Abstract This work describes a semi-empirical dynamic model for predicting ammonia volatilization from field-applied slurry. Total volatilization is the sum of first-order transfer from two pools: a "fast" pool representing slurry in direct contact with the atmosphere, and a “slow” one representing fractions less available for emission due to infiltration or other processes. This simple structure is sufficient for reproducing the characteristic course of emission over time. Values for parameters that quantify effects of the following predictor variables on partitioning and transfer rates were estimated from a large data set of emission from cattle and pig slurry (490 field plots in 6 countries from the ALFAM2 database): slurry dry matter, application method, application rate, incorporation (shallow or deep), air temperature, wind speed, and rainfall rate. The effects of acidification were estimated using a smaller dataset. Model predictions generally matched the measured course of emission over time in a reserved data subset used for evaluation, although the model over- or under-estimated emission for many individual plots. Mean error was ca. 12% of applied total ammoniacal nitrogen (and as much as 82% of measured emission) for 72 h cumulative emission, and model efficiency (fraction of observed variation explained by the model) was 0.5–0.7. Most of the explanatory power of the model was related to application method. The magnitude and sign of (apparent) model error varied among countries, highlighting the need to understand why measured emission varies among locations. The new model may be a useful tool for predicting fertilizer efficiency of field-applied slurries, assessing emission factors, and quantifying the impact of mitigation. The model can readily be applied or extended, and is available as an R package (ALFAM2, https://github.com/sashahafner/ALFAM2 ) or a simple spreadsheet ( http://www.alfam.dk ).

中文翻译：

---

用于估算现场应用泥浆中氨挥发的灵活半经验模型

摘要 这项工作描述了一种用于预测现场应用泥浆中氨挥发的半经验动态模型。总挥发是来自两个池的一级转移的总和：“快”池代表与大气直接接触的泥浆，“慢”池代表由于渗透或其他过程而无法排放的部分。这种简单的结构足以再现随时间变化的特征发射过程。量化以下预测变量对分配和转移率影响的参数值是根据牛和猪粪便排放的大型数据集估算的（来自 ALFAM2 数据库的 6 个国家的 490 个现场图）：泥浆干物质、应用方法、施用率、掺入（浅或深）、气温、风速、和降雨率。使用较小的数据集估计酸化的影响。模型预测通常与用于评估的保留数据子集中随时间测量的排放过程相匹配，尽管模型高估或低估了许多单独地块的排放。平均误差约为。应用 12% 的总氨氮（和多达 82% 的测量排放）用于 72 小时累积排放，模型效率（模型解释的观察到的变化的比例）为 0.5-0.7。该模型的大部分解释力与应用方法有关。（表观）模型误差的幅度和符号因国家而异，突出表明需要了解为什么测量的排放因地点而异。新模型可能是预测田间施用泥浆肥料效率的有用工具，评估排放因子，并量化减缓的影响。该模型可以很容易地应用或扩展，并且可以作为 R 包（ALFAM2，https://github.com/sashahafner/ALFAM2）或简单的电子表格（http://www.alfam.dk）使用。