WMO RSO Intercomparison Experiment held in Brazil in 2001. Preliminary Results
Reinaldo B. da Silveira-INMET[1]
Gilberto F. Fisch-ACA/IAE/CTA[2]
Luiz Augusto T. Machado-ACA/IAE/CTA[2]
Alaor M. Dall’Antonia Jr.-INMET[1]
1. Introduction
This paper describes some preliminary results of the WMO RSO Intercomparison Experiment (WMO, 2000), held at the Brazilian Air Force Satellite/Rocket Launch Centre (CLA), situated at Alcantara city, State of Maranhão, at the latitude 2◦ 18’ and longitude of 44◦ 2.
The experiment, sponsored by the Instituto Nacional de Meteorologia (INMET-Brazil), was carried out from 21st May to 06th of June of 2001, including short periods of testing and packing. Its main goal includes the evaluation of the operational conditions of GPS, temperature and humidity sensors of operational radiosonde systems. We shall present the procedures carried out to perform and analyze data from 43 radiosonde observations, when 3, 4 or 5 distinct equipments, from different manufacture companies, flew simultaneously, using the same balloon. More information about this experiment will be available soon in a Technical WMO Report. Basically, we will show the available data sets, the assembly of this data on a data archive system and some results of the analysis among the radiosonde types.
The manufactures who took place at the experiment were Dr. Graw Messgeraete GmbH&CO, from Germany, using the DFM-97 radiosonde model; MODEM (Geolink), from France, using the GL-98 radiosonde type; InterMet Systems, from USA; Sippican, from USA, using MKII radiosonde; Vaisala, from Finland, using RS80 and RS90 radiosondes and Meteolabor, from Switzerland, using the Snow White humidity sensor and the Thygan, the humidity calibrator. Besides the radiosonde, the experiment had also a radar tracking and an automatic weather station, the Millos 500, both belonging to the CLA and local people from the Radar and Meteorological stations operated them, respectively. The Vaisala team brought in the laser ceilometer for measuring cloud heights.
The radiosondes measure pressure, temperature and humidity. Moreover, with the exception of MKII and GL-98, which compute the height indirectly via hypsometric equation (Wallace and Hobbs, 1977), the radiosondes also measure height directly. The wind is derived from satellite GPS system, using the differential GPS technique.
The 43 soundings were performed every 6 hours, starting at 00UTC. Flights at 00:00 and 18:00 UTC had 4 radiosondes and the 2000g balloon was used. RS80, RS90, (MKII + Snow White) and either GL98 or DFM-97 radiosondes flew at these times. Flights 06:00 and 12:00 had 3 radiosondes and the 1200g balloon was used. The radiosondes in these flights were RS80, MKII and either GL98 or DFM-97.
2. Data and post-processing method
Besides the data provided by the radiosondes, the experiment made possible a great amount of data, such surface observations of temperature, humidity, radiation, clouds, pressure and winds. Also available was a C-Band radar data, which is used to evaluate the GPS wind data.
The data source comprises also satellite images and synoptic mesoscale observations, as well as outputs from the MBAR, the high resolution NWP model of INMET. The meteorological surface parameters, which were observed during the RSO Intercomparison experiment, were measured through an automatic weather station from Vaisala Oyj (Millos 500). The data was measured by sampling at 1 observation each 5 min and hourly and daily values were processed after the field data collection. The weather situation can be split into two periods: a dry period from 20th May until 1st June (11.4 mm) and a wet period from 2nd June to 10th of June (76 mm). This wet period is likely to be associated with a large-scale disturbance over the region. During the dry conditions, there is only one event of rain (May 24 with 8.0 mm.day-1). Considering the whole period, the air temperature showed values between 24 C and 27 C and the relative humidity ranged from 75 up to 90 %. The lowest temperature values and highest moisture conditions were observed during the wet conditions. The integrated solar energy varied from 12 up to 24 MJ.m-2.day-1 and the values around 12 until 15 MJ.m-2.day-1 were observed during the rain. The wind speed values are between 1.0 and 3.5 m.s-1.
