the date of acceptable predicted yield was achieved 62 and 65 days prior to wheat maturity,respectively.In a normal-yielding year（1983–1984）,inadequate precipitation after the jointing stage in most historical years led to the underestimation of wheat yield and the date of accurate yield prediction was delayed to 7–22 days prior to maturity for different N inputs.（4）The irrigation schedule optimization considering the annual variability reduced irrigation water and improved irrigation water use efficiencyThe growth-stage based irrigation simulation showed that wintering to booting stages are key to winter wheat yield formation.Concidering high yield,high efficiency,water-saving and labor-saving,the recommendable irrigation schedule was irrigating 60 mm at wintering,jointing and booting stages each time.Yearly optimization of irrigation schedule,considering the variation of weather during each growing season,could reduced irrigation water by 53 mm,frequency by 0.88 times,and improved irrigation water use efficiency 0.83 kg m-3 while keeping the average yield basically stable at 8514 kg hm-2.The available-soil-water based irrigation simulation showed that the optimized irrigation schedule was irrigation lower threshold of 25%and upper threshold of 75%.Yearly optimization of irrigation schedule,could reduced irrigation water by 34 mm,frequency by once,and improved irrigation water use efficiency 0.6 kg m-3while keeping the average yield basically stable at 8572 kg hm-2.（5）The N optimization considering the annual variability increased crop yieldThe optimum N application rate for high yield was affected by water supply during the growing season.Under rainfed condition,yearly optimized N rate varied greatly each year and averaged 103 kg hm-2,which was 3 kg hm-2 lower than the long-term optimized N rate.The mean yield of yearly optimization was 4578 kg hm-2,which was 354 kg hm-2 yr-1 higher than long-term optimization.Considering the annual weather variation,yearly optimization could increase 861￥hm-2 yr-1.The precipitation in March was the most important factor for rainfed yield and optimized N rate and explained 74%and 56%of their variation,respectively.Under irrigated condition,both the crop yield and optimized N were higher than those under rainfed condition.Yearly optimized N rate averaged 137 kg hm-2,which was not less than the long-term optimization.However,the average yield of yearly optimization was 8443 kg hm-2,which means 386 kg hm-2 higher yield and 923￥hm-2 yr-1 more net profit.Full irrigation greatly eliminated the yield uncertainty caused by rainfall variation.Therefore,daily maximum temperature,daily minimum temperature and daily solar radiation were the most important factors for irrigated yield and optimized N rate and explained 43%,18%and 17%of the variation of optimized N rate,respectively.（6）The irrigation schedule optimization considering the annual variability increased economic incomeThe N application rate optimization for high net profit showed that under rainfed conditions,long-term optimized N rate was 101 kg hm-2,which was 5 kg hm-2 less than that for high yield.Considering the annual weather variation,yearly optimization reducing N rate 3 kg hm-2 yr-1 while increasing yield 249 kg hm-2 yr-1 increased net profit 608￥hm-2 yr-1.Under full irrigation,the annual variation of net profit decreased.The long-term optimized N rate was 135kg hm-2,which was 2 kg hm-2 less than that for high yield.Compared to long-term optimization,yearly optimization reduced N rate 4 kg hm-2 yr-1,increased yield 77 kg hm-2 yr-1,which means increasing net profit of 513￥hm-2 yr-1.higher N fertilizer prices could increase the economic loss by excess N application and was one of the potential measures to avoid over-fertilization of nitrogen. |
