TOGA COARE was a multidisciplinary, international research effort that investigated the scientific phenomena associated with the interaction between the atmosphere and the ocean in the warm pool region of the western Pacific. The field experiment phase of the program took place from 1 November 1992 through 28 February 1993 and involved the deployment of oceanographic ships and buoys, several ship and land based Doppler radars, multiple low and high level aircraft equipped with Doppler radar and other airborne sensors, as well as a variety of surface based instruments for in situ observations.
Six Integrated Sounding Systems (ISS) were installed and operated for various lengths of time during the TOGA COARE Intensive Observing Period (IOP) and Enhanced Monitoring Period (EMP). The ISS was developed jointly by the Surface and Sounding Systems Facility of the National Center for Atmospheric Research Atmospheric Technology Division and the Aeronomy Laboratory of the National Oceanic and Atmospheric Administration. The ISS combines four separate subsystems:
- Balloonborne radiosonde navaid (Loran or Omega) sounding system - Enhanced surface observing station - 915 MHz Doppler clear-air wind profiling radar - Radio Acoustic Sounding System (RASS).
The ISSs were located at Kavieng and Manus Island (Papua New Guinea), Nauru (Republic of Nauru), Kapingamarangi (Federated States of Micronesia), and on two research vessels, Shiyan #3 (Experiment #3) and Kexue #1 (Science #1).
ISS soundings are available for the following time periods:
Kapingamarangi (FSM), WMO#91434 IOP: 00 06 12 18 UTC EMP: 00 12 UTC Data available: 110792-290693
Kavieng (PNG), WMO#94076 IOP: 00 06 12 18 UTC Data available: 011192-010393
Manus (PNG), WMO#94044 IOP: 00 06 12 18 (IOP) Before 1 Nov 92: 00 UTC After 1 Mar 93: 00 12 UTC Data available: 010792-270992, 181092-301192, 241293-220193, 280193-300693
Nauru, WMO #91530 IOP: 00 06 12 18 UTC EMP: 00 12 UTC Data available: 020992-300693
R/V Kexue #1 IOP: 00 06 12 18 UTC Data available: 091192-111292, 151292-230193, 280193-190293
R/V Shiyan #3 IOP: 00 06 12 18 UTC Data available: 101192-111292,181292-220193,260193-180293
1. Surface Data
The ISS surface meteorological instrument installation includes several sensors mounted on two separate towers as well as a rain gauge mounted independently. An anemometer is mounted on the top of a ten-meter tower. Temperature and humidity sensors are mounted on the end of a one-meter boom attached to the ten-meter tower at two meters above the surface. The temperature and humidity sensors are aspirated and protected with a radiation shield. The pressure sensor is housed in the box containing the Campbell CR 10 datalogger. That box is mounted on the ten-meter tower at one meter above the surface.
The radiation sensors are mounted on a one-meter boom on the top of a separate two-meter tower. The standard ISS radiation sensors include an up-looking solar radiation sensor and a net radiation sensor. In situations which require more complete radiation measurements, additional sensors can be added.
The output from all the sensors is directed to the Campbell datalogger for processing. The Campbell datalogger, which is independently programmable, typically generates one-minute average data which are sent via RS-232 to the ISS Sun workstation. The data input to the Campbell datalogger are five-second sample data.
Florida State University under the direction of Dr. James O'Brien and Dr. David Legler, produced a comprehensive TOGA COARE surface meteorological dataset, which includes the quality-controlled ISS surface met data. Data, which are in netCDF as well, can be accessed through the WWW ( http://www.coaps.fsu.edu/COARE/ ). The quality control process was executed in two steps. All surface meteorological data were assessed beginning with a low-level scan. This scan included, at a minimum, automated checks for coding errors, spurious spikes, instrument drift or failure, and unrealistic values. Each data record from each platform was visually inspected and all data points were flagged as either passing or failing the low-level inspection. The second step in the QC process will be the inter-comparisons of data across platforms/instruments. Comparisons between these data will be exploited to verify the data from the platforms and differences noted.
