Seasonal weather prediction for Canadian farmers

加拿大农民几乎所有的作物生产决策都是根据短期、中期和长期天气因素做出的。加拿大有许多服务可以提供有用的信息，农民可以使用这些信息来做出基于天气的短期决策。农户依靠气候学的科学知识来生成“气候正常”指标，以满足全季和多季经营规划的需要。然而，在加拿大最重要的农业区，可靠的中期天气预报(有时被称为趋势预报)的可用性存在显著差距。这项研究的目的是发展和改进区域和特定地点的中期天气预报，并弥合短期预报和气候学之间的差距，短期预报通常提供未来10至14天的天气信息，气候学提供目前用于季节长期作物模型交付的长期历史平均值。举一个具体的例子，天气创新(WIN)公司为许多作物生产物候模型。它们提供了作物生命周期阶段的时间预测，使用了一系列输入，包括要为其制作模型的地点、种植日期和田间使用的品种。在生长季节的每一天，作物模型都会使用三种类型的天气文件进行更新。首先，从田间气象站观测到的天气或由附近气象站网络插入的天气提供了该作物自种植日期以来的实际天气体验。其次，短期10至14天的预测被用来更新模型，以预测近期作物发展预期。最后，用气候学方法计算了生长季平衡的气候常态数据。在过去，预测较长期的天气是“平均”的，这是分析作物模型所能提供的最佳结果。然而，气象学的进步现在可以提供足够的中期预报精度来改进模式。这项研究将测试其中一些中期模型和统计过程，以便在短期预测和作物成熟或收获之间的时间段内更准确地预测作物物候。

Canadian farmers base almost all crop production decisions by reflecting on short term, medium term and long range weather factors. There are many services available in Canada that provide useful information that farmers can use for short term weather based decisions. Farmers rely on the science of climatology to generate indexes of "climate normal" to satisfy the need for full season and multi-season business planning purposes. However, there is a significant gap in the availability of reliable, medium range weather forecasts, sometimes referred to as trend forecasts, for most important farming regions in Canada. The purpose of this research is to develop and improve regional and site-specific medium range weather forecasts and bridge the gap between short term forecasts, which usually provide weather information for the next ten to fourteen days, and climatology which provides the long term historical averages that are currently used for season long crop model delivery. For a specific example, the Weather INnovations (WIN) company produces phenology models for many crops. These provide a forecast of the timing of crop life cycle stages, using a range of inputs including the location, the planting date and the cultivar used in the field, for which the model is to be produced. Each day in the growing season the crop model is updated using three types of weather files. First, the observed weather from in-field weather stations or interpolated by a network of nearby stations provides that crop's actual weather experience since its planting date. Second, the short term ten to fourteen day forecast is used to update the model for near-term crop development expectations. Lastly, climatology is used for the climate normal data for the balance of the growing season. In the past, expecting the weather to be "average" in the longer term was the best that could be provided to analytical crop models. However, the advances in meteorology can now provide sufficient accuracy in medium range forecast to improve models. This research will test some of those medium range models and statistical processes so crop phenology forecasting can be more accurate in the period between the short term forecast and when crop maturity or harvest occurs.