原文發(fā)表于 2017年7月21日 ，是由英國(guó)氣象信息部門（Met Office Informatics Lab, UK）發(fā)表的。

Authors list ：Rachel Prudden, Niall Robinson, Alberto Arribas , Charles Ewen

In the 1950s, there was a revolution in weather forecasting. Advances in technology made it possible to simulate the atmosphere using dynamical models, quickly and accurately enough to be used for operational forecasts. Dynamical models are now a central part of weather forecasting. Starting from basic physical laws, they make it possible to predict events such as storms before they have even begun to form.

二十世紀(jì)五十年代，天氣預(yù)報(bào)有了革命性的變化。技術(shù)進(jìn)步使我們可以使用模式來(lái)模擬大氣運(yùn)動(dòng)，這種方法在預(yù)報(bào)業(yè)務(wù)中是快速而準(zhǔn)確的。模式直到現(xiàn)在仍是天氣預(yù)報(bào)的核心。通過(guò)基本的物理學(xué)原理，模式可以在暴風(fēng)雨形成之前便做出預(yù)測(cè)。

A crucial challenge in the coming decade will be the integration of direct physical simulations on the one hand, and data-driven approaches on the other. Such a hybrid approach holds many opportunities for weather forecasting, as well as countless other fields.

未來(lái)十年的一個(gè)關(guān)鍵挑戰(zhàn)將是直接物理模擬與數(shù)據(jù)驅(qū)動(dòng)方式融合應(yīng)用。這種混合方式為天氣預(yù)報(bào)以及無(wú)數(shù)其他領(lǐng)域帶來(lái)許多機(jī)會(huì)（可能性）。

From model to outcomes 從模式到結(jié)果

• Localisation and super-resolution (downscaling) 局地和超高分辨率（降尺度）
• Links to the real world 與其他領(lǐng)域結(jié)合

Operational weather models are usually run at a resolution of between 1km and 10km, that is, everything within the same square kilometer is represented by a single grid cell. This resolution is fine enough to capture a wide range of phenomena, but will obviously be unable to capture very localised details.

目前業(yè)務(wù)運(yùn)行的天氣模式的空間分辨率在1公里和10公里之間，這意味著在這個(gè)分辨率網(wǎng)格內(nèi)只有一個(gè)值。這個(gè)分辨率對(duì)于一個(gè)大尺度的天氣現(xiàn)象是夠用的，但是對(duì)于一些局地性的天氣卻是不夠的。

It may be possible to perform this kind of localisation using models trained on historical data, providing a mapping between the large-scale predictions of the simulation and the small-scale effects. This is an area of active research which could make forecasts more useful for day-to-day activities.

可以嘗試使用歷史數(shù)據(jù)訓(xùn)練的模型（機(jī)器學(xué)習(xí)的方法）來(lái)預(yù)測(cè)局地效應(yīng)，之后建立一個(gè)大尺度模型預(yù)測(cè)與小規(guī)模效應(yīng)之間的映射關(guān)系。此類研究現(xiàn)在非常活躍，有助于提升天氣預(yù)測(cè)對(duì)日常活動(dòng)的價(jià)值。

As well as predicting weather at finer scales, similar techniques could help to link weather forecasts with their broader impacts. Many things are affected by the weather, either directly or indirectly; these include traffic, hayfever, flight delays, and hospital admissions. While some effects may not be easy to simulate, using data-driven models could help to provide advance warning of significant impacts.

除了在更細(xì)微的尺度上預(yù)測(cè)天氣，類似的技術(shù)可以幫助將天氣預(yù)報(bào)與更廣泛的領(lǐng)域聯(lián)系起來(lái)。許多事情直接或間接地受到天氣的影響，包括交通、花粉過(guò)敏、飛行延誤和住院率，這些事情不容易通過(guò)模型來(lái)推理，但可以使用數(shù)據(jù)驅(qū)動(dòng)的模型來(lái)預(yù)測(cè)進(jìn)而提供預(yù)警。

Emulation

• Faster components (emulation) 局部加速
• Hybrid models 混合模式

Once a machine learning model has been trained, it is often much faster to run than a full simulation. This is the motivation for a technique called model emulation. The idea is to build a fast statistical model which closely approximates a far more expensive simulation. Emulators are already being applied to problems such as climate sensitivity. An area of current interest is using the same tools to speed up some components of the weather model.

