AIAA Conference Preprint

Cecilia M.I.R. Girz* and A.E. MacDonald, NOAA Forecast Systems Laboratory, Boulder, CO
Robert L. Anderson*, Basic Automation, Boulder, CO
Tim Lachenmeier*, GSSL, Inc., Hillsboro, OR
Fernando Caracena and Randall Collander, NOAA Forecast Systems Laboratory, Boulder, CO

The Global Air-ocean IN-situ System (GAINS) is a program to develop the systems needed for an operational global, in-situ observing system. The operational program, intended to begin in 2006, is conceived as a network of 400 high-tech balloons, evenly distributed over the globe. Floating between 60,000 and 75,000 ft for one year with an initial payload of 500 lbs., these superpressure balloons are the vehicles for dropping sondes from the lower stratosphere to monitor Earth’s meteorological, oceanic, and atmospheric chemistry environment. The observing capabilities of GAINS support the mission of the National Atmospheric and Oceanic Administration to describe and predict change in Earth’s environment. This paper gives an overview of the GAINS program, discussing the GAINS concept, motivation for this observing system, the seven-year development program, status of the program, and objectives for Fiscal Year 1999.

In 1995 the NOAA’ Forecast Systems Laboratory began work on a system designed to increase the number of operational soundings taken over ocean areas. The goal of this program was to develop a "shear-directed" balloon to take soundings in the troposphere¹, and to use these data to improve U.S. weather forecasts. Recent field experiments have shown that meteorological soundings in the data-void Pacific Ocean are of critical importance to an accurate 24- to 48-hour forecast in the western United States (and a 5-day forecast for the East Coast). Results from the NORPEX ‘98² and Winter Storms ‘99 field experiments attest to the value of high-resolution, targeted sounding data taken in upper jets and frontal systems heading toward the West Coast. The tropospheric sounding balloon was meant to fill that need.

The initial concept of a tropospheric balloon ran up against insurmountable difficulties related to engineering limitations and aviation safety issues. The balloon itself was the sounding device, taking meteoro

logical data as it ascended and descended two times per day between 150 ft and 35,000 ft. However, pump requirements of high flow rates coupled with low weight and low power requirements ruled out almost all pumps on the market. And those that were available severely restricted possible soundings to one per day, at most. Moreover, the safety hazard these vertically ascending and descending balloons posed to general and commercial aviation, even over the less-traveled ocean regions, was too great a risk. However, the viable aspects of this effort have been recast as a stratospheric platform for environmental monitoring in GAINS.

The Global Air-Ocean IN-situ System (GAINS) is a new global observing system to augment current environmental observing and monitoring networks. GAINS is a network of superpressure balloons designed to float in the stratosphere for up to one year, carrying onboard sensors and hundreds of dropsondes to acquire meteorological, air chemistry, and climate data in remote regions of the globe.

The GAINS balloons fill a void in the current global observing systems. Radiosondes and surface stations give detailed and accurate in-situ readings over land areas, while coverage of oceans and polar regions is quite poor. Remote sensing from satellites provides good coverage of the entire globe, but remote sensors are indirect measurements and represent averages over large volumes having less detail than in-situ measurements. By providing high accuracy sondes distributed over the global atmosphere, GAINS could make existing satellite observations more useful by "anchoring" them to known values. More formally, a Kalman filter analysis³ shows that the best observations over extended areas come when very accurate point observations are combined with horizontal and vertical gradient information. The satellite and balloon-based systems would represent a truly complementary global observing system, in which each component can sense the environment in ways the other cannot, and whose combination offer much of the detailed knowledge of oceans and atmosphere needed for future science.

The GAINS balloons should be thought of as a platform, rather than as a specific set of sensors. As a platform, they can be used in a variety of ways. While GAINS has much to offer NOAA’s operational services that are oriented toward real-time use of data for shorter-term weather forecasts or longer-term climate assessments, the GAINS platform could have a significant role to play in developing and testing new environmental sensors, and in acquiring data sets otherwise difficult to collect. As an example, recent work indicates that very high accuracy may be achieved from carbon dioxide sondes⁴. This suggests that an experiment could be designed in which several hundred carbon dioxide sondes are released over the globe to determine a detailed global carbon budget, including sources and sinks. Published results⁵ based on limited data suggest it well be important to understand the carbon budget much better than we can with the current monitoring system.

