临近预报

Encyclopedie environnement - prevision immediate -

  长期以来,气象雷达和卫星是临近预报主要的输入数据。基于云层和降水区域的移动情况进行外推的方法提供了短期内的预测。如今,高分辨率数值预测已经足够成熟,可以成为即时预测工具的一部分,并能够突破几方面的限制:最大成熟度、可用参数等。近年来,数据融合方法在丰富输出数据、减少不连续性等方面取得了新的突破,被应用于多个方面。

1. 临近预报的发展历史

  “临近预报”包含了常规天气预报的所有内容,预报周期根据定义有所差异,最长可达几个小时。对于临近预报的起源尚存在争议,但毫无疑问的是20世纪60年代雷达观测系统和地球静止轨道卫星的发展使其得以在更大空间尺度上实现。具体来说,地球观测系统提供了高频率的观测数据,使人们得以观测和发现大尺度地理空间范围内气候系统的演变规律。

  近年来,许多因素例如科学合作计划(如 COST 78-European Cooperation in Science and Technology,1983-1988),以及第二代气象卫星 (Meteosat Second Generation)等新一代气象卫星的发射等提高了即时预报的精度。世界气象组织推广了一些即时预报项目并进行展示,特别是北京奥运会(2008年)和温哥华冬奥会(2010年)的合作项目。这篇文章的封面图片就说明了即时天气预报能够改善体育活动观赛体验。

  长期以来,即时预报主要用于危险气候、龙卷风、暴雨和洪水预报等方面。除了“保证货物和人员安全”,即时预报还可以在优化经济成本方面发挥优势。例如,航空业也主要依靠“临近预报”开展运作(特别是预测短期内机场气象变化)。

2. 卫星数据的贡献

  地球静止轨道卫星具有以下几方面的优点:可以广泛地覆盖从热带到南北半球中纬度地区的广大地带,具有高时间分辨率(每5分钟观测一次的第二代气象卫星)和水平分辨率可达到中尺度(第二代气象卫星在欧洲上空的红外波段分辨率为4-5公里,可见光波段具有1公里分辨率)。法国气象局增加了一些第二代气象卫星为临近预报提供服务,这些工作在欧洲气象卫星应用组织领导下的国际合作框架内展开。该合作框架由临近预报卫星应用设施机构(SAFNWC(Satellite Application Facility for Nowcasting))领导。相比于地球静止卫星,低轨道卫星具有更高的空间分辨率,并搭载了具有更宽检测范围的设备。另一方面,对于某一特定区域,低轨道卫星每天有效观测次较少,更可能在地球同步卫星覆盖范围有限的高纬度地区发挥其独特作用。SAFNWC每隔两三年就会发布一个新版本的软件,给用户提供他们感兴趣的应用领域的即时预报产品。在 SAFNWC的框架内,法国气象局负责研发地球静止卫星和对流层数据产品的代码。

  以SAFNWC的RDT(Rapidly Developing Thunderstorm,快速发展的雷暴)产品[1][2]为例。RDT 能够检测、追踪、表征和预测对流单元的运动。从工作原理上看,RDT 基于不同波段的亮度温度值的差异来分析变化趋势。用户也可以使用其他来源的数据;如其他 SAFNWC 产品(云和风产品)、模型数据、来自地面闪电网络(如气象网络)的数据。此外,RDT 给出了观测和预测的对流区轮廓,以及描述对流单元特征的属性,特别是它们的严重性等数据;冷却速率、雷电活动情况、是否存在雷暴云、云的垂直分层等。

环境百科全书-临近预报-2017年2月14日上午10点的气象云图
图1. 2017年2月14日上午10点的气象云图。从GOES-E、METEOSAT 10 (MSG 3)、METEOSAT 7和HIMAWARI-8卫星数据中分析获得的对流层像素信息。左下角用轮廓线(不同颜色对应不同云层的发展阶段)放大了西非区域,黄色线表示历史演化轨迹,箭头表示位移方向。右侧为图例。

  天气预报员使用RDT来补充或替换覆盖率较差(如山区)或没有覆盖(海洋地区)的雷达数据。它还在 HAIC (高海拔冰雪)项目中进行了测试,该项目研究高海拔地区的大气冰核[3],以评估其是否可以作为研究大气高冰核浓度地区的工具[4]。在HAIC项目期间,RDT达到TRL (技术就绪指数)等级的5级水平,表明它已足够可靠成熟。

