Seasonal weather prediction for Canadian farmers

加拿大农民几乎所有的作物生产决策都是基于对短期、中期和长期天气因素的考虑。 加拿大有许多服务提供有用的信息，农民可以使用这些信息来做出基于天气的短期决策。农民依靠气候科学来产生“气候正常”指数，以满足全季节和多季节商业规划的需要。然而，在加拿大最重要的农业地区，可靠的中期天气预报（有时称为趋势预报）的可用性存在很大差距。本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（通常提供未来10至14天的天气信息）与气候学（提供目前用于季节作物模型交付的长期历史平均值）之间的差距。举一个具体的例子，Weather Innovations（WIN）公司为许多作物生产物候模型。这些提供了作物生命周期阶段的时间预测，使用一系列输入，包括位置，种植日期和田间使用的品种，模型将被制作。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内插的天气提供了作物自种植日期以来的实际天气经验。第二，短期10至14天的预测是用来更新模型的近期作物发展的预期。最后，气候学用于生长季节平衡的气候正常数据。在过去，预期天气在较长时间内是“平均”的，这是可以提供给分析作物模型的最好的结果。 然而，气象学的进步现在可以为中期预报提供足够的准确性来改进模型。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期内更准确地进行作物物候预测。

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.