REQUIREMENTS OF DATA FROM AUTOMATIC WEATHER STATIONS (Status Report)

Transcription

1 REQUIREMENTS OF DATA FROM AUTOMATIC WEATHER STATIONS (Status Report) E. Rudel Central Institute for Meteorology and Geodynamics, Hohe Warte 38, A Vienna, Austria INTRODUCTION Automatic meteorological observation is not an idea of our century, although the first practically used systems were installed in the 1960ies. Already in the 1870ies the Dutch instrument designer Olland developed on the suggestion of Buys-Ballot a so called telemeteograph with which meteorological measurements could be sent via telegrapher to meteorological centers. At the same time a similar principle was developed in Belgium. The whole system was not very successful because of the high costs for installation and service. (1) But the demand for automation was increasing with the interlink of economy with weather. Long distances to the registration instruments, installed appropriately from a meteorological point of view, can considerably deteriorate the quality of the data. Furthermore, the hourly evaluation of the generally analogous registrations of different meteorological parameters, is increasingly necessary for the diverse purposes of technical meteorology and climatology of health-resorts. This, however, entails great expense of personal and costs. In the 1960ies and 70ies the considerable progress of micro processing techniques caused the development of low-cost automatic systems for the recording of meteorological and climatological data. Without going into details we have to distinguish four generations and the real step forward was that most of the tasks of these systems are executed by the software and not only by the hardware. The trend to automation of surface data acquisition systems has started already in the seventies of this century, has risen in the eighties and nineties and can be expected to continue in the next decades worldwide. In about thirty years automated surface weather systems stations will replace conventional weather observing systems at most sites. The future surface observational networks will consist largely of unmanned stations with a temporal resolution of as little as 10 Minutes and a spatial resolution of less than 50 km. There are big advantages to the new systems: Automation of surface weather observations reduces costs, increases area coverage and provides data continuously at frequent intervals and for any observation time. On top of it these systems eliminate the subjectivity inherent in manual observations

2 such as visibility, cloudiness and estimates of the winds. This reflects the requirements of all users of near real time synoptic data. It also reflects the trends to reducing model grid scale and the need for more observations to be available in shorter timescales. But also the users of climatological, hydrological, and agrometeorological applications have a growing advantage of the ongoing installation of AWSs. The digital evaluation of different meteorological elements needed for the different applications in climatology was never as easy as with the outputs of autostations. The initial motivation for the use of autostations in climatology or agrometeorology was to get more evaluated meteorological information and especially data in hourly frequency. The motivation for a move to automated observations in climatology is primarily financial, because evaluation of analogous registration is expensive. In the past climatologists have estimated daily mean temperature from a range of methods (e.g. averaging maximum and minimum temperatures or averaging all three-hourly temperatures for a day) and nowadays an autostation can provide exact values in a very easy and cheap way. (3) COMPOSITE OBSERVING SYSTEMS There was a lot of progress in recent times on measuring and observing meteorological variables by the application of direct and remote surface and space based techniques. Various types of instruments and equipment applied for these observations are already operated in automatic networks either on national or regional levels (e.g. weather radar networks). There is done significant work on examining benefits of composite observing methodologies which is being carried out in NAOS (North American Observing System) and within EUMETNET (EUCOS) (Europe). This approach enables the possibility that data obtained from different sources can be combined with the objective of achieving more complete and objective information on the status of the atmosphere. Due to this possible combination of observations performed by various types of instruments which are mainly amalgamated at central points within the NMHSs concerned, the continued need for some visual or subjective observations traditionally done by human observers needs to be reconsidered. These observations were urgently needed in the past for obtaining, as far as possible, a complete picture of the atmosphere. They might significantly be reduced for those Services which are already applying these advanced techniques. It was considered that it might be applied world-wide within the next 10 to 20 years. The considerations may relate to such parameters as cloud cover, cloud type, but probably also to the intensity of precipitation and thunderstorms, etc. These data could in several cases, now or in the near future, be obtained or derived either directly from sophisticated instruments, combinations of them, or by remote measurements, such as surface based remote observations (weather radars, windprofilers, etc.) or space based platforms (satellites). However, the direct or remotely generated data and the combination of these measurements by sophisticated algorithms are