  现在法国气象局生产的RDT产品能够满足包括航空业用户在内的全球各行业用户的需求。图1说明了RDT的覆盖范围,它是基于多颗具有不同空间和时间分辨率的卫星生成的产品。根据各种卫星的重访周期和可用性开展短期分析或预测(“planned for now” data)从而达到协同和组合使用的目的。

3. 雷达数据的贡献

  与卫星观测相比,气象雷达利用空气柱的最大反射率来确定对流的强度并推断到达地面的降水量。气象服务也使用其他类型的雷达测量;如多普勒效应,它提供了对径向风分量,可识别强烈的剪切力,进行极化测量[5],描述水文气象类型等。

  区域方法允许跟踪大量的降水云层单元,而“面向对象”的方法则跟踪单个云层。通过对过去一段时期内云层的动向的记录,即可将雷达提供的信息用于推断未来一段特定时期的气象状况。

环境百科全书-临近预报-雷达反射率图像上的2PIR位移场示意图
图2. 雷达反射率图像上的2PIR位移场示意图,示意图为2002年5月6日的数据,叠加在基于2PIR 方法计算的位移场上

  2PIR方法(Immediate Precipitation Forecasting by Radar Imaging,雷达成像即时降水预报)始于90年代,并被迅速应用于即时预报,例如对天气高度敏感的体育活动的气象预报。2PIR 方法自面世以来以被用于各行各业,如今被法国气象局用于法国本土和国外未来一小时降雨量的自动化预报。该方法基于雷达反射率,首先通过互相关分析当时观测到的图像与前 10 分钟图像之间的雷达回波位移,从而形成一个规则定义的位移速度场。并基于此,结合最后观察到的反射率进行外推。2PIR 方法的主要局限在于来自雷达数据的限制和地形的影响。

  对于航空业,ASPOC3D 服务(信号和风暴预报在空中交通管制中的应用)的基础是:

  • 闪电观测网络。
  • 雷达网络结合卫星观测可以确定了整个法国的对流分布和云顶高度,(1公里空间分辨率,5分钟时间分辨率)。
环境百科全书-临近预报-雷达和卫星测量反演的对流风险等级分布
图3. 雷达和卫星测量反演的对流风险等级分布。蓝色轮廓对应于相应的严重程度。

  对流单体是由四个层次反射率的空间轮廓定义。这些具有三维属性的信息被转换成对航空有重要意义云的参数,例如:对流云的识别、轮廓特征的表征、云顶高度、云中心最大反射率值、证实云已经到达风暴阶段的闪电等。随后,每个参数被传输给终端用户,并被集成到控制台或模型中。图3显示了这些“对流”对象。

环境百科全书-临近预报-APIC网站页面截图
图4. 2016年10月14日APIC网站页面截图

  在 2010年6月15日法国东南部发生严重洪水之后,为防发生异常局部降雨,当局提供了持续的气象服务。气象服务旨在以一种完全自动的方式为市长提供决策支持以应对洪水。APIC 服务(市级强降雨警报)根据法国雷达网络的数据计算累积降雨量,并与统计数据进行对比以评估观测到的降水情况是否存在异常,根据严重程度可分为两个等级。当被确定为异常等级,或此后降水持续增加时,便向用户发送警告。通过在线地图网站(图 4)可以查看当前的警告、暂时不可用区域以及没有该服务的区域分布。为了论证面向城市普及这种预警服务的可能性,法国气象局基于过去两年中等质量的雷达降水观测数据开展了测试。APIC 服务于2011年12月正式开通,如今已经覆盖了超过92%的城市。

4. 数值天气预报对即时预报的贡献

  目前的即时预报主要基于观测数据进行外推。数值模拟技术的进步和计算性能的提高使得我们可以进一步缩短预报周期,有望克服基于观测外推的方法的局限性(凭空创造的结构和固定不变的强度,过去的传统方法均未考虑到位移或强度演变方面的影响)。

  与此同时,精细的数值模式现在能够逼真地模拟对于临近预报有意义的尺度现象,并与密集的观测进行数据融合。数值预报使得即时预报成为可能:将最新的观测数据进行同化,满足最新预测时限。这些信息的高时效性要求需要它们在一定时间范围内能够实时使用。数值预报模型需要临近预报能够在短时间内提供最新信息,所以我们不得不在预报次数、观测数据的更新和运算时间之间做出取舍。如果模拟的物理过程与精细网格模型的物理过程保持一致,则必须开发专门用于临近预报的数值模型,以应对各种挑战。

