From observations to forecasts - Part 9: what is decadal forecasting?
From observations to forecasts - Part 9: what is decadal forecasting?
复制标题
从观察到预测 - 第 9 部分：什么是年代际预测？
作者：
Collins M
2010年底，英国遭遇了初冬的严寒和大雪，导致圣诞旅行混乱，管道爆裂，垃圾收集被取消。当时许多人的想法是，这是可能持续多年的寒冬模式的一部分吗？这或许是一个十年预测或许能够回答的问题？这篇文章试图解释什么是十年预测，并描述其崛起背后的科学。我们还打算揭露一些科学和技术挑战，包括那些已经成功解决的挑战和那些有待未来解决的挑战。十年预报不应与天气预报混为一谈。不可能提前一周左右预测天气的准确细节，所以预测未来十年的天气是不可能的。季节性预报的进步意味着，对世界某些地区来说，准确(确定性)预测下一季发生干旱或洪水的可能性是可能的，特别是在热带辐合带等雨带受到厄尔尼诺等现象强烈影响的地方。但十年预报并不是把这些预报的确定性方面延伸到未来许多个季节。与许多季节性预测一样，十年预测本质上是概率性的。例如，其目的是预测未来十年(或部分时间)与十年平均气温或降雨量相比，天气可能变暖或变冷、变干或变湿的可能性。有可能给出一段指定的多年内的季节平均值的概率，甚至给出寒冷、温暖、潮湿或干燥季节的概率。不过，总的来说，这些信号是微妙的，我们真正能希望的是评估这些概率是如何相对于某个气候概率发生变化的。
Towards the end of 2010 the United Kingdom was gripped by an early winter spell of cold and snow that caused Christmas travel chaos, burst pipes and cancelled rubbish collections. The thought in many minds at the time was is this part of a pattern of cold winters that might continue for many years? Is this perhaps a question that decadal forecasting might be able to answer? This article is an attempt to explain what decadal forecasting is and describe the science behind its rise in prominence. We also aim to expose some of the scientific and technical challenges, including those that have already been tackled with success and those that remain for the future. Decadal forecasts should not be confused with weather forecasts. It is not possible to predict the precise details of weather any more than a week or so in advance, so forecasting weather a decade ahead is impossible. Advances in seasonal forecasting mean that, for some parts of the world, it is possible to be accurate (deterministic) about the chance of, for example, a drought or flood in the coming season, particularly where rain bands such as the Intertropical Convergence Zones are strongly affected by phenomena such as El Niño. But decadal forecasting is not about extending the deterministic aspects of those forecasts many seasons ahead in the future. Like many seasonal forecasts, decadal forecasts are inherently probabilistic in nature. The aim is to predict, for example, the probability that it may be warmer or colder, drier or wetter over the coming decade (or parts of it) relative to the decadelong average temperature or precipitation. It may be possible to produce probabilities for seasonal averages within a specified period of many years, or even give probabilities of runs of cold or warm or wet or dry seasons. In general, though, the signals are subtle and all we can really hope for is to assess how the probabilities are changed with respect to some climatological probability.