Crop models have evolved over the past decade to incorporate more soil-related processes. While this may open avenues to support farmers regarding fertilization practices, it also widens the pitfalls related to model parametrization. Open-access georeferenced soil databases are often a solution for modelers to derive soil parameters. However, they can potentially add to model uncertainty depending on database resolution and the variability of the characteristics it contains. Fertimap is an online spatial database recently released in Morocco. In this study, we aim at assessing how Fertimap could support the use of crop model in the rainfed wheat production areas of Morocco. Data including local soil analysis, farmers’ practices, wheat biomass, and yield were collected on 126 farmers’ fields distributed across the rainfed wheat production area in Morocco from 2018 to 2020. Data were first used to parameterize, calibrate, and assess the model, using site-specific data to infer soil parameters. Then, the impact of soil data source on model uncertainty was assessed by rerunning the simulations while using alternatively locally measured soil inputs or inputs extracted from Fertimap. To disentangle the effect of data source from model sensitivity on model outputs, the model’s sensitivity to labile phosphorus, pH, and organic carbon parameters was also tested. The APSIM-wheat model was found to reasonably simulate wheat phenological stages, biomass, and yield. The comparison of model outputs using one or another source of soil data indicated that using Fertimap had no significant effect on the model’s outputs. This study provides the first assessment of the APSIM-wheat model for simulation of widely used wheat cultivars in Moroccan rainfed areas. It is also the first proof of the practical utility of Fertimap database for modeling purposes in Morocco. This preliminary study delivers a robust basis for model-assisted agricultural advising to take off in Morocco.

作物模型在过去十年中不断发展，纳入了更多与土壤相关的过程。虽然这可能会为支持农民施肥实践开辟途径，但它也扩大了与模型参数化相关的陷阱。开放访问的地理参考土壤数据库通常是建模者获取土壤参数的解决方案。但是，它们可能会增加模型的不确定性，具体取决于数据库分辨率及其包含的特征的可变性。Fertimap 是最近在摩洛哥发布的在线空间数据库。在这项研究中，我们旨在评估 Fertimap 如何支持在摩洛哥雨育小麦生产区使用作物模型。数据包括当地土壤分析、农民的做法、小麦生物量、从 2018 年到 2020 年，在摩洛哥雨育小麦生产区的 126 个农民田地中收集了数据。数据首先用于参数化、校准和评估模型，使用特定地点的数据来推断土壤参数。然后，通过重新运行模拟，同时使用本地测量的土壤输入或从 Fertimap 提取的输入，评估土壤数据源对模型不确定性的影响。为了区分数据源与模型敏感性对模型输出的影响，还测试了模型对不稳定磷、pH 和有机碳参数的敏感性。发现 APSIM-wheat 模型可以合理地模拟小麦物候阶段、生物量和产量。使用一种或另一种土壤数据来源对模型输出的比较表明，使用 Fertimap 对模型的输出没有显着影响。本研究首次评估了 APSIM-wheat 模型，用于模拟摩洛哥雨养地区广泛使用的小麦品种。这也是第一个证明 Fertimap 数据库在摩洛哥用于建模目的的实用性。这项初步研究为模型辅助农业建议在摩洛哥起飞提供了坚实的基础。