3 no longer anymore available as "point observations" as obtained from observing stations. The advantage of these "new" data is that they might be more representative for users than traditional point measurements since they provide a more integral picture of the area of concern and may make obsolete some of the traditionally needed visual observations. This approach could replace, and perhaps extend, in the near future or in a longer-term, several traditional point measurements and, especially, visual observations. However, it is clear that especially for the application of indirect or remote observing methods, such as weather radars when applied for precipitation amount and intensity measurements, and for satellite observations, surface observations are still crucial for now and the foreseeable future, for in situ calibration purposes and "ground truth". The need for a higher quality of this kind of observations and real-time availability will increase. These expected needs call for a clear statement of requirements of observations to be done automatically by AWSs (5,8). THE RECOMMENDATIONS OF CIMO Unless otherwise specified by the user, the Automatic Weather Station should provide for a basic suite of sensors which includes atmospheric pressure, air temperature, relative humidity or wet bulb depression, wind speed, wind direction and rainfall. This together with a few spare data channels plus engineering housekeeping channels, such as battery voltage and system vandalism alarms, indicates that a system of 16 channels may be required. However the user may specify additional quantities such as soil temperatures at several depths and non-meteorological quantities, and this may require the number of channels to be increased. This decision requires considerable thought and care. Meteorological sensors, if they are to provide data of a quality comparable with data obtained and archived by the National Meteorological Agency, should meet the following criteria:- total system (including sensor, measurement system and data processing) accuracy and resolution should be in accordance with WMO requirements. Because modern data processing systems have good performance characteristics, the main source of error is likely to be in the sensors and the associated interface circuitry; sensors offered should be well established with proven performance characteristics comparable with meteorological sensors widely in use by meteorological agencies throughout the world; the sensors should either be known and acceptable to the local national meteorological service or well-documented proof, from an acceptable source (such as a WMO or national intercomparison), of their characteristics and performance should be provided; long term stability and reliability should be assured.

4 In relation to air temperature and humidity (or wet bulb depression) measurement, the sensors need to be protected from the effects of solar and other radiation by exposure in a properly designed instrument shelter. Such a shelter: should be of wood (suitably treated with preservatives) or UV-stabilized plastic construction. Metal construction will not be acceptable unless it is aspirated and/or very strong evidence of a satisfactory performance is presented; should preferably be of the white, opaque, double louvre type; should be of a proven design. Unless the type is known to the national meteorological service, documented evidence of performance characteristics should be provided; may be aspirated or un-aspirated. It should be noted that a screen which requires aspiration to achieve the required performance levels may not be suitable if the AWS is to be used at a location where external power supplies are not available or are unreliable. In measuring wind speed and wind direction the standard exposure is at a height of 10 m, but a 2 m wind-run anemometer may be required in addition to or instead of the 10 m anemometer. Other exposure requirements may be specified by the user in order to address problems with the exposure site however these may render the measurements unacceptable for use by the national meteorological service. For ease of maintenance an anemometer mast should preferably be of the fold-down type. It should be capable of withstanding windspeeds of 50 ms -1, unless the user intends it for use in locations liable to experience strong winds such as those subject to tropical cyclones, in which case it should be capable of withstanding windspeeds of 75 ms -1.(6) PRESENT WEATHER SENSORS In two WMO experts meeting, one was held 1995 in Trappes (France) and one 1999 in DeBilt (Netherlands), the participants agreed that an AWS cannot report "present weather" or, more general, visual observations, in a manner as is done by a human observer nor should an AWS be expected to do so since an AWS observes and reports weather differently. Autostations provide consistent information while human observers characteristically show significant subjectivity, uncertainty, and variation especially when the parameters to be observed are not well defined. (7,8) But it was also stated that in many cases no clear and agreed definitions of "present weather", visual, or subjective observations exist so far. Even more significant, there is presently no clear statement available on the actual and future requirements of data users. In considering this unfortunate situation and noting that many of the "present weather" variables were introduced several decades ago to overcome deficiencies in the direct observation or measurement of variables in the atmosphere, the requirements defined at that early time have now to be significantly reviewed in the light of present and future needs.

