Paper presented at the 1999 American Meteorological Society Conference
15th International Conference on Interactive Information and Processing Systems (IIPS)



National Weather Service, Office of Systems Operations


Most balloon inflations and flight train preparations for National Weather Service (NWS) Upper Air soundings have been performed in High Bay inflation buildings. These buildings are approximately 6.7 meters high with a 4.3 meters high radome mounted on top. The radome provides protection for the Automatic RadioTheodolite (ART) antenna and receiver. The inflation shelter itself is a 6.7 meter by 9.1 meter building with two rooms. Hydrogen or Helium gas used for inflating the balloons is stored in the smaller of the two rooms, which is approximately 3 by 6.7 meters. The inflation room is an approximately 6.1 meters cube. Two 4.3 meters wide by 5.5 meters high doors are mounted on opposite sides of the inflation room. The inflation shelter is typically located in the center of a 40,470 square meter tract of land that must be kept clear of all obstructions.

The relatively large size of the building and the tract of land are necessary. When inflated, the balloon can be as large as six feet in diameter at its widest point horizontally and close to 10 feet vertically. When carried out by a six foot tall person, the top of the balloon may be 16 feet or more high, and on windy days will by blown from side to side. Contact with a sharp object or abrasive surface can easily burst the balloon. At release, the balloon must be kept well clear and down wind of the High Bay. On windy days it must also be hundreds of feet away from any objects down wind from it. The flight train consists of a balloon, a parachute connected about 5 feet below the balloon and a radiosonde connected about 100 feet below the balloon. The balloon ascends at approximately 5 meters per second, but during high winds can be blown horizontally much faster. In order to prevent the radiosonde from hitting the ground during high winds, a de-reeler is attached immediately below the parachute. The de-reeler slowly pays out the string attached to the radiosonde. Even with the de-reeler, when a balloon is released during high winds, the balloon is sometimes blown hundreds of feet horizontally while the radiosonde is still within 3 meters of the ground.

The facilities and procedures for preparation and launch have been developed over time and are well suited to operational requirements. The NWS would normally be unwilling to consider deviating from these proven methods. However, during the last 10 years, the NWS has been in the process of replacing it=s old and obsolete equipment, through its Modernization Program. The Modernization has also necessitated the restructuring of field operations. Many field offices have moved; some to much more densely populated areas. Of particular concern are some offices that have been moved to University campuses where there is no available open space large enough to support a conventional inflation shelter. At some locations it will be necessary to launch balloons from a roof top.

Roof top releases will be necessary at approximately six locations. There are other sites where limited open space prohibits the use of a conventional shelter. A slightly smaller version of the conventional high bay has been built at some of the new upper air sites. However, this new design costs approximately $150,000.00. If a less expensive alternative were available, it would be used at the small number of sites that require relocation, but have not been moved yet.


In assessing this situation, it was determined that a new type of enclosure was needed. The National Center for Atmospheric Research , in Boulder, Colorado was contracted to develop a new type Balloon Inflation and Launch Shelter (BILS) in conjunction with development of a proof-of-concept NAVAID (a system utilizing navigational aids, such as LORAN or GPS for wind-finding) radiosonde sounding system. NCAR was given three very general objectives for the BILS development:

1. Capable of launching balloons from highly populated areas and from roof top.

2. Capable of semi-automated release.

3. Portable and easy to set-up. (Required for multiple site field testing.)

2.1 The NCAR Design

The design criteria for the BILS established during the initial phase of development was intended to meet the fixed site and transportable sounding system needs of both the NWS and the atmospheric research community. The design also addressed, in a preliminary fashion, a semi-automated balloon launch capability under the control of a system computer.

The BILS was to:

1. Accommodate a variety of balloon sizes.

2. Accommodate a variety of navaid radiosondes.

3. Reliably function in extremes of weather.

4. Be operable by a single individual

5. Be installed in either roof top or ground-based modes.

6. Have a long useful life with minimal maintenance.

7. Be adaptable to semi-automated launches.

8. Have moderate initial cost.

9. Be transportable.

Specifically, the BILS was an approximately 2.4 meter cube structure designed to protect personnel and equipment during the preparation and launch (automatic or manual) of a radiosonde flight train. It was constructed of fiberglass siding placed over 4 cm square aluminum tube welded frame. The roof was coated with a ultraviolet (UV) resistant rubber coating. The BILS was configured so that it could be attached to a 2.4 meter wide by 3.7 meter long transportable equipment shelter (TES), used for housing radiosonde tracking equipment and computers. The TES was not designed for operational use but was to be used as work space for field evaluations if office space was not available. The entire structure is blocked, leveled, and secured with tie-downs.

