Validation of models for plant disease management is a crucial part in the development of decision support systems in plant protection. Bespoke field trials are usually conducted to determine the performance of a model under practical conditions. However, field trials are very resource-demanding, and the use of already existing field trial data could significantly reduce costs for model validation. In this study, we took this novel approach to verify the performance of models for determining the need of fungicide applications against leaf blotch diseases in wheat by utilising historical weather data and yield data available from fungicide efficacy field trials. Two models based on humidity factors were used in the study. To estimate how specific humidity settings in the two models affect the number of recommended fungicide treatments per season, historical weather data from a 5-year period from weather stations in Denmark, Sweden, Norway, Finland, and Lithuania was used. The model output shows major differences between seasons and regions, typically recommending between one and three treatments per season. To determine the prediction potential of the models, data on yield gains from either one or two fungicide applications in fungicide efficacy trials conducted in wheat over a 5-year period in the five countries was utilised. The yield responses from fungicide treatments in the efficacy trials varied considerably between years and countries, as did the proportion of predictions of profitable treatments. In general, there was a tendency for the models to overestimate the need to apply fungicides (low specificity), but they rarely failed to recommend an application that was needed (high sensitivity). Despite the importance of having specific trials across regions in order to adjust models to local cropping and weather conditions, our study shows that historical weather data and existing field trial data have the potential to be used in model validation.

植物病害管理模型的验证是植物保护决策支持系统开发的关键部分。通常进行定制的现场试验以确定模型在实际条件下的性能。然而，现场试验非常需要资源，使用现有的现场试验数据可以显着降低模型验证的成本。在这项研究中，我们采用这种新方法来验证模型的性能，该模型通过利用历史天气数据和来自杀菌剂功效田间试验的产量数据来确定小麦叶斑病的杀菌剂应用需求。研究中使用了两个基于湿度因素的模型。为了估计两个模型中的特定湿度设置如何影响每个季节推荐的杀菌剂处理次数，使用来自丹麦、瑞典、挪威、芬兰和立陶宛气象站的 5 年历史气象数据。模型输出显示了季节和地区之间的主要差异，通常每个季节推荐一到三种治疗方法。为了确定模型的预测潜力，使用了在五个国家进行的 5 年期间在小麦中进行的杀菌剂功效试验中使用一种或两种杀菌剂的产量增益数据。功效试验中杀菌剂处理的产量反应在年份和国家之间差异很大，盈利处理的预测比例也是如此。一般来说，模型倾向于高估使用杀菌剂的需要（低特异性），但他们很少不推荐所需的应用程序（高灵敏度）。尽管跨地区进行特定试验以根据当地作物和天气条件调整模型很重要，但我们的研究表明，历史天气数据和现有的田间试验数据有可能用于模型验证。