  法国气象局开发了AROME-FR精细网格模型(查阅:天气预报介绍),用于创建一个专门用于即时预报的数值模型:AROME-PI,该模型已于2016年3月投入使用。

  AROME-PI 集成了 AROME-FR 的所有特征:二者具有相同的代码版本,相同的3d-var 数据同化方法,相同的空间分辨率(1.3 km)和相同的耦合模型(ARPEGE)。然而,有些属性被调整以满足即时预报的需要。例如,AROMEPI可以实现15分钟时间周期的预报,而 AROME-FR 则为1小时。该项目旨在促进模型在即时预测中的使用以及多源数据与其他预测模型的融合。

  这个模型提供了大量且高效的信息。如今系统已经投入使用,开始为法国气象预报员提供各种综合信息并发出危险预警。

5. 向前迈出了新的一步:数据融合

  数据融合方法[6]预计将在即时预测(卫星,雷达和中尺度数值预测之间的组合方法)中得到了广泛应用。

  例如欧洲气象卫星应用组织临近预报中心,除了卫星数据之外还使用了闪电观测网络和来自气候数值模型的数据。

  欧洲气象卫星组织的 MPE 产品(多传感器降水估计)是另一个例子。在某一区域,Meteosat7 卫星或 MSG 卫星可以与 DMSP(国防气象卫星计划)卫星(该卫星配备了 SSMIS (特殊传感器微波成像仪/探测仪)仪器)的观测数据进行联合应用。这些卫星数据的融合使得人们可以通过统计学习算法从地球静止卫星数据中估计降水发生的概率。与此同时,在每次卫星过境时可以通过地面观测积累的结果对观测数据不断进行校准。

  丰富的多源数据可以优化单一使用雷达数据的预报精度。以下两个例子可以具体直观地证明这一事实:

  • 由Lannion空间气象中心开发的SAFNWC云顶产品得到了广泛应用,特别是在航空领域。它提供了对流云垂直方向分布的信息,而在水平方向上的二维轮廓则由法国水文气象雷达数据拼接合成。
  • 相对螺旋度[7]是数值预报的一个重要指标,能够很好地表征龙卷风来袭前的云层对流情况。基于法国气象局的雷达观测可以计算这个指标。

  在刚开始的几个月,基于观测资料的外推结果表明站点数据与卫星观测的吻合度较高。AROME-PI 有助于补偿观测外推中已知的缺陷(例如由于数据可用性的限制或数据缺失造成的误差等)。科学家们面临的最大挑战是如何有效集成每种方法的优点以不间断地获得预报截止前0-3小时内的信息。法国气象中心自2016年6月以来开展了结合专家知识、雷达外推以及 AROME-PI 两种模式的预报,旨在获得最优的结果。根据每个专家的历史预测结果与地面观测记录的一致性,来确定分配给每种模式所占的权重。

6. 不久的将来

  “临近预报”脱胎于雷达和卫星,但随着它的发展,对于上述二者的依赖性已越来越低。数值预报的进步使得人们可以在更短的周期内开展预报,还能实时更新数据。这一研究结果表明,除了传统的外推方法和数值预测之外,还可以考虑采取数据融合的方法。

  临近预报的发展大大得益于宏观对地观测系统的进步。MTG (即第三代气象观测卫星—)将搭载 FCI (柔性组合成像仪)辐射计,具有更高的光谱、水平和时间分辨率。随着 LI (Lightning 成像仪)探测器伴随 MTG 卫星被发射升空,以大气对流数据为代表的更多气象观测信息将能够被用于校正、实时处理和验证预报的准确性。而雷达技术则有望在垂直观测、双极化和新参数的改进方面取得预期进展。

 


参考资料及说明

封面照片: 罗兰·加洛斯在雨中(2010)
[1] Mahfouf, J.-F., Moisselin, J.-M., Autonès, F., Vidot, J.,2017, Apport de l’observation satellitaire pour la prévision du temps, La Météorologie, to be published

[2] Autonès, F., Moisselin, J.-M., 2016, Algorithm Theoretical Basis Document for Convection Products, NWC/CDOP2/GEO/MFT/MFT/SCI/ATBD/Convection, seehttp://www.nwcsaf.org/indexscientificdocumentation.html

