NDBC TAO: Oceans '94

TOGA-TAO Array Sampling Schemes and Sensor Evaluations

L.J. Mangum, H.P. Freitag, and M.J. McPhaden

TAO Project Office Pacific Marine Environmental Laboratory National Oceanic and Atmospheric Administration Seattle, WA 98115

Proceedings of the Oceans '94 OSATES, Brest, France, 13-16 September 1994, Vol II, pp. 402-406.

Abstract The TAO array will be fully deployed in the Pacific Ocean by the end of 1994, with a total of 64 ATLAS wind and temperature moorings and 5 PROTEUS/current meter moorings. Implementation of the TAO array began in the latter half of TOGA, and will be continued after TOGA ends in December 1994 as part of national and international programs to improve short-term climate analyses and prediction. The array is directed by NOAA's Pacific Marine Environmental Laboratory (PMEL) and involves international cooperation with scientists and organizations in U.S.A., Japan, France, Korea, and Taiwan. A review of the TAO program, sampling information, and an evaluation of sensor performance are presented.


The sudden onset of the 1982-83 El Niño-Southern Oscillation (ENSO) event, which was not detected until several months after the start of the event, clearly demonstrated the lack of real-time observations in the equatorial Pacific. PMEL, which had been involved since 1976 with the first successful equatorial taut-line surface moorings, intensified the development efforts for a low-cost, year-long mooring which would transmit critical surface and subsurface data in real time. First deployments of a prototype system were made in 1984. Under the direction of the late Dr. S.P. Hayes, ATLAS thermistor chain moorings were deployed as part of the Equatorial Pacific Ocean Climate (EPOCS) program in the eastern equatorial Pacific along 110W and 140W in 1985. Additional moorings were deployed during the same time period in the western Pacific in cooperation with the Institut Français de Recherche Scientifique pour le Développment en Cooperation (ORSTOM) in New Caledonia. By the end of 1985, there were nine ATLAS moorings and two current meter moorings deployed in the Pacific. From this modest beginning, Hayes envisioned a much larger array that would span the Pacific Ocean and make oceanographic and meteorological data available in real time for short-term climate studies, in particular those focusing on ENSO. In early 1987, Hayes first presented the concept of the TOGA Tropical Atmosphere Ocean (TAO) program, which called for an expansion of the number of surface moorings from 11 to almost 70 during the second half of TOGA (1990-1995) [1]. During the next 3 years, international support for the TAO program grew, particularly in the western Pacific. It is presently supported through cooperation of the United States, Japan, Korea, Taiwan, and France.

Fig. 1. Implementation of the TAO array.

Fig. 1 shows the growth of the TAO array from near the beginning in 1988, the status before the major expansion, and its present configuration. A description of the recent 1991-93 ENSO event using TAO data is provided in [2].


Standard measurements on all TAO ATLAS and PROTEUS moorings include surface winds, air temperature, relative humidity, sea surface temperature (SST), and subsurface temperatures to a depth of 500 m (Fig. 2).

Fig.2. Schematic diagram of ATLAS wind and thermistor chain mooring. These systems are deployed in the deep ocean in water to 6000-m deep.

PROTEUS moorings also sample subsurface currents from an acoustic Doppler current profiler in the upper 300 m [3]. ATLAS moorings are deployed for a 1-year nominal deployment and the PROTEUS moorings are designed for a 6-month deployment. All moorings and instrumentation are assembled by the TAO group at PMEL.

Surface and subsurface data from the TAO array are telemetered in real time via Service Argos. These data are available to the community directly from PMEL through Internet anonymous ftp transfers, and through the TAO workstation display [4]. ATLAS mooring data are also distributed via the GTS, as described in a later section.

The on-board instrumentation and sampling for these moorings has evolved during the implementation of the array. Real-time wind data from the current meter moorings was tested in 1983 [5]; wind sensors were added to the ATLAS moorings in 1986 for real-time transmissions. Relative humidity sensors were tested beginning in 1989 to provide information for boundary layer studies and have now become part of the standard instrumentation on each mooring. Early deployments averaged winds for 2 hours, and for early ATLAS moorings, temperatures were averaged over 8 hours.

The present averaging and sampling schemes were first introduced in 1990 and existing electronic packages were upgraded as possible during the past 3 years. For wind, air temperature, SST, and relative humidity, both daily averaged and the most recent hourly value for each satellite overpass are transmitted. In addition, daily averaged subsurface temperatures from ATLAS moorings and daily averaged current profiles from PROTEUS moorings are telemetered. Air temperature, sea surface temperature, subsurface temperatures, and relative humidity are sampled every 10 minutes starting at 10 minutes past the hour, and averaged at the top of the hour. The time word for temperatures and humidity is the end of the averaging period. Winds are sampled at a rate of 2hz for 6 minutes centered on the top of the hour and then vector averaged. The 6-minute sampling period was chosen to remove effects of buoy motion on speed and direction. The time word for winds is centered in the sampling period.