In order to have the whole radiosonde data under the same conditions, after the collection of the radiosonde data a post-processing was necessary to remove spurious data as well to adjust the time for each radiosonde. Thus, a time adjustment was applied by using the temperature, as it is the parameter that presents more agreement among the radiosondes, and as the RS80 was the radiosonde that participates in most of the flights, the temperature profile of RS80 was used as reference to adjust the time of the other radiosondes. One would observe that the choice of RS80 does not interfere in the results of the Intercomparison as far as parameters are concerned, since it only is used to drive the time of the other radiosondes.
The maximum time offset was considered in the time interval of about 20 seconds. In order to assure the best adjustment, this time step was defined larger than the largest time offset occurred during the radiosonde trial. Thus, the time offset was adjusted considering only the average time by which the radiosonde crosses a layer slight larger than the mixed layer, i.e., 160 seconds. The layer including the mixed layer and few meters higher has a larger temperature dynamics (vertical temperature variations). By using this layer it was assured a good adjustment, without needs of including all the radiosonde patches, which probably would add time offset due to a specific radiosonde system. We have then applied a mean squared error algorithm in the temperature profile, using about 20 seconds lag, for each flight between RS-80 and each one of the other radiosondes participating in the flight. The absolute error of the minimum time lag was considered as the time offset of each radiosonde related to the RS80 flight.
3. Results
A intercomparison data analysis was performed for zonal (east-west) and meridional (north-south) wind components, humidity, temperature, pressure and height. We will focus on some results of wind components and relative humidity analyses. Thus, the analyses can be summarized as comparisons of the vertical profiles for distinct weather conditions, as combinations of day, night, wet and dry conditions. Computing averages made further analyses and other statistics, such as deviations from average (bias), root mean squared error (RMS) between a radiosonde and a reference.
3.1 Wind component analysis
The radiosondes used at the experiment produced similar wind measurements. This can be seen from figure 1, which shows the vertical profile for flights 02 and 18. The figure 1 displays the zonal and meridional wind components for the radiosondes participating in those flights, as well the sondes average profile. A graph of the deviation from this average is also included in this figure. One can note that the meridional component of wind is weaker than the zonal component, as well as the deviation from the average. This is a consequence of the predominance of the trade wind pattern in the equatorial region. The main differences, although small (about 5m/s), occurred in flight 02 between 4200 and 5300 seconds (22 to 27 km). RS80 presents differences at levels between 3600 and 3800 seconds (18 to 19 km) in the flight 18.
Figure 2 displays the average profiles of both wind components as well as differences between pairs of radiosondes, where the references were, respectively, RS80, RS90, MKII and GL98. The average profiles were computed by considering all individual flights of each radiosonde type. The comparison shows that radiosondes agree well until 12500 metres, when the differences start to appear. The relative difference profiles show that the GPS wind components computed with DFM-97 and GL98 present the main differences related to that computed with Vaisala’s radiosondes and MKII. The maximum difference is of order of 12 m/s at 28 km, which are about 40 % of the maximum of the zonal component, which is considerable large. Unfortunately, the radar data could not yet be used as reference, as it is still under processing. It would help to explain the differences found in the analysis.
Figure 1: Vertical profiles of flight 02 (above) and flight 18 (below), for zonal (left) and meridional (center to right) wind components, for MKII, DFM97, RS90 (flight 02) and average of each level. The latter is also for zonal (center to left) and meridional (right) wind components.
Figure 2: Comparison of zonal and meridional wind components for all radiosondes. The 2 first plots are average profiles of the components. The other plots refer to differences from RS80, RS90, MKII and GL98, respectively.
3.2 Relative humidity analysis
There are a number of remarks considering the measurements of relative humidity. For simplicity we illustrate some of the problems as follows. Figure 3 shows the average profile of the reference sensor Snow White and of the radiosondes in the experiment.
Figure 3: Average relative humidity profiles for the radiosondes in the experiment, including the Snow White sensor.