[3] Brenguier, J.-L., Bouttier, F., Moisselin, J.-M., 2015, Les nouveaux services météorologiques pour l’aviation, La Météorologie – n° 91 – novembre 2015 pp47-53

[4] Gounou, A., Moisselin, J.-M., Autonès, F., Brenguier, J.-L., Levaillant, D., Defer, E., Turner, S., Parol, F., Dezitter, F., Grandin, A.,2015, The RDT nowcasting tool for detecting convective areas associated with high ice water content during HAIC/HIWC field campaign. Paper for the 2015 SAE conference on Icing

[5] 雷达极化测量包括连续发射水平极化波和垂直极化波。通过比较两种反馈回波,可以确定目标的球形度,从而确定水文气象的类型。

[6] 数据融合必须区别于《气象数据同化》一文中描述的数据同化。在同化过程中,必须注意保持变量之间的某些平衡,以便能够平衡地初始化预报模式。在聚变中,我们没有这个约束,因此我们可以定义一个更接近观测值的状态。

[7] 在流体力学中,螺旋度(速度矢量与涡旋矢量的标量积)通常被认为是衡量局部旋转对流体粒子所产生的夹带效应的量度。


环境百科全书由环境和能源百科全书协会出版 (www.a3e.fr),该协会与格勒诺布尔阿尔卑斯大学和格勒诺布尔INP有合同关系,并由法国科学院赞助。

引用这篇文章: MOISSELIN Jean-Marc (2024年3月11日), 临近预报, 环境百科全书,咨询于 2024年12月9日 [在线ISSN 2555-0950]网址: https://www.encyclopedie-environnement.org/zh/air-zh/immediate-forecasting/.

环境百科全书中的文章是根据知识共享BY-NC-SA许可条款提供的,该许可授权复制的条件是:引用来源,不作商业使用,共享相同的初始条件,并且在每次重复使用或分发时复制知识共享BY-NC-SA许可声明。

Nowcasting

Encyclopedie environnement - prevision immediate -

Weather radars and satellites have long provided the main input data for nowcasting. Extrapolation methods based on recent movements of clouds and precipitating areas provided a very short-term forecast. Now, high-resolution numerical prediction is mature enough to be part of the immediate prediction tools and allows to push back several limits: maximum maturity, available parameters. And recently, data fusion methods have led to a new breakthrough by enriching outputs and reducing discontinuities when several methods are used.

1. The recent history of Nowcasting

Nowcasting covers all weather forecasts up to a few hours, a limit that varies according to definitions. It is a structured activity available over large territories, the origin of which will be placed without trying to be too precise in the 1960s with the rise of radar observation systems and geostationary satellites. Indeed, these observation systems offer a high frequency of measurement, making it possible to understand the evolution of the phenomena monitored over a wide geographical coverage.

Over the years, many factors have improved immediate forecasting, such as the sharing of work or initiatives (e.g. COST 78 – European Cooperation in Science and Technology, 1983-1988) and the arrival of new meteorological satellites such as MSG (Meteosat Second Generation). The World Meteorological Organization (WMO) has promoted programmes to demonstrate immediate forecasting, particularly in conjunction with the Beijing (2008) and Vancouver (2010) Olympic Games. The cover image of this article illustrates the contribution of immediate forecasting in a sports audience.

For a long time, immediate forecasting focused on dangerous phenomena, cyclones, storms, heavy rainfall and floods. In addition to this “security of goods and people” component, there is now an economic optimization component. Aeronautics also works largely in “nowcast” mode (in particular to announce very short-term changes in weather conditions at an airport).

2. The contribution of satellite data

In immediate anticipation, geostationary satellites have several advantages: a wide geographical coverage from the tropics to the mid-latitudes of each hemisphere, a high measurement rate (e.g. fast scanning every 5 minutes of MSG), and a horizontal resolution allowing access to the mesoscale (infrared resolution of 4-5 km from MSG over Europe and 1 km in high resolution visible). MSG has added satellites to the range of nowcasting tools at Météo-France, within an international collaborative framework led by the meteorological satellite agency EUMETSAT: SAFNWC (Satellite Application Facility for Nowcasting). Low-orbit satellites offer a better geographical resolution than geostationary satellites, as well as a wider range of instruments. On the other hand, for a given area, they only capture a few sequences per day. Their immediate use is more likely to be at high latitudes where geostationary satellite coverage is limited. Every 2 or 3 years, SAFNWC delivers a new version of software that allows user services to offer nowcasting products in their field of interest. Météo-France is responsible, within the framework of SAFNWC, for developing codes for cloud and convection products from geostationary satellites.

