An Intercomparison of Skill and Overconfidence/Underconfidence of the Wintertime North Atlantic Oscillation in Multimodel Seasonal Forecasts
An Intercomparison of Skill and Overconfidence/Underconfidence of the Wintertime North Atlantic Oscillation in Multimodel Seasonal Forecasts
复制标题
DOI：
10.1029/2018gl078838
复制
发表时间：
2018-08-16
影响因子：
5.2
通讯作者：
Scaife, A. A.
中科院分区：
文献类型：
--
作者：
Baker, L. H.;Shaffrey, L. C.;Scaife, A. A.
最近对各个季节预报系统的研究表明，冬季北大西洋涛动(NAO)可以巧妙地进行预报。然而，也有人提出，这些巧妙的预测往往信心不足，这意味着预测中存在太高比例的不可预测的噪音。我们评估了季节预报系统对欧洲季节到年际预报(EUROSIP)多模式集成系统的技能和过度自信/信心不足的贡献。在所研究的七个系统中，有五个系统具有在2至4个月的提前期预测冬季NAO的重要技能。这些熟练的系统中有四个对预测NAO信心不足。多模合唱团(合奏人数为126人)既有技巧，又明显缺乏自信。随着整体规模的增加，自信心不足变得更加明显。所有或大多数模型都很好地预测了后转播期的某些年份。这意味着北大西洋涛动的共同遥相关和驱动因素正在被EUROSIP季节预报所捕捉到。本文简单地比较了七个季节预报系统，特别是冬季北大西洋涛动(NAO)。冬季NAO是英国和欧洲冬季天气变化的主要驱动因素，能够预测未来几个季节的NAO对农业、能源、卫生、交通和水资源管理等许多不同部门都有潜在的好处。我们发现，在所研究的七个系统中，有五个系统可以熟练地预测NAO，而多模式集成的技能甚至更高。这些熟练的系统中有四个被发现信心不足，这意味着模型中存在太高比例的不可预测的噪音。信心不足会使充分利用预测技能变得更加困难。然而，有一个系统是巧妙的，但并不缺乏信心。我们还发现，在共同的年份中，所有熟练的系统都能很好地预报NAO。这是一个重要的结果，因为它意味着NAO可预测性的共同驱动因素正在被这些系统捕获。这些结果对我们理解季节预报系统和NAO的可预报性有重要贡献。
Recent studies of individual seasonal forecast systems have shown that the wintertime North Atlantic Oscillation (NAO) can be skillfully forecast. However, it has also been suggested that these skillful forecasts tend to be underconfident, meaning that there is too high a proportion of unpredictable noise in the forecasts. We assess the skill and overconfidence/underconfidence of the seasonal forecast systems contributing to the EUROpean Seasonal to Interannual Prediction (EUROSIP) multimodel ensemble system. Five of the seven systems studied have significant skill for forecasting the wintertime NAO at 2- to 4-month lead times. Four of these skillful systems are underconfident for forecasting the NAO. A multimodel ensemble (ensemble size 126 members) is both skillful and clearly underconfident. Underconfidence becomes more pronounced as the ensemble size increases. Certain years in the hindcast period are well forecast by all or most models. This implies that common teleconnections and drivers of the NAO are being captured by the EUROSIP seasonal forecasts.Plain Language Summary In this paper we provide an intercomparison of seven seasonal forecast systems, with particular focus on the wintertime North Atlantic Oscillation (NAO). The wintertime NAO is the main driver of winter weather variability in the United Kingdom and Europe, and being able to forecast the NAO for the season ahead has potential benefits for many different sectors such as agriculture, energy, health, transport, and water resource management. We show that five of the seven systems studied can skillfully forecast the NAO, and a multimodel ensemble has even higher skill. Four of these skillful systems are found to be underconfident, which means that there is too high a proportion of unpredictable noise in the model. Being underconfident makes it more difficult to fully utilize the skill of a forecast. However, one system is skillful but not underconfident. We also find that there are common years in which the NAO is well forecast by all the skillful systems. This is an important result because it implies that common drivers of NAO predictability are being captured by these systems. These results are an important contribution to our understanding of seasonal forecasts systems and the predictability of the NAO.