2. Upper-air Data
The balloonborne radiosonde navaid (Loran or Omega) sounding system is the standard NCAR "CLASS" sounding system. This sounding system typically uses the Vaisala RS-80 L or the Vaisala RS-80 N radiosondes which use Loran and Omega radionavigation signals respectively for windfinding. Launching configurations can vary depending on the installation. An ISS site may have an enclosed air conditioned launcher, a "bag" launcher, or possibly no launcher at all. The balloon and sonde are secured to structures available before release in the latter case.
Final, quality-controlled upper-air data are available from UCAR/OFPS though the CODIAC system, http://data.eol.ucar.edu/codiac/. The data have passed through the JOSS Quality Control process after the original data were converted into the JOSS Quality Control (ASCII) Format (QCF). Data were processed, quality controlled and archived in the highest "native" resolution. JOSS' QC process consists of three stages: a) automatic internal consistency checks, which include 17 automatic checks to evaluate the internal consistency of each individual sounding ("reasonable limit" check); b) vertical consistency checks which include comparisons between adjacent vertical levels; c) visual examination of each sounding. To establish the consistency of upper air measurements across the entire COARE domain, spatial quality of the sounding data was assessed by comparing neighboring data. The general methodology used by JOSS is to produce statistics (mean, standard, deviation, variance) from a simple weighted averaging technique.
This dataset underwent significant processing by NCAR/SSSF prior to its arrival at OFPS. The reprocessing efforts and documentation files regarding the quality of each sounding are available for each sounding location from the TCIPO and UCAR/JOSS.
3. SUDS-generated skew-t plots and analysis files for TOGA COARE ISS upper-air data
SUDS-generated skew-t plots and analysis files for TOGA COARE ISS upper-air data are available from NCAR/ATD. Since sounding data for a project are gathered across many days from several stations, the NCAR/ATD archive organizes the information in two ways, by day and by station. For a given project, all of the sounding information for a date are grouped together in the "by day" documents, and all of the information for a station are grouped together in the "by station" documents. Moreover, to make the archive more usable, each sounding reference has a color identification that indicates the sounding's launch station.
For more information, contact David Ecoff at NCAR/ATD (email@example.com).
+ Surface Humidity Data: The humidity sensor failed at Kapingamarangi at 16Z on the 17 November 1992 and, due to shipping problems, it was not replaced until 02 February 1993. There are portions of good data after 17 November, as the sensor began to ... drop out intermittently at that time. After the 25 November at 06Z there were no good humidity data. A replacement humidity sensor and a psychrometer were sent to Kapingamarangi. The psychrometer arrived before the replacement sensor. Psychrometer data were taken but recorded separately beginning 4 January 1993 at 17Z. The data were taken every hour through the end of February (with a number of gaps). That data are available in ASCII format. They are tabulated by date and sounding time (sounding times are GMT). The data format is: local time (GMT + 11)-Temperature-Wet Bulb Temperature-.....
+ Surface Pressure and Temperature Data: The surface pressure data appear noisy at times. This can be seen in the time series plots as a series of a few points offset up to one millibar from the mean pressure trace. This is seen over several time periods: 9-15 December 1992; 14-19 January 1993; 7 February 1993; 10-15 February 1993; 27 March - 22 April 1993; and 25-31 May 1993. There are a few erroneous spikes in the temperature data. Most of these occurred 19-23 February 1993.
+ Surface Data Gaps: There are several data gaps when no surface data were recorded. These were caused by various things - short term disk problems, power interruptions, etc. The gaps are 23-26 August 1992 (72 hours); 3-6 October 1992 (60 hours); 15-16 October 1992 (20 hours); 5 December 1992 (7 hours); 8 March 1993 (4 hours); 21 April 1993 (20 hours); and 21 June 1993 (21 hours). There were other shorter data gaps as well. These were typically on the order of two hours or less.
+ Results from FSU Quality Control: Some variables in these data are considered to be in poor condition. Several problems occurred in the pressure data. A preprocessing program flagged many values as either out of the acceptable range or for being greater than four standard deviations away from the da Silva climatological mean (da Silva, et al. 1994). Numerous discontinuities occurred just after small breaks in the data and were flagged as suspect data. Finally, rhythmic spikes occurred on many days. These one millibar, downward spikes usually occurred near the even hour and in some cases lasted for several days. These values were flagged as spikes although they may be the result of a data logger or instrument malfunction.