In this section we focus on calibration of surface meteorological data from all TAO moorings and subsurface temperature data from ATLAS moorings. Accuracy of subsurface current meter mooring measurements has been discussed elsewhere [6,7,8].

TAO moorings were developed by PMEL using sensors available from commercial vendors. PMEL's Engineering Development Division designed the electronics hardware and software packages which digitize, record, and pass the sensor data to the Argos transmitter. TAO technicians construct the electronic instrumentation and are responsible for all sensor calibrations and checkout. Pre-deployment and post-recovery calibrations are performed on all sensors at PMEL, enabling estimates of sensor accuracy and drift. The instrumental error for each type of sensor is a combination of initial accuracy when a sensor is deployed (RMS errors for pre-deployment calibrations) and drift of the sensors over the deployment period (difference between pre- and post-deployment calibrations). Reference [9] contains complete details of calibration methods and results for the temperature and humidity sensors. A similar study for wind sensors is presently underway. Table 1 contains a summary of sensor information and accuracies for the standard TAO sensors.

Air temperature and humidity measurements are made at a height of 3 m using Rotronic Instrument model MP-100 probes mounted in a baffled housing which is designed to reduce the effects of solar heating. An analog to digital converter is used to convert the output voltage to a digital signal which is stored in memory. Calibrations are a two-stage process where the sensor is calibrated separately from the electronics (I/O) boards. The air temperature sensors are calibrated at PMEL by emersion in a controlled water bath and measuring their output voltages at seven temperatures ranging from 14C to 32C. The humidity sensor was calibrated in a calibration chamber using saturated-salt solutions as humidity standards. I/O boards were calibrated by applying a known voltage and recording the digital output. RMS errors and drifts were larger for the sensors then for the I/O boards. The accuracy of the air temperature was within manufacturer's specifications.

TABLE 1. Resolution and accuracies are listed for the standard ATLAS sensors. The manufacturer's specifications are listed as available. The resolution is determined by the onboard sensor sampling software. The single calibration RMS is the RMS error of individual pre- or post- deployment calibrations. Pre/post calibration differences are a measure of the calibration drift over a deployment. TBD indicates errors to be determined. Resolution and accuracy for surface variables from PROTEUS moorings are similar except for relative humidity as discussed in the text (see reference [9]).

The relative humidity errors were larger than specified and it is believed that these are due to calibration method as well as environmental fouling of the sensors due to salt encrustation after 1-year periods in the field. PROTEUS humidity accuracies were considerably better (1.8%) than the ATLAS accuracies (4.1%). Calibrations of these two groups of sensors were performed separately using slightly different techniques (sensor filter caps were taken off prior to calibration of PROTEUS sensors and left on for ATLAS sensors). Rotronic states that clogged filters may significantly increase the sensor response time. Different deployment lengths (6 months for PROTEUS and 1 year for ATLAS) may also contribute to the errors. A new calibration chamber for relative humidity has been recently procured and is capable of simultaneously calibrating up to 20 sensors through automated procedures that will decrease technique errors as well as allow for further studies of sensor response. With the higher accuracies of the PROTEUS relative humidities and the new calibration chamber, we hope that relative humidity errors will meet manufacturer's specifications.

Water temperatures to a depth of 500 m are measured with Yellow Springs Instrument Company 46006 thermistors. At each subsurface depth, the temperature interface converts the resistance of the thermistor into a voltage, then uses a voltage to frequency (V/F) converter to time multiplex the data up to the surface electronics package. The thermistor and V/F counter were calibrated as one unit for subsurface temperatures by immersing the potted boards and resistor into a controlled water bath and recording the digital output. SST sensors are mounted on the bridle of the buoy at a nominal depth of 1 m. SST instrumental errors are within manufacturer's specifications. Subsurface temperatures show a larger drift between calibrations that may be partly a result of errors in the calibration data base (due to the large number of calibrations that were entered manually until new automated procedures were developed 1 year ago), as well as instrument performance. Further studies are underway to examine long term stability of these sensors.

Winds are measured at a height of 4 meters above the ocean surface using an RM Young wind sensor (model number 05103). Its range is 0 to 50 m/s, with a threshold velocity of less than 0.4 m/s as specified by the manufacturer. The electronics package also contains a fluxgate compass (either EG&G or KVH) which is sampled at the same rate and this is used to obtain wind directions relative to true north. Resolution of the compass and vane is better than 1.4 degrees. Maximum single calibration residuals for winds are typically 0.2 m/s for speed and 3 degrees for direction. A small number of pre- and post-deployment calibration pairs suggests drift of less than 0.5 m/s for speed and 10 degrees for direction. This assessment of wind sensor drift should be considered preliminary until such time as a more thorough analysis is completed.


