Crop N decision support tools are typically based on either empirical relationships that lack mechanistic underpinnings or simulation models that are too complex to use on farms with limited input data. We developed an N mineralization model for corn that lies between these endpoints; it includes a mechanistic model structure reflecting microbial and texture controls on N mineralization but requires just a few simple inputs: soil texture soil C and N concentration and cover crop N content and carbon to nitgrogen ratio (C/N). We evaluated a previous version of the model with an independent dataset to determine the accuracy in predictions of unfertilized corn (Zea mays L.) yield across a wider range of soil texture, cover crop, and growing season precipitation conditions. We tested three assumptions used in the original model: (1) soil C/N is equal to 10, (2) yield does not need to be adjusted for growing season precipitation, and (3) sand content controls humification efficiency (ε). The best new model used measured values for soil C/N, had a summertime precipitation adjustment, and included both sand and clay content as predictors of ε (root mean square error [RMSE] = 1.43 Mg ha−1; r2 = 0.69). In the new model, clay has a stronger influence than sand on ε, corresponding to lower predicted mineralization rates on fine‐textured soils. The new model had a reasonable validation fit (RMSE = 1.71 Mg ha−1; r2 = 0.56) using an independent dataset. Our results indicate the new model is an improvement over the previous version because it predicts unfertilized corn yield for a wider range of conditions.

中文翻译：

---

改进预测未施肥玉米产量的氮矿化模型

作物氮决策支持工具通常基于缺乏机械基础的经验关系或过于复杂而无法在输入数据有限的农场中使用的模拟模型。我们开发了一个位于这些端点之间的玉米氮矿化模型；它包括一个反映微生物和质地对氮矿化控制的机械模型结构，但只需要一些简单的输入：土壤质地、土壤碳和氮浓度、覆盖作物氮含量以及碳氮比 (C/N)。我们使用独立数据集评估了该模型的早期版本，以确定未受精玉米预测的准确性（玉米L.）在更广泛的土壤质地、覆盖作物和生长季节降水条件下的产量。我们测试了原始模型中使用的三个假设：（1）土壤 C/N 等于 10，（2）产量不需要根据生长季节降水进行调整，（3）沙子含量控制腐殖化效率（ε）。最好的新模型使用土壤 C/N 的测量值，进行夏季降水调整，并包括沙子和粘土含量作为预测因子ε（均方根误差 [RMSE] = 1.43 毫克·公顷−1;r2= 0.69）。在新模型中，粘土比沙子的影响更大ε，对应于细质地土壤的较低预测矿化率。新模型具有合理的验证拟合（RMSE = 1.71 Mg ha−1;r2= 0.56）使用独立数据集。我们的结果表明，新模型比之前的版本有所改进，因为它可以预测更广泛条件下的未受精玉米产量。