Let us take as an example the RDT product [1], [2] (Rapidly Developing Thunderstorm), from SAFNWC. RDT detects, tracks, characterizes and predicts the movement of convective cells. The RDT is based on the brightness temperature values of different channels or channel differences, as well as their trend between two images. The user has the possibility to use other data sources: other SAFNWC products (cloud and wind products), model data, data from ground lightning networks such as the Meteorological Network. Finally, the RDT gives the contours of the observed and predicted convective zones, as well as the attributes that characterize the convective cells and in particular their severity: cooling rate, electrical activity, presence or not of overshooting tops, vertical extension of the cloud, etc.

Encyclopedie environnement - prevision immediate - cellules convectives - immediate forecasting
Figure 1. 14/2/2017 RTD cells at 10:00 am. Top extract from a mosaic of convective cells analyzed from data from the GOES-E, METEOSAT 10 (MSG 3), METEOSAT 7 and HIMAWARI-8 satellites. Bottom left, zoom on West Africa with contours (colors according to the cloud’s stage of development), past trajectory (in yellow) and displacement (arrows) of the cells. The large cell to the southwest of the image is described with some attributes. On the right is the legend that is applied to this visualization.

Forecasters use RDT to complement or replace radar data in poorly covered (mountains) or uncovered (oceans) areas . It has also been tested in the HAIC (High Altitude Ice Crystals) project to study ice crystals at high altitudes [3] to assess its potential as a tool for reporting areas of high crystal concentration [4]. During the HAIC project, the RDT reached level 5 of the TRL (Technology Readiness Level) scale, which measures the level of maturity of a technology.

The production of RDT by Météo-France now makes it possible to satisfy users with global needs, in particular aviation. Figure 1 illustrates the RDT coverage that continuously uses several satellites with varying spatial and temporal resolutions. Depending on the frequencies and availability of the various satellites, short-term analyses or forecasts (“planned for now” data) are used in the assembly to synchronize the information.

3. The contribution of radar data

Compared to satellite measurement, weather radars provide more usable information. The maximum reflectivities on an air column are used to determine the intensity of convection. The amount of precipitation that reaches the ground can be deduced from this. Meteorological services also use other types of radar measurements: the Doppler effect, which provides access to the radial wind component and then identifies strong shear forces, polarimetric measurements [5], which characterize hydrometeorological types, etc.

Zone methods allow to track large sets of precipitating cells while “object” methods allow to track individual cells. Knowledge of past movements makes it possible to extrapolate the information provided by the radar to a characteristic one-hour deadline.

Encyclopedie environnement - prevision immediate - champ deplacement 2pir - immediate forecasting
Figure 2. An illustration of the 2PIR displacement field, case of 05/06/2002, on the radar reflectivity image is superimposed on the displacement field calculated with the 2PIR method.

The 2PIR method (Immediate Precipitation Forecasting by Radar Imaging) was initiated in the 1990s and then rapidly used for immediate forecasting needs, such as precipitation forecasting at weather-sensitive sporting events. Its use has since diversified and is now Météo-France’s main automatic tool for forecasting rainfall in the coming hour, both in mainland France and overseas territories. Based on the radar reflectivity mosaics, this method first analyses the displacement of radar echoes between the observed image at the time and the one preceding it by 10 minutes, by cross-correlation, resulting in the development of a displacement velocity field defined on a regular grid. This field is then applied to the last observed mosaic to derive mosaics extrapolated up to 1 hour after the last observed mosaic, for reflectivity images as well as for 5-minute QPE (Quantitative Precipitations Estimate) images. Operating by successive advections, it makes it possible to prolong the observed movements of precipitation by reproducing both rotational movements and convergences or divergences (see Figure 2). The main limitations of the 2PIR method, in addition to those resulting from the radar data that feeds it, are linked to the hypothesis of a constant intensity advection and the failure to take into account the orographic effects.

For aeronautics, the ASPOC3D service (Application of signalling and storm forecasting for air traffic control) is based on:

  • The Météorage lightning network
  • Reflectivity measurements of the radar network coupled with satellite observations to identify the convection and altitude of cloud tops throughout France, with a resolution of 1 km and a refresh time of 5 minutes.
Encyclopedie environnement - prevision immediate - risque de convection derive - convection derive - immediate forecasting
Figure 3. Convection risk derived from radar and satellite measurements. The colored contours correspond to different levels of severity.