One can note the large deviation of Snow White after 2000 seconds, from the radiosondes and the large dispersion of mean values, as displayed in figure 3. Table I gives figures, which illustrate the average values of bias and RMS, within three successive layers. The first layer goes from surface to 600 seconds (~3000 m); the second layer is between 600 and 1600 seconds (3 to 8 km) and the third layer extending from 1600 seconds up to the top of the sounding. Although the values presented in table I were smoothed out by the average process, some conclusions can be gathered from these results. The RS80, RS90 and GL98 radiosondes produced close measurements of relative humidity. Whereas major differences occur for levels beyond 8000 m, we do not have yet means to indicate which equipment is giving the correct values. The MKII over estimates the relative humidity values when compared to other radiosondes, presenting high values for RMS. The Snow White behaves quite well at low levels, but the RMS increases considerably with the time, reaching very high values at the top of the sounding. The DFM-97 under estimates the relative humidity values when compared to other radiosondes. However, it is strongly correlated to RS90, mainly at levels below 8000 m.
Table I. Comparison of pairs of average bias and RMS, at 3 layers, considering the radiosondes and the Snow White reference.
Relative differences / Average valuesBIAS (%) / RMS (%)
1st layer / 2nd layer / 3rd layer / 1st layer / 2nd layer / 3rd layer
RS80-RS90 / -1.52 / +1.04 / -5.14 / 3.42 / 4.58 / 7.76
Sippi-RS90 / +7.25 / -1.99 / +1.29 / 9.27 / 13.61 / 13.42
Geo-RS90 / -1.42 / +1.17 / -4.58 / 3.85 / 5.38 / 8.37
Graw –RS90 / -4.39 / +0.40 / +10.86 / 6.16 / 7.52 / 17.09
Snow –RS90 / +0.61 / -1.28 / +24.11 / 4.81 / 12.40 / 34.40
Sippi –RS80 / +7.17 / -2.62 / +2.91 / 10.08 / 13.97 / 13.25
Geo-RS80 / +1.20 / +1.66 / +1.92 / 5.25 / 7.48 / 6.93
Graw –RS80 / -3.84 / -0.98 / +13.58 / 5.74 / 6.21 / 17.93
Snow –RS80 / +2.37 / +0.60 / +28.19 / 5.70 / 13.01 / 39.34
Geo-Sippi / -6.79 / +3.90 / -0.52 / 9.89 / 12.87 / 15.43
Graw –Sippi / -9.65 / +2.55 / +6.97 / 12.40 / 16.52 / 15.31
Snow –Sippi / -6.79 / +3.27 / +24.07 / 8.01 / 11.46 / 35.74
Snow –Geo / +3.10 / -1.81 / +28.02 / 5.41 / 9.74 / 36.98
Snow-Graw / +3.28 / +0.07 / +12.92 / 6.83 / 13.20 / 27.43
Acknowledgements
The WMO/INMET RSO Intercomparison Experiment was possible thanked to many institutions either in Brazil or outside the country. The authors wish to thank all of them as well as the participant vendors, operators and those who direct or indirectly collaborated for the success of the Experiment. Part of the Experiment was sponsored by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) under contract #2000/15124-7. The authors are also grateful to Dr. Rosa Marques, who helped in the interpretation of meteorological data and synoptic conditions happening during the experiment; Luiz Fernando Sapucci, for the analysis of the relative humidity data and Daniela Cristina Felix Pereira, who helped in the analysis of GPS wind data.
References
Wallace, J. M., and Hobbs, P. V., 1977: Atmospheric Science. Academic Press, INC. San Diego California, 467 pp.
WMO, 2000, Final Technical Report of the International Organizing Committee for the WMO Intercomparison of GPS Radiosondes. Brasilia, Brazil.
[1] Corresponding address: Instituto Nacional de Meteorologia
Eixo Monumental – Via S1- Brasilia-DF. CEP 70680-900-Brazil
Email and .
[2] Corresponding address: Centro de Técnico Aeroespacial (CTA/IAE/ACA)
Praça Marechal Eduardo Gomes, 50. CEP 12228-904.
São José dos Campos- SP-Brazil. Email: and
[2]
[1]