Improving TV weather broadcasts with technological advancements: two cases from a 20 year perspective

Authors


Abstract

The introduction of technological advancements in TV weather broadcasts has been a constant trend, particularly in the last 20 years. Rapidly evolving techniques, both from the point of view of new meteorological products or the use of advanced computer generated graphics, have offered a wide choice of possibilities to enhance information content and communication capability. However, despite the potential advantage of technological innovation, a case-by-case careful feasibility analysis is always required, including not only technical and economical aspects, but most importantly communication issues bearing in mind the education and acceptance of the audience.

In this paper two specific cases of technological advances in TV weather broadcasts, seen from the perspective of the last 20 years, are examined: the introduction of weather radar images and the use of animated characters as weathercasters. These cases are illustrated from the experience acquired in TVC, the public Television of Catalonia (NE Spain). Direct access to weather radar data was first made possible within the framework of a joint collaboration agreement with the University of Barcelona and the Meteorological Service of Catalonia, which installed a new C-band Doppler radar. The use of animated characters as weathercasters was the result of developing and advanced human-like interface accessing an already existing database fed automatically by NWP forecasts. In each case a description of both technical aspects and also of the benefits and challenges seen from the communication point of view found in their introduction in TV weather broadcasts is provided. Copyright © 2010 Royal Meteorological Society

1. Introduction

The progressive introduction of technological changes in the form of new visualization systems or meteorological products in weather television broadcasts has followed a constant trend since the early times of TV weathercasting (McAllen, 1979; Henson, 1990). However, in the last 20 years, especially, the rhythm of change has accelerated as a result of the rapid progress in meteorology, particularly in numerical weather prediction (NWP) and remote sensing, and computer visualization technologies. Soon after new meteorological products, such as satellite or weather radar imagery, or new versions of advanced computer generated graphics have been available, they have been considered for potential use or implemented by the broadcast industry. This progress is evident in highly competitive markets of meteorological products and services such as in the United States of America and, to a lesser extent, in Europe, given the big differences in those two geographical areas, as recently examined by Pettifer (2008).

The substantial weight of broadcast meteorology in the US market is well illustrated by the existence of a specific section devoted to this area within the American Meteorological Society (AMS), which, among other activities, operates a professional accreditation system for weathercasters and organizes periodically specialized symposia, the AMS Broadcast Meteorology Conferences. In this framework, it is not strange that television weather broadcasters have been thoroughly examined in the past, mostly from the US perspective, in a number of publications, for example remarking on their potential role as science communicators (Earl and Pasternack, 1991; Posegate, 2008; Wilson, 2008; Doherty and Barnhurst, 2009), in objective forecast verification analysis (Driscoll, 1988), or commenting various aspects such as public credibility (Grenci, 2005; Socci, 2007). References focused in other areas such as Europe are relatively scarce, despite the high number of national and regional television weather programmes routinely broadcasted there.

The objective of this paper is to describe two specific cases of technological innovation in TV broadcasts, seen from the perspective of the last 20 years: the introduction of weather radar images and, the use of animated characters as automated weathercasters. These cases are illustrated from the experience acquired in TVC, the public Television of Catalonia (NE Spain) in close collaboration with the Meteorological Service of Catalonia (SMC). In each case a description of both technical aspects and also of the benefits and challenges seen from the communication point of view found in their introduction in TV weather broadcasts is provided.

The organization of the paper is as follows. Firstly, a brief introduction to the area of interest and institutions involved is given. A description of the weather radar system and visualization used in the broadcasts is then presented. Later, an overview of the production chain of animated human-like weathercasters and related applications is given. Finally, the paper concludes with a summary and outlook.

2. Background

The public television of Catalonia, TVC, is part of the Catalan Corporation of Radio and Television (CCRTV), owned by the regional administration, and broadcasts over this region (the northeast of the Iberian Peninsula) and other neighbouring areas which share the same language such as the Balearic Islands in the Mediterranean Sea. Catalonia has a geographical extension about 32 000 km2 and approximately 7 million inhabitants, mostly living in the metropolitan area of Barcelona. This extension and population density make this region comparable to others served by local TV stations in the United States of America as described by Trobec (2007), which provide more specific local detail in TV weathercasts compared to general bulletins covering the whole country. The TVC meteorological section plays an active role in SAM (Audiovisual Meteorological Services), a provider of meteorological information for Internet, mobile devices and TV in Spain within the Activa Multimedia group, a subsidiary of the CCRTV.

On the other hand, the Meteorological Service of Catalonia is also part of the regional administration of Catalonia and is responsible for providing meteorological products and services to the regional civil protection and water authorities and other local organizations, and disseminating warnings for hazardous weather such as heavy rainfall events. Other activities such as support to civil aviation or defence are performed at state level by the Meteorology State Agency (AEMET, formerly INM).