The sampling strategy envisioned is a global network of hundreds of balloons that are 120 ft in diameter, carry a payload of 780 lbs.and stay up for a year floating between 60,000 and 75,000 ft. A typical budget of sondes could include 900 meteorological sondes, and a number of other types of sondes such as chemistry sondes and ocean sondes. Chemistry sondes might include carbon dioxide sondes with accuracy better than 1 part per million, as well as ozone and particulate measuring sondes. It may also be possible to drop XBTs (expendable bathythermographs) to measure temperature, salinity and current in the upper parts of the ocean. Each of these sondes presents development and engineering challenges, some of which are substantial.

Dispersal of hundreds of balloons globally appears to be possible, as discussed below. A network of 100 evenly spaced balloons, for example, means one balloon per a 20-degree rectangle of latitude and longitude. Coverage of a 10-degree rectangle of latitude and longitude requires a 400-balloon network. The success of keeping a network of operational balloons evenly distributed will depend on a number of factors, some of which cannot be known until the operational system is implemented. The altitude diversity available, the speed at which the balloons can be repositioned in the vertical, the accuracy of the analysis and prediction models, and operational requirements will all determine the distribution function. There is a nonlinear effect–the more balloons there are in the stratosphere, the better the knowledge of winds in the stratosphere. Thus, the precision in routing the balloon ensemble should increase in a nonlinear fashion as the number of balloons increases.

Of primary importance to GAINS is the safety of this system to life and property on the ground as well as in the air. We have built safety considerations into this program from the beginning. For aviation safety, for instance, the only real collision hazard to aviation are objects whose existence and position are unknown. In the case of the GAINS network, the Global Positioning System (GPS) should be able to locate each balloon at all times within a few meters. This information can be fed through satellite communications into air traffic control data, and ultimately into the cockpit itself, with the aircraft being able to adjust will in advance of potential close passes. Additionally, commercial air traffic is concentrated at levels below 45,000 ft, while the GAINS balloons are designed to operate above controlled domestic airspace. For those times when a balloon is traversing controlled airspace, on launch and termination for example, the onboard transponder is air traffic control’s direct link regarding the balloon’s position. For increased safety, it is desirable to have launch and recovery operations in sparsely inhabited regions, such as deserts, where air traffic and human activities are minimal.

The design of the sondes is also a safety issue. Frangibility, the ability of an object to break into small pieces when it collides with another object, is critical. A development target is a sonde whose total mass is approximately 150 grams, with effects (in the unlikely event of a collision with an aircraft) similar to that of a small bird. The so-called "goose law" of the FAA’s FAR101 prescribes that aircraft should be built to withstand collision with an object (or ingestion into an engine) of about the weight and density of a large bird, namely, 2 kg over a 1000 cubic centimeter volume. Chemistry, particulate, and ocean sondes would also have to conform to the frangibility requirements, and operations would have to be limited to controlled conditions for the nonconforming instruments.

While the GAINS balloon advects with the speed and direction of the wind at its float altitude, the key to keeping these balloons dispersed globally is "shear direction." Taking advantage of the change of wind speed or wind direction with height, a balloon can change its vector motion by changing altitude which is effected by a change in density of the balloon. The GAINS superpressure balloon (Fig. 1) of the canniballoon or cocoon design⁶ is composed of two concentric polyurethane cells (containing helium and air) enclosed in a Spectra^TM envelope, which keeps the balloon at a constant volume. As air is pumped into the air cell, the density of the balloon increases and the balloon loses altitude; conversely, a release of air causes the balloon to rise.

Small changes in altitude produce wide divergence in the balloon’s trajectory. Simulations of four balloon flights (Fig. 2) illustrate this point. Each balloon is "launched" from Tillamook, OR, with identical ascent rates. Upon reaching a specified altitude, ranging from 55,000 to 70,000 ft in 5,000 ft increments, the balloon advects with the wind at that level for 48 hours, and then terminates with identical descent rates. Landing locations vary by a thousand miles from the central valley of California to the Colorado Rocky Mountains. (See below for details on the sounding data used to produce these plots.)

Although not a new idea, scientific ballooning has experienced a resurgence in recent years. New materials, such as Spectra^TM, make superpressure balloons more robust over their metallized Mylar cousins of a few decades ago. Miniaturization of electronics, the development of GPS, and improvements in solar power are advantageous technologies in developing GAINS. A brief overview of the GAINS vehicle and instrument systems follows. Two companion papers discuss the GAINS vehicle⁶ and instruments⁷ in further detail.