As noted in the ISS Report (Miller 1994), the humidity sensor, in a Vaisala Humicap Model HMP35C, began malfunctioning at 16Z on the 17th of November, 1992 and was not replaced until the 2nd of February, 1993. Normal relative humidity values for Kapingamarangi are in the range of fifty to ninety percent, however, the humicap started reporting values between ten and fifty percent. The drops to the lower relative humidity values were very abrupt, often occurring in one or two minutes. It is notable that the relative humidity trends were still realistic (i.e. the diurnal cycle was present in the data) even though the absolute value of the relative humidity was unrealistically low. The relative humidity data were flagged during these periods as a malfunctioning sensor.
Because the temperature sensor and humidity sensor are coupled in an HMP35C, many temperature values were flagged as suspect data or as a malfunction. Other suspect data flags were added to interesting discontinuities which occurred just after temperature values were out of the realistic range. These flags were usually associated with the discontinuities which occurred in the data.
The wind direction, wind speed, and precipitation data do appear to be in very good condition, and there were 26 days that had no flags added.
2. Kavieng Station
+ Surface Pressure Data: The surface pressure sensor did not operate correctly from start-up through 00Z 16 November 1992. A replacement was installed at that time and performed reliably through the remainder of the project.
+ Surface Data Gaps: There were a few gaps in the surface data at Kavieng. The longest gap ran from 5 November at 08Z until 16 November 1992 at 00Z. There were two other significant gaps, one from 16Z 6 December until 00Z 7 December 1992, and one from 00Z until 20Z on 22 December 1992.
+ Radiation Data: During January 1993, the up-looking and down-looking pairs of pyranometers and pyrgeometers were switched (a standard practice to detect any bias in the instruments). The up-looking pyranometer became the down-looking pyranometer and vice-versa. The up-looking pyrgeometer became the down-looking pyrgeometer and vice-versa. However, due to a number of factors, things were not switched properly (both hardware and software) and some confusion resulted. Corrected data from this period can be obtained from the ARM PROBE archive; contact Chuck Pavloski (firstname.lastname@example.org). Note that two data fields in the Kavieng surface data files are not used. The information in them should be ignored. These fields are sol and soldiff (Kavieng only!).
+ Results from FSU Quality Control: The data from Kavieng are in very good condition, however, two main problems occurred during the IOP. The primary reason for flags was a malfunctioning atmospheric pressure sensor which reported a constant pressure of 800 mb from 0000 UTC on November 1, 1992 through 2400 UTC on November 4, 1992. No pressure data were available from 0000 UTC on November 5, 1992 through 2400 UTC on November 15, 1992. The sensor was replaced and the new sensor appeared to function properly from 0000 UTC on November 16, 1992 through the end of the IOP (Miller 1994). The second problem was the lack of data from 0000 UTC on November 5, 1992 through 2400 UTC on November 15, 1992. Data were not collected for any of the variables for these eleven days.
The data analyst does not foresee any problems in using this data. The 22 interesting feature flags were added to the wind direction, wind speed, temperature, and relative humidity profiles to highlight what are believed to be convective events (TCIPO 1993) which produced dramatic changes in these variables. Satellite imagery was used to identify these events. Flags were placed at peak wind speeds and the corresponding directions and at the beginnings and ends of significant drops or rises. Minor changes were noticed in other variables but were not flagged.
3. PRC R/V Kexue 1 (Research Vessel)
+ Earth-relative Surface Wind Data: The shipboard surface wind measurements were made with the standard RM Young Wind Monitor. They were corrected for ship motion using ship speed obtained from a GPS navigator and ship direction (heading of the ship's longitudinal axis) obtained from a magnetometer. The GPS direction (obtained from ship position differencing) was not adequate for use with the wind algorithm when the ship was "on site" (anchored or drifting). Comparisons of the magnetometer data to the GPS directional data during cruises have generated some concern regarding the operation of the magnetometer. In comparisons done for the PRC R/V Kexue 1, differences in the two directions have been seen to vary, at times as much as 20 degrees. This was the case after magnetic declination was taken into account. Thus there is an unknown uncertainty in the wind direction measurement. In addition to the magnetometer uncertainty, there were periods when the GPS navigator failed. During these periods the surface wind data were not corrected for ship motion. The uncorrected wind data are available but they will be of limited use. The direction will be bad (unless the longitudinal axis of the ship happens to be pointing to the north). The speed may be useful if the ship isn't moving or is only moving very slowly relative to the real wind (often the case when the ship was on station). Periods when the GPS failed and no corrected winds were obtained are 1-9 November 1992 (prior to first deployment on station); 13-14 December 1992 (44 hours); and 11-28 January 1993 (17 days).