An approximately 2.4 by 2 meter rooftop hatch of the same general construction was opened for radiosonde release. The normally closed position of the hatch covered a 1.8 meter diameter hole in the roof. Air deflectors, made of UV resistant fiberglass, were mounted around the top edge of three exterior walls. The deflector for the fourth side was part of the hatch cover. During a launch sequence the hatch cover was rotated 90 degrees to open, the balloon was released, and the hatch cover rotated back. During normal operations, the hatch was automatically controlled through the computer workstation or with a hand-held control. Electrical contact switches at either end of the hatch travel cycle both stop the geared electric motor and set contacts to reverse the direction of the motor to insure proper motor rotation and hatch direction. A simple hand lever was also available for manual operation in the event to a power failure. Counterweights were added to balance the hatch cover.

Heating and air conditioning for the TES were provided by commercially available units. To ventilate the radiosonde inside the BILS, a small electric fan was used to draw outside air through a hole in the side of the BILS. Flexible duct work was insulated and attached to this hole. Outside air was directed over the sonde by this ducting. Radiosonde ventilation was normally automatically controlled by the computer.

Helium was contained in high pressure steel cylinders that were equipped with a hand valve for gas release. These cylinders were strapped to the outside wall of the BILS. A pressure regulator was attached to the cylinder to reduce the high pressure (approximately 2000 psi) to an easily managed pressure (approximately 20 psi). From this point on, only a low pressure line to the balloon was needed. This line was either rubber or plastic hose of 1.3 cm diameter for safe handling. Gas flow into the balloon was controlled by a quarter-turn valve.

The balloon holder was made of soft fabric webbing. This webbing held the balloon during inflation and helped keep it from hitting the BILS walls prior to release. To keep the balloon from rising and hold it in place during windy conditions, two wide fabric strips were draped over the top of the balloon and held by two solenoid-controlled clamps until the moment of release. When activated by the computer or manually, a small weight on the end of each strip quickly pulls the strips off and away, freeing the balloon for ascent.

A whip antenna, approximately 2 meters in length, was mounted on the roof to the BILS. The purpose of this antenna was to provide a navaid signal to the navaid receiver on the radiosonde prior to launch. The radiosonde antenna was ineffective before release because it was tightly wound around the radiosonde de-reeler device. A short piece of wire between the whip antenna and the sonde was disconnected at launch, allowing the de-reeler to deploy the sonde antenna. Within a few seconds of launch, the navaid antenna on the sonde was fully deployed and signals were translated to the navaid receiver in the normal fashion.


This proof-of -concept system was completed in 1993. Field testing was conducted in 1994 and 1995. Table 1 shows a list of upper air sites that participated in the field evaluation of the BILS, the dates of the test, and the reason the site was selected.

Table 1: BILS Field Evaluation Sites

Station                     Test Dates                 Reason Site chosen
S. Ste Marie, MI       Jan-Feb, 1994             Snow/cold
Caribou, ME             Jan- Feb, 1994            Snow/cold
Int'l Falls, MN           Feb-Mar, 1994             Snow cold
Glasgow, MT            Feb-Apr, 1994             High winds
Norman, OK             May-Jul, 1994             High temp/thunderstorms
Desert Rock, NV       Jul-Aug, 1994             High temperature
Tallahassee, FL        Aug-Sep, 1994            High temperature
Sterling, VA              Nov '93 - Aug '95         NWS Test and Evaluation Branch

About 20 flights were conducted at each site to evaluate the general performance of the BILS under site-specific weather conditions. A total of 157 flights were made at the BILS evaluation sites. Many of the flights took place during fair weather, but there were a number of releases during periods of light snow or rain. Launches took place in temperatures ranging from -281C to over 401C and humidities ranging from 96% to 11%. Since the BILS was at each field site for only a few weeks, it could not be determined how well the materials used in the construction of BILS hold up to solar radiation and temperature and humidity extremes over long periods of time.