5 Individual sensors, multi-sensor systems, combination of available information or measurements, and sophisticated algorithms are already available or can be developed if there is a need for observing relevant parameters. It has, however, to be considered that the automation of subjective and visual observation is an expensive undertaking and the requirements need to be evaluated thoroughly before considering their implementation. The automation of visual and subjective observations has to be reconsidered within the light that automated systems perform differently than human observers (i.e. it has to be based on a more objective and well defined basis). If this can be done, widely homogeneous observations can be achieved globally both within and outside of NMHSs. According to the common understanding of meteorologists, visual or subjective observations were more urgently needed in the past than nowadays (or even in future) due to the previously insufficient or generally missing measurements of various variables in the atmosphere. That is to say, the subjective observations were in several cases used as indirect means for characterizing the status of the atmosphere, especially for forecasting purposes (such as the type, coverage, and height of clouds). In addition, quantitative measurements were not sufficiently available or generally not yet possible at this earlier stage so that qualitative information had to be provided instead. These mainly subjective observations were, especially if they were not well defined, very unreliable and subjective (such as the characterization of precipitation as "drizzle", "slight", "moderate", and "heavy"). Improvements of presently available systems are ongoing and there are some individual sensors, multi-sensor systems, and sophisticated algorithms already available, in testing, or in development which may widely meet future needs. However, before further efforts will be undertaken in this regard, the future requirements have to be defined clearly. INHOMOGENEITIES Going back to the classical elements like air-temperature, precipitation or wind conditions.: The new generation of weather stations cause many changes in sensor design, in observation techniques, in the interrogation time and data processing algorithms and this will inevitably introduce inhomogeneities into the climatic record of sites with a long history of conventional observations. Data measured on the same place and with the same environmental conditions but by different systems usually show slight discrepancies. Due to one of the most important questions of today Changing Climate? there should be brought special emphasis on the homogeneity of the meteorological data series. Different guidelines and standards for archiving data of AWS are used on national levels and in some of the results of time series one can see quite clearly the influence of the changes of measuring and averaging techniques. (4)

6 Climatic records, at least those which are readily available, are normally mixtures of both apparent and real variations. Factors causing long-term climatic changes which may be influenced by changing the system from a conventional to an automated are the following: Changes in observing times Changes in averaging methods Stations relocation Change in design (e.g. screen, aspiration) Changes in location height Changes in calibration method The deviations can be divided into systematic and stochastic ones. If the systematic differences are not corrected, inhomogeneities in the climatological series are the consequence. Intercomparisons of different systems over a long period of time(e.g. one or two years) on some representative locations covering different weather conditions are necessary to test the compatibility and to find out correction algorithms. Several statistical methods are available, which can show whether any bias is included in the data records. Some methods provide also an indication of its location, but the causes cannot be revealed by any statistical methods. Therefore nearly all the NMHSs which introduced Automated Surface Weather Observing Systems undertake increased efforts to test and evaluate new algorithms for auto-stations which ensure a kind of Climate Data Continuity. STANDARDIZATION: Regarding the AWS sensor specifications (accuracy, data sampling frequencies, etc.), observation/measurement methodologies, data preprocessing methods/algorithms, and station siting and sensor exposure, there is recommended to follow the most recent guidelines established by WMO. A standardized set of procedures regarding current and past station/network operation (i.e. metadata) is needed. The total system of the AWS, its accuracy and resolution should be in accordance with present and future WMO requirements (and for all related programs such as GCOS, WCP, WCIRD and IPCC). Anticipating a gradual conversion of manually-operated climatological stations to AWS, there is an increasing need to maintain series homogeneity for purposes of climate change detection. As much as possible, side by side comparison of converted stations should be made for at least one year, and possibly even longer for climate parameters with high interannual variability. Stations with historical and global significance should be maintained in their current state as long as possible.

