Improvement of Numerical Weather Prediction for Complex Terrain

。我们的动机是更好地预测与天气有关的灾害，并帮助提高与天气有关的清洁能源(风能和水能)的效率和安全性。为了实现这一目标，我们的研究目标是提高对加拿大西部等复杂山区和沿海地形的数值天气预报技能。 数值天气预报(NWP)是一种使用高性能计算机对大气流体动力学方程进行数值积分的计算工具。用于求解这些方程的大型、公开可用的NWP计算机代码被称为“模型”。没有一个模型是完美的，所以我们对计算机输出进行后处理，以纠正任何系统误差。我们还利用了来自许多模式的NWP预报的“集合”来解释大气的混乱性质。可以对集合预报的分布进行校准，以估计不同预报结果的概率--这是能源资源最佳管理的重要信息。 我们的以下活动将有助于实现我们的目标。灾害性天气事件往往相对罕见且难以预测，因此我们将改进利用过去类似罕见事件的“模拟天气集合”的方法。水电水库和河流洪水预报依赖于不完善的水文模型，因此我们将使用一组NWP模型的输出来驱动一组水文模型，以提高技能。高分辨率天气预报将应用于加拿大西部部署的智能电表的数据，并将得到改善。我们将测试一种新型的遗传编程是否可以帮助预测风速的急剧变化，从而调节风力发电。我们将改进数值预报模式，提高山区预报的分辨率，以便更好地预报陡峭狭窄山谷交通走廊沿线的降雨引发的山体滑坡、泥石流、铁路冲刷、局部洪水和雪崩。这项工作将有助于拯救生命，降低能源成本，使交通更加可靠，并帮助加拿大的工业蓬勃发展。

. Our motivation is to better predict weather-related disasters, and to help enhance the efficiency and safety of weather-related clean energy (wind and hydro). To achieve this, our research goal is to improve the skill of numerical weather prediction for complex mountainous and coastal terrains such as in Western Canada. Numerical weather prediction (NWP) is a computational tool that uses high-performance computers to numerically integrate the atmospheric fluid-dynamics equations. The large, publically available NWP computer codes that solve these equations are called "models". No model is perfect, so we post-process the computer output to correct any systematic errors. We also utilize an "ensemble" of NWP forecasts from many models to account for the chaotic nature of the atmosphere. The spread of the ensemble forecasts can be calibrated to estimate the probability of different forecast outcomes -- important information for the optimum management of energy resources. Our following activities will help achieve our goals. Hazardous weather events are often relatively rare and hard to forecast, so we will improve ways to utilize "analog weather ensembles" from past rare events that were similar. Hydro reservoirs and river flood forecasts rely on imperfect hydrologic models, so we will drive an ensemble of hydrologic models with output from an ensemble of NWP models to improve skill. High-resolution weather forecasts will be applied to, and will be improved by, data from smart electric meters being deployed in W. Canada. We will test whether a new type of genetic programming can help predict sharp changes in wind speed that modulate wind-power production. We will enhance an NWP model to enable finer-resolution forecasts in mountainous terrain for better prediction of rainfall-triggered landslides, debris flows, rail washouts, local flooding, and snow avalanches along transportation corridors in steep narrow mountain valleys. This work will help save lives, reduce energy costs, make transportation more reliable, and help Canadian industries to thrive.