Hourly surface data (wind speed and direction, air temperature, and sea surface temperature) from the ATLAS moorings are presently transmitted over the GTS network in DRIFTER format by Service Argos. In addition, daily averaged subsurface temperatures are sent out in a separate GTS message for each buoy. Calibration information for each sensor is sent to Service Argos by PMEL when new sensors are deployed. Daily status checks on the entire array are performed at PMEL to detect any sensor failures, and if problems are found, the sensor is removed from the GTS distribution. New Argos software that has been running since early 1994 enables PMEL to automatically make updates in the status of sensors at Service Argos whenever problems are detected in the array.

ATLAS buoys only transmit for two 4-hour windows each day to conserve transmitter battery power during the 1-year deployment. In the tropical Pacific, there are typically three to four satellite overflights per day during this 8-hour period. On each satellite pass, the most recent hourly data is received as well as the previous daily averaged data. Transmission windows were originally set up to try to maximize satellite coverage for the array and the buoys transmit from 0600 1000 and 1200 1600 (buoy local time) along each longitude. Unfortunately one of the polar-orbiting satellites has been drifting, which reduces the number of transmissions that are presently seen each day. Table 2 lists the number of ATLAS buoys along each longitude as a function of time (GMT hours) that the buoys actually transmit. The last column is the maximum number of platforms that transmit during each hour. Maximum transmissions are centered near 2100 z; no data transmissions should be seen between 0600 z and 1200 z. Since there are typically only three satellite passes per day, the number of GTS surface messages from the ATLAS moorings during any specified time period will be considerably less than those listed in Table 2. A typical number of surface GTS messages seen during any 24-hour period would be 170 for the entire array.

TABLE 2. Transmission information for ATLAS moorings in TAO Array across the equatorial Pacific.

Although only 3-4 individual hourly values per buoy each day are available through the real-time transmissions, the complete hourly time series of surface values (air temperature, sea surface temperature, surface winds, and relative humidity) are available from onboard RAM data storage when the electronics package is recovered. These high resolution surface data files are now available via anonymous FTP file transfer as well as from the new release of the TAO Software Display package [4].


The TAO array has been designed to provide real-time temperature and wind data from the tropical Pacific to the oceanographic and meteorological community. These data are available directly from PMEL via Internet anonymous ftp file transfer and through a new release of the TAO Display Software [4]. ATLAS data are also available in real time via the GTS. The last two scheduled moorings for the array will be deployed by the end of 1994, completing the array that was envisioned over 7 years ago. The array will be continued as part of TOGA follow-on programs focusing on improved short-term climate analyses and prediction, namely the Global Ocean Observing System (GOOS), the Global Climate Observing System (GCOS), the World Climate Research Programs Study of Climate Variability (CLIVAR), and the US Global Ocean-Atmosphere-Land System (GOALS) program.


[1] S.P. Hayes, L.J. Mangum, J. Picaut, A. Sumi, K. Takeuchi, "TOGA-TAO: a moored array for real-time measurements in the tropical Pacific Ocean," Bull. Am. Meteorol. Soc., vol. 72, pp. 339-347, 1991.

[2] M.J. McPhaden, "TOGA-TAO and the 1991-1993 El Nino-Southern Oscillation event," Oceanography, vol. 6 (2), pp. 36-44, 1993.

[3] M.J. McPhaden, H.B. Milburn, A.I. Nakamura, and A.J. Shepherd, "PROTEUS Profile telemetry of upper ocean currents," Proceedings of MTS 90 Conference, Washington, D.C., 26-28 September 1990, Marine Technology Society, Washington, D.C., pp. 353-357.

[4] N.N. Soreide, D.C. McClurg, W.H. Zhu, M.J. McPhaden, L.J. Mangum, and D.W. Denbo, "Distribution and Display of TOGA-TAO Buoy Data," Proceedings of Oceans 94, this issue.

[5] D. Halpern, H.P. Freitag, and A. Shepherd, "Applications of Argos and RAMS measurements in equatorial Pacific ocean-atmosphere interaction studies." Proceedings of Oceans, September 10-12, 1984.

[6] D. Halpern, "Comparison of upper ocean VACM and VMCM observations in the equatorial Pacific," J. Atmos. Ocean. Tech., vol. 4(1), pp. 84-93, 1987.

[7] M.E. McCarty and M.J. McPhaden, "Mean Seasonal Cycles and Interannual variations at 0, 165E during 1986-1992," NOAA Tech. Memo., ERL PMEL-98, 64 pp., March, 1993.

[8] P.E. Plimpton, H.P. Freitag, and M.J. McPhaden, "Correcting moored ADCP data for fish-bias errors at 0, 110W and 0, 140W," NOAA Tech. Report, 49 pp., 1995.

[9] H.P. Freitag, F. Yue, L.J. Mangum, M.J. McPhaden, J. Neander, L.D. Stratton, "Calibration procedures and accuracy estimates of TOGA-TAO temperature, relative humidity and radiation measurements," NOAA Tech. Report, 32 pp., December, 1994