The GAINS vehicle is a sewn balloon made from Spectra^TM fabric. Concentric air and helium cells of polyurethane control ballast and lift. The full-scale balloon is a 120-ft diameter sphere that can carry a payload of 780 lbs. Maximum altitude of this fully loaded balloon is 75,000 ft. Two smaller prototypes (PII and PIII) are tested this year. The PII vehicle is a 16-ft diameter balloon, carrying a 30-lb. payload to 23,000 ft. Two 1- to 3-h-long flights are planned. The quarter-scale prototype, PIII, will be tested in early summer 1999 for 1- to 2-day flights. The PIII balloon is 60 ft in diameter, carries a payload of 200 lbs., and reaches an altitude of 60,000 ft.

GAINS is not the first time a cocoon has been used; the prototypes draw on several predecessors. A 25-ft cocoon balloon hangs in the lobby of the Oregon Museum of Science and Industry. And a 3.5-ft-diameter Spectra^TM superpressure balloon flew in 1997 during ACE-2, the second Aerosol Characterization Experiment, taking data in the boundary layer⁸.

The GAINS instruments consist of observation, location, command and control, communications, termination, power, vertical control, and safety systems that are essential for real-time tracking of the balloon and real-time data acquisition, for coordination with FAA Air Traffic Control staff during launch and termination, and for recovery of the balloon after flight termination. Internal and external state sensors on the balloon and a complement of dropsondes acquire environmental data. GPS provides 3-dimensional location of the balloon within several meters. An onboard microprocessor controls the balloon’s functions, and collects data for transmission through line-of-sight and over-the-horizon communications channels. Redundant termination devices are activated by line-of-sight and over-the-horizon command to bring the balloon down on schedule or under emergency conditions. Solar cells with rechargeable batteries, plus emergency back-up batteries power the payload. Pumps, vents and valves add air and release air or helium for altitude control. A safety and recovery system (aircraft transponder, radio beacons and visual lights and streamers) provides for safe operation in controlled airspace and recovery on the ground. The payload is housed in a fiberglass torus closely coupled to the balloon to minimize oscillatory motion during ascent and descent.

Three test flights were conducted in 1998. The GAINS instruments operated in the stratosphere on two flights using zero-pressure balloons. These flights were performed in collaboration with New Mexico State University Physical Science Laboratory. The purpose of these flights was to qualify the GAINS instrument in the stratosphere during an 7-h daytime and a 12-h nighttime test. Successes achieved in these tests included:

The vehicle for the third flight was the 16-ft-diameter GAINS Prototype II (PII) balloon. In this test the balloon travelled 155 miles in four hours at an altitude of 29,000 ft. New systems tested were:

Intelligent use of the GAINS balloon requires a strong modeling component for the prediction of balloon trajectories, which have been computed to simulate control of an operational network of balloons as well as for long-term flight planning and for launch control. As discussed below, each of these applications relies on a unique dataset.

In addition to the engineering challenges of testing the balloon vehicle and its instrumentation for use in the extreme environment of the stratosphere, the second great challenge for GAINS is to design a network and develop a flight management system for 400 balloons. The key to filling in the large gaps in the current sounding network is to control the position and distribution of the stratospheric balloons to maximize atmospheric sampling.

The balloon management problem is a complex one because it requires the use of real-time observations and forecast information about the global atmosphere. The stratosphere itself is not sampled as well as the troposphere; radio contact is lost on many soundings before they reach these levels. Normally, aircraft do not fly as high as the stratosphere, and satellite winds derived from cloud drift by necessity characterize the troposphere, the location of clouds. Hence, stratospheric winds are not as well sampled, nor as well predicted as winds in the troposphere.

Preliminary balloon management studies have been aimed at defining the problem in a tractable form, and in developing low-level control algorithms (for single balloons) to be used by higher-order control structures (for balloon networks). Several balloon directing strategies are described. These results represent an initial set of numerical experiments whose goal is an automatic control algorithm for optimizing balloon spacing. The zeroeth-level algorithm is to select a float altitude and allow the balloon to drift with local winds at this preselected level. The next level of control is to change the balloon drift altitude periodically to take advantage of the change in winds with height in order to bring the balloon toward a more favorable location.