A formal collaboration agreement regulates fluent information exchange between TVC and the SMC, which also prepares weather information for the mass media, such as several newspapers and radio stations. Historically, among the early activities of the SMC, which began in 1921, was the dissemination of weather bulletins—the first weather radiobroadcasts in 1927 by the SMC founder, Eduard Fontserè, are considered among the first ones in southern Europe (Batlló Ortiz, 2002; Roca Rosell et al., 2004).

3. Weather radar images in TV

The intense activity and competition in the US TV broadcast market led to many stations acquiring their own weather radar systems, particularly in the 1990s after National Weather Service radar data were available (Robertson and Droegemeier, 1990), so that stations could differentiate their offer from the rest. This widespread availability of radar data caused some potential problems of communication, for example regarding the use of the term ‘live radar’ which motivated a special statement of the AMS (2000). In other parts of the world the situation was very different and radar imagery was much more scarce and unknown to the viewers.

For example, since their start during the mid-1980s, in TVC weathercasts the main sources of information displayed were sea level pressure maps and Meteosat satellite imagery, mostly in the thermal infrared channel, allowing presenters to show viewers cloud cover and convective development areas. A big change took place in 1997 when first weather radar images were used and broadcast, allowing presenters to show radar reflectivity fields associated with precipitation. To the authors' knowledge this was the first time radar images in TV were broadcast in Spain. This was possible after a collaboration agreement was signed between several institutions including the SMC, TVC and the University of Barcelona, which acquired and installed a weather radar for research purposes in Vallirana (near Barcelona), currently integrated in the SMC four-radar network.

3.1. The Vallirana weather radar

The radar system, model TDR-3070 manufactured by Kavouras, Inc., was a radar unit widely sold during the late 1990s to many US TV local stations. It is a C-band Doppler radar, with a 3 m antenna and a low peak power transmitter (about 8 kW) based in Travelling Wave Tube (TWT) technology, which allows pulse compression to enhance resolution and sensitivity (O'Hora and Bech, 2007). Radar data are acquired automatically every 6 min covering 15 antenna elevation angles, Plan Position Indicator (PPI) products, with a volumetric short range mode of 130 km and a base scan long range mode of 250 km. The antenna was originally installed on a 25 m tower, which was modified and shortened in 2003 to improve maintenance operations. A summary of technical characteristics of the radar is given in Table I. The TDR-3070 radar was originally equipped with a graphical terminal display designed for broadcasting purposes with real-time visualization, zoom and animation capabilities. Background geographical maps at different scales were customized to show the coastline, main cities and roads of the area covered. The highest resolution radar images, covering a 25 km circle, were designed to provide ‘street level’ visualization details, a standard term used by several commercial radar vendors. As noted by Speheger and Smith (2006) and Newman et al. (2008) the use of these maps is in potential conflict with the real resolution of radar measurements or other potential errors related to radar observations such as wind drift of precipitation below observation level at high resolution (Collier, 1999). This and other factors hamper or limit the use of weather radar quantitative precipitation estimates, as in the recent study described by Collier (2009).

Table I. Technical characteristics of the TDR-3070 Kavouras weather radar
Antenna feed typeOffset
Antenna beam width1.5°
Antenna gain42 dB
Transmitter typeTWT
Transmitter peak power8 kW
Operating frequency5600–5650 MHz
Pulse widths1, 5, 30, 40 µs
PRF600, 1000, 1200 Hz

The radar was installed in 1996 in the hill of Puig Bernat, in the town of Vallirana, near Barcelona city (Figure 1). One year later, upon the installation of a 2 Mb microwave link to allow remote operation and data transmission, the SMC started to use the radar routinely for weather surveillance tasks and the images were made available to TVC. The first implementation of the data transfer was through a video link which converted the VGA output of the radar display into a PAL signal which was then transmitted to TVC. Radar reflectivity PPI images were used by weathercasters in live broadcasts, either in real-time mode (‘live radar’), showing a characteristic rotating line indicating the antenna pointing direction, or in animation mode, displaying a sequence of past images, typically spanning a few hours. In both cases this visualization type was controlled by the SMC terminal and simply received as a video image at TVC: the transmission in ‘live radar’ mode required changing temporally the standard radar volumetric data acquisition to a continuous low elevation PPI, an operation typically coordinated between TVC and SMC through a telephone call. In 2000 a new radar data acquisition and visualization software allowed transferring radar images as graphic files from the SMC to TVC, giving better local control of animated sequences by TVC weathercasters. However, with this new software, the continuous low elevation scan, ‘live radar’ mode, was no longer available as it interrupted the volumetric radar data acquisition required by enhanced radar precipitation estimates for quantitative applications (Sánchez-Diezma et al., 2002). Since then, radar data transmission was mainly performed through graphic file transfers from SMC to TVC, though the video link was also kept as a backup system to broadcast radar animations generated locally at SMC.

