SWEEP: Space Weather Empirical Ensemble Package

太阳的大气层以太阳风的形式流入行星际空间。太阳爆发导致这种流动的巨大变化，可能对地球、经济和社会造成巨大破坏。这种干扰可以通过对恶劣空间天气的预警来减少，本项目的主要目的是通过预报来改进这一预警系统。空间天气预报是一个相对较新的领域，英国气象局有一个专门的部门为各种组织提供预报和警报。这些预测是基于望远镜对太阳及其延伸的大气层（日冕）的测量。这些数据被持续监测，以防可能影响地球的大型火山喷发事件。这些数据还驱动了太阳风的大规模计算模型，给出了地球状况的预测。这个项目的任务是提供一个与当前系统一起运行的新系统，并提供某些关键改进。拟议的项目被称为空间天气经验集合包（SWEEP）。SWEEP从多个观测太阳和日冕的望远镜获取数据，并使用新颖的方法创建日冕磁场、日冕密度和日冕物质抛射（cme）的多个三维地图。因此，基于不同类型的数据和方法，存在多个独立的信息流。这些不同的流被用来驱动一个高效的太阳风模型，在地球上提供多种预测。SWEEP将把这些多重预测与最近对近地太阳风的测量结果进行比较，计算出一个“技能分数”，这是对每种预测表现的衡量标准。在不同的条件下，某些预报流可能比其他预报流表现得更好。然后使用技能分数将各个预测流合并为最终的主要预测。这种方法具有模块化的优点。如果一个望远镜没有收集到数据，其他的预报流将继续运行——因此系统是稳健的。SWEEP使用了一种高效的太阳风模型——比目前的模型快几个数量级。这允许采用“集合”方法进行预测。我们可以运行这个模型几百次，并且对每一次在太阳附近输入模型的数据应用随机的小变化。这导致了地球附近的大量预报。然后，最后的输出可以作为最可能的预测给出，并对预测的不确定性进行估计。与地球天气预报不同的是，有几十个或几百个气象站直接对地球大气进行采样，而观测太阳和日冕的望远镜很少，而且价格昂贵。太阳大气的性质必须从望远镜的远程数据中间接推断出来。我们可以从这些数据中提取的任何新信息都是有价值的，并且可以在不花费建造新望远镜的情况下提高当前的预测性能。SWEEP利用了几种新的日冕分析工具和模型，其中一些以前没有用于预测。SWEEP将是一个完全自动化的系统，这是对现有系统的又一改进。模块化的基础也允许未来的发展——在未来几年新的望远镜上线时，可以直接添加其他预测流，或者合并新开发的数据分析方法或模型。这个最初为期2.5年的项目致力于将现有工具调整和组合成一个功能齐全的SWEEP。到项目结束时，它将在英国气象局全面投入使用。在接下来的几十年里，我们希望SWEEP将成为一个关键的预测工具，帮助保护社会免受空间天气的损害。

The Sun's atmosphere flows into interplanetary space as the solar wind. Solar eruptions cause large variations in this flow that can cause huge disruption to Earth, economy and society. This disruption can be reduced given advance warning of severe space weather, and the main purpose of this project is to improve this warning system through forecasts.Space weather forecasting is a relatively new field, and the UK Met Office has a dedicated unit providing forecasts and alerts for various organisations. The forecasts are based on telescope measurements of the Sun and it's extended atmosphere (the corona). These data are continually monitored for large eruptive events which may impact Earth. The data also drive large-scale computational models of the solar wind, giving a forecast of conditions at Earth. This project is tasked with providing a new system that will run alongside the current system, and provide certain key improvements.The proposed project is called Space Weather Empirical Ensemble Package (SWEEP). SWEEP takes in data from multiple telescopes observing the Sun and the corona, and uses novel methods to create multiple three-dimensional maps of the coronal magnetic field, the coronal density, and of coronal mass ejections (CMEs). Thus there are multiple, independent streams of information based on different types of data and methods. These multiple streams are used to drive a highly-efficient model of the solar wind, giving multiple forecasts at Earth. SWEEP will compare these multiple forecasts against recent measurements of the solar wind near Earth, calculating a 'skill score', which is a measure of each forecast's performance. Certain forecast streams may perform better than others under different conditions. The skill scores are then used to combine the individual forecast streams into a final main forecast. This approach has the advantage of being modular. If one telescope fails to collect data, the other forecast streams continue to operate - the system is therefore robust.SWEEP uses a highly-efficient solar wind model - several orders of magnitude faster than the current model. This allows an 'ensemble' approach to forecasting. We can run the model several hundred times, and for each instance apply random small variations on the data fed into the model near the Sun. This leads to a large set, or an ensemble, of forecasts near Earth. The final output can then be given as the most probable forecast, with an estimate of the forecast uncertainty. Unlike Earth weather forecasting, where there are dozens, or hundreds, of weather stations sampling Earth's atmosphere directly, telescopes that observe the Sun and corona are few and expensive. The properties of the Sun's atmosphere must be indirectly inferred from the telescope remote data. Any new information that we can extract from these data is valuable, and may improve current forecasting performance without the cost of building new telescopes. SWEEP exploits several new analysis tools and models of the solar corona, some of which have not previously been used for forecasting. SWEEP will be a fully automated system, which is another improvement from the current system. The modular basis also allows for future developments - it will be straightforward to add other forecast streams as new telescopes come online in future years, or to incorporate newly-developed data analysis methods or models.This initial 2.5 year project is dedicated to adapting and combining existing tools into a fully functional SWEEP. By the end of the project it will be fully operational at the Met Office. Over the next few decades, we hope that SWEEP will be a key forecasting tool that helps protect society from the damage of space weather.