In these studies a balloon's trajectory is computed for a year using the 1997 National Centers for Environmental Prediction (NCEP) reanalysis data to simulate the effects of stratospheric winds. Reanalysis data are available every six hours at 17 mandatory sounding levels on a global 2.5 by 2.5 degree grid in longitude and latitude. The 6-h kinematic data are interpolated linearly to the particular temporal and spatial location of the balloon to simulate balloon drift. The release point for all balloon simulations is Tillamook, OR (45.42N 123.813 W), and the integration time step to compute the balloon's trajectory is 0.1 h.

For this zeroeth-level test, the balloon was allowed to ascend to a predetermined pressure level and drift passively with the wind for a fixed period. As can be seen (Fig.3), uncorrected balloon drift at a fixed level for a full year is too erratic to be of use. This is not a good strategy because the trajectory takes the balloon over most of the land areas of the northern hemisphere poleward of 15N. Some of the overflights are likely to be politically sensitive, and at times balloons may cluster tightly over a small area, at other times they may scatter but be located near conventional sounding sites.

Two sets of control algorithms are developed based on the concept of maintaining the balloon on a designated latitude circle by adjusting its altitude. Restoring a balloon to a preselected latitude is achieved by selecting an altitude that has favorable winds, based on either instantaneous meridional winds, or a persistence forecast for the trajectory. For both algorithms, a 12-h period (the interval between synoptic soundings taken around the globe by participating weather services) has been chosen as the time of drift, and a change in altitude can be made at 0000 and 1200 UTC.

The first option corresponds to the management of balloon trajectories using only analyses of observational data and no numerical weather prediction (NWP) model output. A correction in flight level is introduced instantaneously by selecting the value of the meridional component of the wind that would quickly restore the flight track to the desired latitudinal circle. This simple steering model is able to keep the balloon on track for most of the year except for a few deep circulations that the balloon encounters during winter, throwing it well off course (Fig. 4). (In a follow-on study, a more sophisticated ramp-up time will be introduced for flight level changes, but the instantaneous change used here provides a good first approximation to the problem.)

In the second option, the balloon trajectory for 12 h is "forecast" for each flight level using persistence of analyzed winds for 12h. The level where the trajectory takes the balloon closest to the specified latitudinal circle is then selected for the flight level for the next 12 h. The reanalysis winds interpolated to specified flight times on that level are taken as the true winds for the next 12 h period. The persistence forecast for a drift latitude of 20N is shown in Fig. 3.

The above models were used in several numerical experiments for various specified latitudinal orbits integrating the balloon trajectory for a full year. Uncontrolled drift at a fixed level (Fig. 3) does not give suitable results. Using the instantaneous meridional component of the wind at decision time to select the best flight level for the next 12 h gives a measure of control over the orbit (Fig. 4). The best results, however, are obtained by using a 12-h persistence forecast (Fig. 5) to find the flight level that takes the balloon closest to the selected latitude within the following 12-h period. Note that because of the effects of extratropical cyclones, the best control of drift latitude is obtained nearest the equator; control deteriorates for latitudes approaching the poles (Fig. 6).

Knowledge of climatologically favorable periods for balloon launch is essential for efficient planning of field operations. Of interest to GAINS is the determination of favorable launch periods by week. A seven-year (1990 through 1996) dataset of North American soundings⁹ is the basis for these predictions, from which an extensive database of average soundings has been computed. Weekly averages are available for individual sounding sites at the two synoptic times. The original data were interpolated to 10-mb intervals, and averages were computed. From this information, the potential for flight operations is assessed with plots such as those in Fig. 2. A simple trajectory model uses assumed ascent and descent rates, float altitude, time at altitude and the average winds to predict the balloon’s path. In the current software, winds from a single sounding valid at the time of launch are used to drift the balloon for the entire period of flight. The next version will update the sounding every 12 h and use the closest sounding station. These changes will be especially important for longer flights.

Trajectory planning is also an important part of preflight operations. Our current trajectory model is similar to that discussed in the previous section. It takes into account ascent and descent rates, float altitude, and time at float altitude. Winds, however, are the current data, acquired from either the closest operational sounding or a forecast pseudo-sounding based on an operational mesoscale model, the RUC-2¹⁰. Pseudo-soundings from RUC-2 are very useful for launch and descent, because these forecasts provide high-resolution guidance, with forecasts produced on a 40-km (24-mile) grid every 3 h; the update cycle will soon become hourly. RUC-2 output is of limited applicability, however, for stratospheric flights because the top levels of the model’s isentropic vertical coordinate rarely extend above the tropopause.