+ Surface Pressure and Temperature Data: There were a few long periods of noisy surface pressure data. This can be seen in the time series plots as a series of a few points offset up to one millibar from the mean pressure trace. The times when this occurred are 25 November through 13 December 1992; 16-25 December 1992; and 13-17 January 1993. Much of the surface temperature data from the PRC R/V Kexue 1 is noisy. Most of the temperature data recorded after 00Z on 20 November 1992 show a variance which is too high. Use caution when using this data. Some sort of averaging would likely salvage some useful information.
+ Surface Data Gaps: There was very little lost data from the PRC R/V Kexue 1. The only significant data gap occurred from 09Z until 22Z on 15 December. There were a few other much shorter (on the order of two hours) data gaps, but overall the data recording system performed quite well.
4. PRC R/V Shiyan 3 (Research Vessel)
+ Earth-relative Surface Wind Data: The shipboard surface wind measurements were made with the standard RM Young Wind Monitor. They were corrected for ship motion using ship speed obtained from a GPS navigator and ship direction (heading of the ship's longitudinal axis) obtained from a magnetometer. The GPS direction (obtained from ship position differencing) was not adequate for use with the wind algorithm when the ship was "on site" (anchored or drifting). Comparisons of the magnetometer data to the GPS directional data during cruises have generated some concern regarding the operation of the magnetometer. In comparisons done for the PRC R/V Shiyan 3, differences in the two directions have been seen to vary, at times as much as 8 degrees. This was the case after magnetic declination was taken into account. Thus, again, there is some uncertainty introduced into the wind direction measurement.
+ Surface Pressure, Temperature, and Humidity Data: There were no ISS surface pressure data from the PRC R/V Shiyan 3 during the first cruise. The sensor fell overboard during a maintenance effort in rough seas. A replacement was installed at the first port of call. Recording of good pressure data resumed at 0330Z 15 December 1992. There were a couple of short periods of bad temperature and humidity data in January 1993. Use caution with data obtained between 00Z and 04Z on both 24 January and 30 January. In addition, there were periods of noisy temperature data. The temperature data variance was too high early in the first cruise (12 November 1992) and was also possibly a bit high in early December 1992.
+ Surface Wind Data: The are periods of very high, spiky wind speed data in the last half of December. This can be seen in the time series plots from roughly 17-31 December 1992. Although the data appear suspect, it is likely that they may be reasonable as they correlate well with periods of significant rain and may be gusts associated with storm cells.
5. Manus Station
+ Humidity Data: The first good surface humidity data at Manus was obtained on 02 October 1992. Prior to that, there were no good humidity data.
+ Radiation Data Initially, a LICOR radiometer was used to measure solar radiation. The output of that instrument was in kilowatts per square meter (kW/m2). That sensor was replaced by an Eppley pyranometer at the end of October 1992. The output of the Eppley sensor was in watts per square meter (W/m2). The difference in output is reflected in the time series plots (the units for the solar radiation data plots of July through October 1992 SHOULD BE "kW/m2"). There was no longwave (ir) radiation sensor at Manus until 1 November 1992. The Eppley Pyrgeometer installed to measure longwave radiation (incoming - the sensor was up-looking) functioned properly for the remainder of the Extended Monitoring Period (through June 1993), although it was not properly calibrated. The output of the sensor was about 80% of what it should have been. Comparisons with incoming longwave radiation data from the nearby Kavieng ISS site showed that the Manus sensor output was roughly 80% of that obtained at Kavieng. The Manus pyrgeometer output consistently stayed at that 80% level throughout the project.
