PeakWeather
收藏资源简介:
PeakWeather是一个高质量的地面气象观测数据集,由瑞士联邦气象和气候办公室的SwissMetNet网络收集。该数据集包括来自302个站点的多样化气象变量,时间分辨率为每10分钟,时间跨度超过8年。数据集还包括从数字高程模型中提取的地形描述符,以提供上下文信息。PeakWeather旨在支持广泛的空间时间任务,包括不同规模的时间序列预测、图结构学习、插值和虚拟传感。该数据集还包括来自当前运行的高分辨率数值天气预测模型的集合预报,作为评估新方法的基准预报。
PeakWeather is a high-quality surface meteorological observation dataset collected by the SwissMetNet network of the Swiss Federal Office of Meteorology and Climatology. This dataset includes diverse meteorological variables from 302 monitoring stations, with a temporal resolution of 10 minutes and a time span of over 8 years. It also includes topographic descriptors extracted from digital elevation models to provide contextual information. PeakWeather aims to support a wide range of spatiotemporal tasks, including time series forecasting at various scales, graph structure learning, interpolation, and virtual sensing. Additionally, the dataset provides ensemble forecasts from currently operational high-resolution numerical weather prediction models, which serve as baseline forecasts for evaluating novel methodologies.




