Precipitation is the major cause of crop yield variation in rainfed agriculture production on the Loess Plateau. As overfertilization is economically and environmentally undesirable, and crop yield and the resulting returns for N input are uncertain when rainfall variability is high, optimizing N management according to the precipitation of fallow season is vital for efficient crop water use and high yield in dryland/rainfed farming systems. Thus, this study established a reference range describing wheat cultivation year types based on fallow season precipitation over 37 years. We conducted an 8-year field experiment using seven different N input rates, to focus on soil water content and its vertical distribution in the 0–300 cm soil layers and to assess the relationship between soil water utilization and yield formation. The final objective was to evaluate how the optimal N input affects crop water consumption, with the aim of maximizing yields. Eight experimental years (2009–2017) were classified as dry and normal years (four of each), with optimum N application rates of 150 and 180 kg N ha⁻¹, respectively. Compared with that used by farmers (210 kg N ha⁻¹), optimum N input saved 26.8% and 12.3% of N fertilizer, increased grain yield by 5.0% and 5.2%, and improved crop water productivity (CWP) by 6.1% and 8.5% in dry and normal years, respectively. Optimized N inputs reduced soil water consumption before the jointing stage and increased the soil water available for grain yield production. Soil water depletion occurred from deeper soil layers with crop development, and soil water storage (SWS) at 80–240 and 200–300 cm depth played an important role in increasing spike number and grain filling, respectively. Furthermore, soil water conservation treatment (increasing rainwater storage via deep plowing at start of the summer fallow season) was unable to eliminate the variability in soil water storage at sowing, and as “drought at sowing” effects on yield were more adverse than “drought in growing season” effects in dryland systems, our results indicate that adjusting N fertilizer inputs based on summer rainfall variation could enhance wheat yield and CWP in the rainfed farming system.

降水是黄土高原雨养农业生产中农作物产量变化的主要原因。由于过量施肥在经济和环境上都是不合需要的，并且在降雨变化较大时，作物的产量和氮素输入的回报不确定，因此，根据休耕季节的降水量优化氮素管理对于有效利用作物水分和旱地/旱地的高产量至关重要。农业系统。因此，本研究建立了一个参考范围，该参考范围基于37年的休耕季节降水来描述小麦栽培年的类型。我们使用7种不同的氮输入速率进行了为期8年的田间试验，重点研究了土壤水分含量及其在0–300 cm土层中的垂直分布，并评估了土壤水分利用与产量形成之间的关系。最终目标是评估最佳氮素输入量如何影响作物耗水量，以最大程度提高产量。八个实验年（2009-2017年）被划分为旱年和正常年（每年四个），最佳氮肥施用量为150和180 kg N ha^-1。与农民使用的相比（210 kg N ha ^-1），在干旱和正常年份，最佳氮输入分别节省了26.8％和12.3％的氮肥，谷物产量分别提高了5.0％和5.2％，作物水分生产率（CWP）提高了6.1％和8.5％。优化的氮输入量可减少拔节前的土壤耗水量，并增加可用于粮食生产的土壤水量。随着作物的生长，土壤深层土壤水分枯竭，而在80–240和200–300 cm深度的土壤水储量（SWS）分别在增加穗数和籽粒灌浆中起重要作用。此外，土壤节水处理（夏季休耕期开始时通过深耕增加雨水蓄积）无法消除播种时土壤蓄水的变化，