機(jī)器學(xué)習(xí)模型一旦被建立，通常是要比完整的數(shù)值模擬工程要快。可以使用一種模式仿真（model emulation）的方法，建立一個(gè)非常接近于數(shù)值模式的統(tǒng)計(jì)學(xué)模型，這種方法已經(jīng)應(yīng)用于氣候敏感性研究。現(xiàn)在比較熱的領(lǐng)域是使用機(jī)器學(xué)習(xí)工具加速天氣模式的部分 組件。

There are some aspects of weather prediction which require a full physical simulation; this is what lets you predict unseen events with confidence. Other places this is not possible or even justified, and a statistical approximation may be the best you can do. This second case is where emulation can be useful in operational forecasting.

天氣預(yù)測(cè)中的一些場(chǎng)景是需要通過(guò)大氣物理模式來(lái)實(shí)現(xiàn)，但有些場(chǎng)景使用模式卻是不可能或不合理的，這些場(chǎng)景下使用統(tǒng)計(jì)學(xué)趨近是最好的選擇，模式仿真（model emulation）在預(yù)報(bào)業(yè)務(wù)中會(huì)有效果。

Beyond emulators, there is broader potential for hybrid models with both learned and simulated components. Such models would combine data-driven and physically-driven approaches. For example, it may be possible to adapt statistical components of the model to the local terrain, based on previous observations.

除了模式仿真（model emulation），建立融合機(jī)器學(xué)習(xí)與數(shù)值模擬的混合模式也是非常有潛力的。這種混合模型可以融合數(shù)據(jù)驅(qū)動(dòng)和物理驅(qū)動(dòng)兩種方法。比如，在局地地形對(duì)天氣影響方面，可以基于前期觀測(cè)的結(jié)果訓(xùn)練模型，融合到數(shù)值模式中。

Descriptive learning 描述學(xué)習(xí)

• Finding features 特征識(shí)別
• Exploring and summarising 信息匯總

An area where machine learning has made dramatic progress is feature detection. You can see examples of this in apps which not only detect your face, but add glasses and a moustache in real-time.

機(jī)器學(xué)習(xí)取得了顯著進(jìn)步的一個(gè)領(lǐng)域是特征檢測(cè)。一些基于機(jī)器學(xué)習(xí)的應(yīng)用程序不僅可以檢測(cè)到您的臉部，還可以實(shí)時(shí)在臉上添加眼鏡和胡子。

There is currently a lot of interest in applying similar methods to hazard detection, especially to storm tracking. Trained experts are able to recognise storms and trace their paths from weather imagery; in principle there is no reason an algorithm could not learn to do the same.

目前有很多研究在使用類似的方法做災(zāi)害監(jiān)測(cè)，特別是風(fēng)暴跟蹤。訓(xùn)練有素的專家能夠識(shí)別風(fēng)暴，并從天氣圖像中追蹤路徑，理論上算法也可以做得到。

Another application could address the challenges posed by data volume and complexity when dealing with data from physical simulations. The fields output by such models are highly multidimensional; making sense of them is a complex task, requiring many “screens” of information. An algorithm which could summarise the salient features and bring them to the forecaster’s attention would help streamline this task.

預(yù)報(bào)員在使用觀測(cè)數(shù)據(jù)和數(shù)值預(yù)報(bào)結(jié)果時(shí)，需要處理大量的多維度的數(shù)據(jù)，理解這些數(shù)據(jù)是一項(xiàng)復(fù)雜的工作，經(jīng)常需要切換多個(gè)屏幕來(lái)查閱信息。通過(guò)算法可以自動(dòng)識(shí)別這些數(shù)據(jù)中的關(guān)鍵信息，然后匯總到預(yù)報(bào)員的桌面，從而簡(jiǎn)化這項(xiàng)工作。

Summary 總結(jié)

Exploring combinations of machine learning and numerical simulation is an area of great interest and promise for the Met Office. Not only does it offer an advance in scientific capability, but the challenges arising from the attempt could drive new research in the field of machine learning. This article has given an outline of a few research directions within meteorology, but a similar story holds across a range of scientific disciplines.

探索機(jī)器學(xué)習(xí)和數(shù)值模擬的組合是 Met Office 非常感興趣且抱有期望的領(lǐng)域。它不僅促進(jìn)了預(yù)報(bào)能力的進(jìn)步，而且可能會(huì)推動(dòng)機(jī)器學(xué)習(xí)領(lǐng)域的新研究。本文概述了氣象學(xué)中的一些研究方向，在其他科學(xué)學(xué)科中，機(jī)器學(xué)習(xí)的應(yīng)用的方向與本文所述類似。

總結(jié)

以上是生活随笔為你收集整理的在天气预报中应用机器学习的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問(wèn)題。

如果覺(jué)得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。