7 Climatologists should have more than a passive interest in the future development of AWSs. As well as making clear what historical observations are required to maintain continuity and homogeneity of the climate record, climatologists should look for opportunities for obtaining more relevant data to their needs. Finally, although it is acknowledged that the needs of climate are generally more stringent than the demands of other stakeholder groups, they are also generally less complex given that the priority surface climate data needs have changed little over the century. Following the ten principles for longterm climate monitoring will undoubtedly have benefits for the broader organization in the long-run. (2, 3) MANAGEMENT OF DATA FROM AUTOMATIC WEATHER STATIONS (AWS) It must be considered that there is a need for compatibility of meteorological data obtained from stations of a staffed network and from Automatic Weather Stations. and there is a need for homogeneity of observed data used for climatological purposes, in the Global Observing System and the Global Climate Change Detection. The autostations must be developed in close connection with all users of the meteorlogical community (forecast meteorologists, climatologists, agrometeorologists, environmental meteorologists). To reach these aims it is clear that responsibility for providing and using data has to be shared between the data collector and the data user. Recommendations for the data collector: We must distinguish if an AWS replaces a standard ordinary staffed meteorological station or if it is installed on a new site. But in any case there is a strong demand for The total system of the AWS, its accuracy and resolution should be in accordance with WMO requirements There must be established standard procedures for collecting overlapping measurements for all significant changes made in instrumentation and observing practices. There must exist detailed documentations of the construction of the data set from the measurements. The list of variables for AWS measurements has to be continually reviewed and updated based on the developments in new technology and new software. The configuration of the AWS must be in this way that the accuracy of the output data set must be nearly identical to the accuracy to the output of an ordinary staffed meteorological station. If the sensor performance of the AWS allows only a less accuracy the ordinary measurement should still be carried out as long as possible. Influences on the climate record of changes in ordinary staffed stations to AWS must be known prior to implementing such changes.

8 Comparison observations and measurements should be carried out and there should be no compromising of the homogeneity of time series. AWS should be fully tested in the field for a carefully selected number of stations. This should span at least one year with comparison data collected for two stations, and possibly even longer for climate parameters with high interannual variability. Document all the procedures and comparisons in a technical or research report. Recommendations for the data user: The users of meteorological data have to look after the metadata, because only the knowledge about the description of the instruments, the calibration methods, the observation reporting practices, the sites and the changes of sites and the data archiving policy can give full information about their continuity and homogeneity. If a site was changed from an ordinary staffed meteorological station to an AWS the user should look after: The documentation of the construction of the data sets. The documentation of the comparison of the observations and measurements. Implement continuity and homogeneity tests on the data series. Work close together and have feedback with instrument engineers, station network operators and climate data managers. CONCLUSIONS The net of autostations is increasing worldwide. The measurements produced by automated surface stations necessitate a change of climate data. In the period these inhomogeneities might have played still no large role. For the Standard Normals of the next period the majority of climate data will be produced by AWS and therefore the homogeneity of the time series has to be carefully handled. There is a necessity to review and prepare techniques for guidance material on data processing and quality control procedures involved in the conversion of conventionally-operated stations to automatically-operated stations. A summary of the various techniques and procedures used by the meteorological services is required and should be published. There is a necessity to recommend new criteria for the quality control and management of data from automatic stations. Needs to be means of identifying (flagging) instruments used and when changes introduced in the data acquisition networks. There should be established guidelines and proposed standards for the implementation and archiving of data from automatic meteorological stations, including averaging techniques and temporal and spatial resolutions. Especially in regard to algorithms for subjective/visual observations.

9 ACKOWLEDGMENTS I have to thank Mr Klaus Schulze,Senior Scientific Officer at the World Weather Watch Department within the World Meteorological Organization for providing me with several very helpful comments and communications. REFERENCES 1. Höhne, W., 1986: Automatische meteorologische Stationen; Entwicklungstendenzen, Systemaspekte und Einsatzprobleme. Zeitschrift für Meteorologie, 36 1, Karl T. R., Derr, V. E., Easterling, D. R., Folland, C. K., Hofmann, D. J., Levitus, S., Nicholls, N., Parker, D. E. and Withee, G.W.1995: Critical issues for long-term climate monitoring. Climate Change, Plummer, N., Collins, D., Trewin, B. and Della-Marta, P. 1999: Automatic weather Stations in Australia A Climate Perspective. Proceedings of the ICEAWS99, in Vienna, September 99, Austria 4. Rudel, E.1997: Report and review about data processing and quality control procedures involved in the conversion ofmanually operated stations to automatically operated stations. World Climate Programme: Data and Monitoring No.31, WMO-TD No Schulze, K Personal communication. 6. WMO, 1995: Guidance Spezification (functional) for a general purpose Automatic Weather Station, Prepared by the CIMO Rapporteur on Functional Specifications for Automatic Weather Stations, 7. WMO, 1997: Final Report of the Expert Meeting on Automation of Visual and Subjective Observations, held in Trappes/Paris, France, from May WMO, 1999: Draft Report of the Expert Meeting on Requirements and Representation of Data from Automatic Weather Stations, held in De Bilt, Netherlands, from April 1999