FSL’s next-generation trajectory model will use a combination of single sounding observations and operational forecasts from the operational Eta model¹¹. Although Eta spans a larger domain, and includes vertical levels up to 30 mb (~80,000 ft), its six-hourly forecast products are not as frequent as those of RUC-2. More refined and potentially more accurate trajectories, particularly for recovery points on multi-day flights, are expected with the addition of Eta forecast pseudo-soundings to the model.

GAINS is a multi-year program embodying a variety of scientific, engineering, economic, and political challenges. A number of critical capabilities of GAINS need to be demonstrated and the important questions need to be resolved before the system can be considered a credible system for operations. Among these issues are balloon performance, feasibility and accuracy of balloon positioning, lifetimes and reliability of instruments in an extreme environment, launch and control schedules for hundreds of balloons, and the scientific value of GAINS observations. A seven-year research and development program (Table 1) is a staged process to test the existing vehicle, sensor, and instrument components of GAINS, build and test rudimentary launch and control systems, promote development of new environmental sensors and the tailoring of current sensors to GAINS specifications, demonstrate a prototype operational network of balloons, and assess the impact of this new observational platform

Since the GAINS balloons drift with the local winds, they cover the entire globe, and regularly occupy the sovereign airspace of every country. GAINS, however, addresses questions of paramount importance to the planet, and it is clear that an operational program must be sanctioned and administered by an international body. Parallels with extant international agreements under the World Meteorological Organization (WMO) for the acquisition and exchange of weather data come to mind. But the quasi-permanent nature of the GAINS balloons and the increased complexity of its instrumentation, move GAINS well beyond the category of the meteorological sonde. GAINS could be directly under the auspices of one of the agencies of the United Nations, such as the WMO, or a new organizational structure specifically identified with global environmental issues.

Year	Development Phase	No. of Balloon Duration
1995	Feasibility Tests PI, PII, PIII	1 balloon flying hrs.
2000	Stage 1	3 balloons flying mos.
2002	Stage 2	6 balloons flying 6 mos.
2004	Demo & Scientific Assessment	30 balloons flying1 year
2006	Phase 1 Global Ops	100 bal’n/1 yr
2008	Phase 2 Global Ops	400 bal’n/1 yr

In FY99, GAINS has one objective–to demonstrate multiday flight capability of the PIII balloon and instruments. This flight will demonstrate prototype and rudimentary versions of all systems, including ground control systems and operations. Recovery and reuse of the vehicle and instruments is built into the flight plan. Several preflight tests in the laboratory, on the ground and in the air have been or will be executed beforehand. Results of this flight will be presented at the conference.

GAINS is an exciting opportunity for augmenting environmental observations in the 21^st century. Our efforts over the past two years have diminished neither the attraction of GAIN’s potential role in answering important environmental and scientific questions, nor our assessment of the feasibility of such a system. Successes in the engineering and development of the GAINS vehicle and the GAINS payload have continued, and test flights have resulted in positive advances. Trajectory modeling results point to the high possibility of operating a reasonably distributed network of balloons.

11DiMego, G., M. Baldwin, T. Black, F. Chen, F. Mesinger, K. Mitchell, D. Parrish, E. Rogers, and Q. Zhao. "Enhancement to the Operational ‘Early’ Eta Analysis and Forecast System at the National Centers for Environmental Prediction." AMS 16th Conference on Weather Analysis and Forecasting, Phoenix, AZ, pp245-246, 1998.

Fig. 2. Divergent trajectories for a 48-h flight simulation. Each balloon is launched from Tillamook, OR, rises at 1000fpm, and floats at altitude for 48 h advected by the observed wind at one of four specified levels (55, 60, 65, and 70 Kft). Winds are 7-year averages based on the 1200 UTC soundings from Salem, OR.

Fig. 3. A one-year trajectory for a balloon released from Tillamook, OR on 1 January 1997 0000 UTC. The balloon drifts at 70mb for the entire period.

Fig. 4. As in Fig. 3, but flight level is changed at 00 and 12Z by selecting the level having the fastest restoring meridional current to a desired drift latitude, 30N in this example.

Fig. 5. As in Fig. 4, but flight level is changed at 00 and 12Z by selecting the level at which the 12-h persistence forecast generates a corresponding trajectory that most closely approaches the target latitude in 12 h.

Fig. 6. A series of balloon drift latitudes that make use of restoring 12-h persistence trajectories for target latitudes of 10, 20, 30, 40 and 50N as functions of Julian day (1997).