Convective cells are defined by the spatial contour of four levels of reflectivity. This almost three-dimensional information is then translated into objects that represent the clouds significant for aviation: identification of convective clouds, characterization of their contour, altitude of their top, maximum reflectivity values reached at the heart of the cloud, the presence of lightning that attests that the cloud has reached the storm stage. Each object can then be transmitted to the user to be integrated into his console or impact model. Figure 3 shows these “convection” objects.

Encyclopedie environnement - prevision immediate - capture ecran extranet apic
Figure 4. Screenshot of the APIC extranet of 14/10/2016

After the severe floods in south-eastern France on 15 June 2010, it seemed necessary to provide an alert service in case of exceptional localized rainfall. The objective is to provide, in a fully automatic way, a decision support service for mayors to activate flood management procedures. The APIC service (Intense Rainfall Warning at the municipal level) calculates the cumulative rainfall at different depths from the water slide images of the French radar network. For each area of interest and for a depth ranging from 1 hour to 24 hours, return period statistics are used to assess whether the observed precipitation is exceptional, distinguishing two levels of severity. When exceptional precipitation is newly diagnosed, or in the event of increased precipitation thereafter, a warning is sent to subscribers. A website (see Figure 4) provides an overview of the situation through a map that shows at any time current warnings, temporary unavailability and areas where the service is not open. In order to determine whether or not the service can be provided for a given municipality, Météo-France uses an estimate of the average quality of precipitation detected by radar over the previous two years. The APIC service was opened in December 2011 and now covers more than 92% of the municipalities.

4. Contribution of numerical weather prediction to immediate forecasting

Current immediate forecasting tools mainly use extrapolation of observed data. Advances in numerical modelling and the increase in computing power now make it possible to consider the use of numerical prediction for the very first forecasting deadlines, in order to overcome the limitations of methods based on the extrapolation of observations (structures cannot be created ex-nihilo, nor evolve in intensity, the impact of the relief in terms of displacement or evolution in intensity is not taken into account).

In parallel, fine-scale numerical models are now able to simulate scale phenomena that are of interest for nowcasting with great realism and to assimilate increasingly frequent observations. They make it possible to increasingly meet the needs of immediate forecasting: to have data on the first updated forecast deadlines with the latest observations. This information must also be available within a time frame that allows it to be used in real time. The challenge for numerical prediction models dedicated to nowcasting deadlines is to provide updated information in a short time frame. This constraint imposes a compromise between the number of updates, the refreshing of observed data and the necessary calculation time. If the simulated physical processes remain identical to those of fine mesh models, the numerical models dedicated to nowcasting must therefore be configured to meet these various challenges.

Météo France has a AROME-FR fine mesh model (read : Introduction to weather forecasting) whose implementation has been adapted to create a numerical model dedicated to immediate forecasting: AROME-PI, in place since March 2016.

AROME-PI includes all the characteristics of AROME-FR: same domain, same version of physical and dynamic code, same 3D-var data assimilation method, same resolution (1.3 km), same coupling model (ARPEGE). However, some settings are adapted to meet the needs of immediate forecasting. For example, AROME-PI outputs are available at a time interval of 15 minutes compared to 1 hour for AROME-FR. The aim is to facilitate the use of the model in immediate forecasting as well as the fusion of data with other highly paced forecasts.

The volume and frequency of information delivered by this model are important. Systems are in place to synthesize the available information and draw attention to dangerous phenomena that facilitate exploitation by Météo France forecasters.

5. A new step forward: data fusion

Data fusion methods [6] are expected to become widespread in immediate forecasting (combined approach between satellite, radar and mesoscale numerical prediction).

Let us take again the example of SAFNWC. The following data are used in addition to satellite data: lightning networks, data from weather numerical models.

The MPE product (Multisensor Precipitation Estimate) of Eumetsat is another example. In a given area, covered by a Meteosat7 or MSG satellite, data from American DMSP (Defense Meteorological Satellite Program) satellites equipped with the SSMIS (Special Sensor Microwave Imager/Sounder) instrument are used. Their measurements make it possible to implement statistical learning to estimate precipitation rates from geostationary satellite data. The statistical diagram is then used and regularly calibrated at each pass of the scrolling satellite.