Figure 1.

The Vallirana Doppler weather radar in 1998 installed in Puig Bernat. Source: SMC. This figure is available in color online at www.interscience.wiley.com/ma

3.2. Use of radar images

An example of the radar real-time visualization type in a broadcast presented by the second author of this paper is shown in Figure 2(a). It corresponds to the heavy rainfall event of 3 December 1998, a remarkable episode with 24 h accumulations exceeding 100 mm in Barcelona city (Casas et al., 2010). Figure 2(b) shows an example of the second visualization type, a TV broadcast displaying an animated radar sequence on the 7 September 2005. On that day there was strong convective activity in Catalonia with intense thunderstorms and a tornado outbreak that affected the southern part of the metropolitan area and Barcelona airport (Bech et al., 2007).

Figure 2.

Two examples of use of weather radar images in TVC live broadcasts: real-time display of the heavy rainfall event of 3 December 1998 (a) and animated sequence of the tornado outbreak event of 5 September 2005 (b). Source: TVC. This figure is available in color online at www.interscience.wiley.com/ma

The use of radar images in TVC broadcasts became quickly popular and was very much appreciated by the presenters and the audience. The possibility of seeing the extension and intensity of estimated precipitation fields was a clear added value to the TV broadcast, previously supported essentially by satellite imagery and sea level pressure forecast maps, improving particularly the description of the current situation and very short range forecasts or nowcasts of precipitation. This has been particularly beneficial to communicating heavy rainfall warnings. In the beginning some difficulties in the interpretation of artefacts in the radar images were evident, for example the presence of sea clutter or spurious echoes caused by anomalous propagation of the radar beam over the sea, which is a frequent situation in this area during some parts of the year (Bech et al., 2000). Despite these initial problems, after some time the regular use of radar images in TVC broadcasts became a standard feature and an essential element in the communication of past and current description of precipitation occurrence and distribution.

4. Animated characters as weathercasters

The use of computer generated or animated characters has increased in recent years, particularly in interactive self-learning applications where they play the role of ‘conversational’ pedagogical agents whose computing and didactical aspects are investigated in a number of studies (Jonson and Rickel, 2000; Andre and Rist, 2001; Gratch et al., 2007; Kunc and Kleindienst, 2007; Louwerse et al., 2008, Wouters et al., 2008).

The development and use of animated human-like weathercasters in the Catalan Corporation of Radio and Television (CCRTV) was the result of several projects carried out in its subsidiary group Activa Multimedia (AM). Two companies of AM participated: Audiovisual Meteorological Services (SAM) and iVAC, which produces interactive virtual automated characters (iVACs) for television, the Internet and mobile devices. The iVAC weathercaster was popularly known as Sam. Other iVACs have been produced by the CCRTV and are present in a number of Spanish TV programmes such as in sport broadcasts or in children programmes. The design and programming of the iVAC system was made in collaboration with the Pompeu Fabra University in Barcelona.

4.1. Step-by-step generation of animated weathercasters

In this case the main objective was the automatic generation of digital video to broadcast meteorological information. This is done in several steps: (1) processing of numerical weather prediction (NWP) output, (2) retrieval of weather forecasts at selected locations in standard terms, (3) generation of a standard text forecast, (4) generation of IVAC (multilingual) text to speech audio, and, (5) generation of iVAC video.

  • 1.The starting point is a database updated twice daily with NWP forecasts from the US GFS model and the mesoscale MASS model (Codina et al., 1997). The GFS data provides 10 day forecasts used as initial and boundary conditions for running the MASS model globally, providing mesoscale forecasts for the first 8 days.
  • 2.A post-processing system is then applied to obtain a present weather description at 15 000 selected locations over the world with 8 h temporal resolution (morning, afternoon, night), high and low temperatures and wind, among other variables. The present weather descriptor is similar to the WMO present weather code of SYNOP observations and may take 80 different values or standard terms (for example clear, mostly clear, scattered showers, isolated showers, or showers). These 15 000 local forecasts descriptions were used, before the iVACS development, to serve a number of automated forecasts (Molina, 2005).
  • 3.Each standard term has a set of three different sentences with some degrees of freedom to allow random generation of a descriptive sentence associated with the forecast. The configuration of the sentences is updated every 8 h, which is the temporal resolution of the forecasts. For example tomorrow morning forecast generated at night ‘Showers are expected tomorrow morning’ would change in the morning to ‘Showers are expected this morning’.
  • 4.The generation of the audio speech is done in 10 different languages: Spanish, English, French, German, Italian, Portuguese, Catalan, Basque, Galician and Russian. This generation requires adding a greeting part (chosen randomly from a list of possible options such as Hi, Hello, How are you?, or How are you doing?) and an introductory item about the location and the time of the day considered. The previously constructed weather forecast is then added, including as well information about temperatures and their evolution. Finally a farewell expression is taken from a possible list of choices, similarly as in the greeting section. A special module developed by the Italian speech technology vendor Loquendo Vocal Technology and Services produces realistic digital audio from the generated text in each language.
  • 5.The text generated previously contains punctuation marks such as commas, colons, semicolons and question marks that play an important role in the iVAC video generation. Each of these marks control some predetermined movements of the face and body of the animated character. For example, a comma makes the iVAC close its eyes, and a question mark produces a slight movement of the head and eyebrows. These movements are also coordinated with changes in the voice tone, also modulated by the text punctuation, in order to obtain a more realistic effect. Geographical locations and description of temperatures may produce the iVAC to point to map or background elements if they appear on the same scene (different scenes are available depending if the iVAC target is a TV programme, mobile phone or internet web page). All possible movements are previously recorded to allow a fast generation of the video animation to simulate a real-time response comparable to a human person.