+ Data Loss: There were a few significant periods of complete surface data loss. There were no surface data obtained from 20 October until 1 November 1992. There were also no data recorded from 16 May 1993 through the end of June 1993 due to disk problems. There were a few other shorter periods, up to two and one-half days, where data were lost. These periods are clearly apparent in the time series plots provided. There were also periods of intermittent surface data in July 1992.
+ Results from FSU Quality Control: The data from Manus are in very good condition. The only problem that occurred, and the result of most of the flags, involves the relative humidity data. The 10,440 flags were added to relative humidity values slightly greater than 100 percent. These values may be the result of fog or, possibly, a miscalibrated relative humidity sensor.
The data analyst does not foresee any problems in using this data. The 48 interesting feature flags were added to the wind direction, wind speed, atmospheric pressure, temperature, and relative humidity profiles to highlight what are believed to be convective events (TCIPO 1993) which produced dramatic changes in these variables. Satellite imagery was used to identify these events. Flags were placed at peak wind speeds and the corresponding wind directions and at the beginnings and ends of significant drops and rises. Minor changes were noticed in other variables but were not flagged.
6. Nauru Station
+ The surface data quality and recovery from Nauru are quite good. There were no data recorded until late August 1992. There is a 24 hour period of missing data 6 and 7 November 1992. There are only a few other short periods, on the order of a couple hours, of missing surface data. Again, these are apparent from the time series plots provided.
+ Results from FSU Quality Control: The data from Nauru were considered to be in good condition except for two problems that occurred in the pressure data. The first problem was the presence of rhythmic, one millibar, downward spikes in the atmospheric pressure data. These spikes were similar to those seen at Kapingamarangi. They occurred near almost every even hour on many days and sometimes lasted for periods of up to several days. The other problem in the Nauru atmospheric pressure data is the large number of values that were flagged for being greater than four standard deviations from the da Silva climatological mean (da Silva, et al. 1994). These pressure values were usually rather low. As a final note, the data analyst does not foresee any problems in using this data.
Cole, H., C. Martin, M. Spowart and W. Ecklund, "Progress Report #1 for the Development of a Shipboard Integrated Sounding System Stable Platform" February 1991, Atmospheric Technology Division, National Center for Atmospheric Research, 22 pages.
da Silva, A.M., C.C. Young and S. Levitus, 1994: Atlas of Surface Marine Data 1994, Volume 1: Algorithms and Procedures. NOAA Atlas Series. In preparation.
Dabbert, W.F., M.C. Michaelis, J.W. Militzer, J.A. Businger, A.C. Delany, and S.R. Semmmer, "NCAR Surface Meteorological Measurement Facilities" 1989 19th Conference on Agricultural and Forest Meteorology, 7-10 March, Charleston, SC.
Gage K.S., W. Ecklund, R. Strauch, E.R. Westwater, and W.L. Smith "The Integrated Sounding System - A New Observing System for Mesoscale Research" 1991, Seventh AMS Symposium on Meteorological Observations and Instrumentation, 13-18 January, New Orleans, LA.
Miller, E.,, 1994: TOGA COARE Integrated Sounding System Data Report, Volume IIa: Surface Data-IFA Sites. NCAR Dabbert, W.F, H.L. Cole, P. Hildebrand, T. Horst, Y.H. Kuo, C. Martin,
TOGA COARE International Project Office (TCIPO), 1992: TOGA COARE Operations Plan, Working Version September 1992. University Corporation for Atmospheric Research, Boulder, CO 80307, 138 pp.
TOGA COARE International Project Office (TCIPO), 1993: TOGA COARE Intensive Observing Period Operations Summary. University Corporation for Atmospheric Research, Boulder, CO 80307, 505 pp.
TOGA COARE International Project Office (TCIPO), 1994: Summary Report of the TOGA COARE International Data Workshop, Toulouse, France, 2 - 11 August 1994, University Corporation for Atmospheric Research, Boulder, CO 80307, 170 pp.
Webster, P.J., and R. Lukas, 1992: TOGA COARE: The Coupled Ocean- Atmosphere Response Experiment. Bull. Am. Meteorol. Soc. 73, 1377-1416.
World Climate Research Programme (WCRP), 1985: Scientific Plan for the TOGA Coupled Ocean-Atmosphere Response Experiment. WCRP Publications Series, No. 3 Addendum, World Meteorological Organization, Geneva, 96 pp.