Additional information can enrich the description of the diagnostics or immediate prediction from the radar data, here are two examples:

  • The SAFNWC cloud top product developed by the Lannion Space Weather Centre is widely used, particularly for aviation. It provides information on the vertical extension of convective clouds, the 2D contours being determined from the mosaic of French hydrometeorological radars.
  • Relative helicity [7] is a diagnosis from numerical prediction and is a good precursor of tornadic potential of convective cells. It is calculated for each cell observed by radar in the Météo-France production.

The extrapolation of observations brings us very close to the situation observed in the first few months. AROME-PI helps to compensate for known defects in the extrapolation of observations (limits in areas of relief, absence of cell creation or disappearance, etc.). The challenge is to take advantage of the advantages of each method to have more relevant and uninterrupted information on the 0h-3h forecast deadlines. The merger developed in Météo-France since June 2016 is based on a method combining two forecasts (called “experts”), radar extrapolation and AROME-PI, in order to obtain a combination close to or better than the best of the two experts. The resulting product is a weighted sum of the numerical forecast fields and extrapolations. The weights assigned to each expert are adjusted in real time, according to the past behaviour of each expert in comparison to the ground truth.

6. The near future

Nowcasting is gradually emancipating itself from the exclusive use of the major observation systems that have led to its emergence: radars and satellites. Progress in numerical prediction makes it possible to have forecasts that can be used at shorter deadlines and with a description of phenomena closer to the latest available observations. This observation leads us to build fusion methods between traditional extrapolation approaches and numerical prediction.

Nowcasting will benefit in the future from advances in the large observing systems it uses. The MTG (Meteosat Third Generation) satellite, which will provide a better spectral, horizontal and temporal resolution with the FCI (Flexible Combined Imager) radiometer. With the LI (Lightning Imager) flash detector embedded in MTG we will have new information which, for convection products in particular, will be used in the adjustment, real-time processing and verification phases. Concerning radar, progress is expected in new parameters that exploit all vertical radar measurement, dual polarization, and other improvements.

 


References and notes

Cover image. Roland Garros in the rain (2010)

[1] Mahfouf, J.-F., Moisselin, J.-M., Autonès, F., Vidot, J., 2017, Apport de l’observation satellitaire pour la prévision du temps, La Météorologie, to be published

[2] Autonès, F., Moisselin, J.-M., 2016, Algorithm Theoretical Basis Document for Convection Products, NWC/CDOP2/GEO/MFT/MFT/SCI/ATBD/Convection, see http://www.nwcsaf.org/indexScientificDocumentation.html

[3] Brenguier, J.-L., Bouttier, F., Moisselin, J.-M., 2015, Les nouveaux services météorologiques pour l’aviation, La Météorologie – n° 91 – novembre 2015 pp47-53

[4] Gounou, A., Moisselin, J.-M., Autonès, F., Brenguier, J.-L., Levaillant, D., Defer, E., Turner, S., Parol, F., Dezitter, F., Grandin, A., 2015, The RDT nowcasting tool for detecting convective areas associated with high ice water content during HAIC/HIWC field campaign. Paper for the 2015 SAE conference on Icing

[5] Radar polarimetry consists in emitting successively horizontally and vertically polarized waves. The comparison of the two feedback echoes makes it possible to determine the degree of sphericity of the target, and therefore the type of hydrometeor.

[6] Data fusion must be distinguished from data assimilation, described in the article Assimilation of Meteorological Data. In assimilation, care must be taken to preserve certain balances between the variables in order to allow a balanced initialization of a forecast model. In fusion, we do not have this constraint, and we can therefore define a state closer to the observations.

[7] In fluid mechanics, helicity (scalar product of the velocity vector by the vortex vector) is often considered as a measure of the entrainment effect that a local rotation will give on a fluid particle.


环境百科全书由环境和能源百科全书协会出版 (www.a3e.fr),该协会与格勒诺布尔阿尔卑斯大学和格勒诺布尔INP有合同关系,并由法国科学院赞助。

引用这篇文章: MOISSELIN Jean-Marc (2024年3月7日), Nowcasting, 环境百科全书,咨询于 2024年12月9日 [在线ISSN 2555-0950]网址: https://www.encyclopedie-environnement.org/en/air-en/immediate-forecasting/.

环境百科全书中的文章是根据知识共享BY-NC-SA许可条款提供的,该许可授权复制的条件是:引用来源,不作商业使用,共享相同的初始条件,并且在每次重复使用或分发时复制知识共享BY-NC-SA许可声明。