4.2. iVAC examples and results

The configuration of each iVAC was done through a specially designed software module, the iVAC editor. A simple interface allows choosing elements of each character such as clothing, accessories, and different views (Figure 3). The video can be produced in several standard digital formats including. avi. flw, or. wmv.

Figure 3.

(a) Animated character design through the iVAC editor and (b) different views of the character and backgrounds. Source: Activa Multimedia. This figure is available in color online at www.interscience.wiley.com/ma

An example of iVAC Sam is shown in Figure 4(a). This is from the TV programme ‘Your Weather’ from the Méteo channel where viewers telephoned to ask for specific forecasts and Sam replied in real time, typically lasting less than a minute. In other programmes Sam appeared giving longer broadcasts, for instance weekend forecasts at ski resorts in Spain and Andorra. Figure 4(b) shows Sam giving information for a ski resort in Catalonia through a mobile phone service. Sam can also be accessed online at http://www.meteosam.com/eng/index.php where it provides weather forecasts for a number of locations across Europe.

Figure 4.

Two examples of Sam, the animated weathercaster. (a) Sam on the ‘Tu Tiempo’ (Your Weather) program on canal Méteo, showing a typical TV complex background. (b) Sam showing weather information at a ski resort in Catalonia adapted for mobile visualization. Source: Activa Multimedia. This figure is available in color online at www.interscience.wiley.com/ma

Once the system is developed, and the NWP forecasts are fed regularly into the data base, the advantages of using animated characters in weathercasting are clear in terms of operation costs, either for automated predefined forecasts or as an interactive a la carte weatherman for mobile devices. However, from this particular experience, after a positive initial reaction in the audience, probably driven by the expectation created by the innovation, there was later some apparent indifference which made it necessary to reconsider some of these applications. While the applications represent a valid option for short automated interactive information exchange (as in interactive TV and mobile platforms) longer and more complex information, such as general weathercasts that may last several minutes, might be more appropriate for traditional human weather presenters. Though the sophisticated processing described above allowed for high quality results, both in the realistic human appearance and in the meteorological information content, the overall communication capability of the animated weathercasters seems still far from the possibilities of their human counterparts.

5. Summary and final remarks

In this paper two technological advances in TV weather broadcasts have been examined and discussed from the perspective of the last 20 years and the experience of TVC, the public Television of Catalonia (northeast Spain): the introduction of weather radar images and the use of animated characters as weathercasters. Direct access to radar data imagery, although normal in many US TV stations, is unusual in Europe. In this case it was possible thanks to a collaboration agreement among TVC, the University of the Barcelona and the Meteorological Service of Catalonia. Despite some initial difficulties, the results of introducing radar observations in TVC weathercasts have been very positive so far and they are now considered a fundamental element to describe the occurrence and spatial distribution of precipitation, and helping in the dissemination of weather warnings when necessary. The use of animated characters as weathercasters was the result of developing a sophisticated human-like interface accessing an already existing database fed automatically by NWP forecasts. The resulting animated digital video sequences have been used successfully both for short automated interactive information exchange (as in interactive TV and mobile platforms) and for longer and non-interactive more complex forecasts, which were finally deemed more appropriate for their human counterparts, taking into account the different communication capabilities of each option. Therefore, despite the general historical trend of incorporating technological innovation in TV weathercasts, it seems appropriate to evaluate, case by case, not only technical and economical aspects, but most importantly communication issues bearing in mind the education and end-user satisfaction.

Acknowledgements

Thanks are due to the Meteorological Service of Catalonia, the Catalan Corporation of Radio and Television (CCRTV) and to Activa Multimedia in particular for all the technical information, graphics, and valuable comments provided for the preparation of this paper.

Ancillary