The BILS radiosonde ventilator was a major problem. Although it was intended to force outside air over the radiosonde, it was rarely able to equalize the sonde temperature and humidity with the outside surface conditions. In cold weather, the sonde temperature was generally warmer and the humidity less than what was measured outside. The opposite was generally true during hot weather. One likely cause for the significant differences was from heated or cooled air entering into the BILS from the TES. These differences can cause problems with radiosonde data quality shortly after release. As the sonde leaves the BILS, it enters air at a different temperature/humidity than what was measured inside the structure. Because there are time lags with the radiosonde temperature and humidity sensors, errors in the measurements can arise if the temperature/humidity inside the BILS differs greatly from the outside conditions. For instance, in cold, dry conditions the radiosonde humidity sensor (i.e. hygristor) can take up to 2 minutes to reach ambient conditions.

BILS releases also took place under varying wind speeds. There were 11 flights where winds exceeded 20 knots at the time of release. During most of these flights the balloon flight train successfully left the BILS. However, at Glasgow, 28 knot winds caused the sonde to hit the ground before being carried aloft.

Another objective of the evaluation was to determine if 500 and 600 gram balloons released from the BILS could attain acceptable ascension rates and termination altitudes. Totex rather than Kaysam balloons were mainly used because, at the time they were better able to maintain a spherical shape when released in high winds. This helped reduce the likelihood that they would burst as they left the BILS roof hatch.

In most flights the balloon was filled with 1.7 to 1.98 cubic meters of helium, which is about 1.5 meters in diameter at ground level. Generally, 600 gram balloons carried heaver payloads and were filled with more gas than 500 gram balloons. Total flight train weight ranged from 750 to 1,120 grams. Even with this range of weights, nearly all of the flights had an average ascension rate of more than 275 meters/minute, which is the NWS balloon ascension rate requirement.

It should also be noted that the ascension rates of 18 flights using a 600 gram balloon filled with only 1.7 cubic meters of helium averaged nearly 300 meters/ minute, with only one flight having an ascension rate of 265 meters/minute. Total payload weights for these flights ranged from 850 to 1,050 grams. These results show that 600 gram balloons can achieve desired ascension rates with less helium than used operationally.

Flight termination altitude results were not as good. Only one of the 500 gram balloon flights and 15 of the 600 gram flights reached altitudes equal to or exceeding 10 mb The termination altitudes achieved with the 500 gram balloons were well below what is usually averaged with 600 gram balloons used in the operational radiosonde network. There could be various causes for the low termination altitudes of the 500 gram balloons. Only one could possibility could be attributable to the BILS. The shrouds, which hold the balloon in place inside the BILS, might be scratching the balloon as they are pulled away at release. At high altitudes when the balloon has expanded to five or six times its diameter at launch, even a very small surface scratch can weaken the balloon enough to cause a premature burst. However, we have not been able to verify if this is actually happening.

There were other ease of use and safety issues noted during the field test. The main complaint from users was on the height of the shelter. The inside height was 7 feet. Depending on the size and type of balloon used, it=s inflated diameter could be as much as 62 feet. Only about 6 inches at ground level were left to accomplish much of the preparation for launch. This made the process unnecessarily difficult and uncomfortable.

The gas fill valve and meter were also concerns. The gas fill valve must be closed manually after the balloon is inflated. Determining the amount of inflation gas used requires subtraction of the previous meter reading form the present one. This can result in incorrect fill levels, which might result in early burst or incorrect ascension rates.

Although some significant problems with the original design were found, the BILS design was preferred over the High Bay by most users. After the incorporation of various enhancements the basic design will be applicable for roof top and other special applications.


Future BILS production will incorporate the following improvements.

1. The radiosonde ventilation rate was increased considerably. Follow-on testing at the NWS Test and Evaluation Center demonstrated that a blower with a flow rate of 15 cubic meters per minute would bring the temperature and relative humidity within the BILS to <11C and <5% RH in approximately 3 minutes even when the interior of the BILS was forced to a 101C temperature differential and a 15 % RH differential. This higher flow rate will have the added benefit of providing greater positive pressure within the BILS immediately prior to release which will result in a faster exit of the balloon and flight train from the hatch which will reduce the likelihood of balloon burst during releases in high winds. Additional sensors will be added inside and outside the BILS to accurately measure temperature and RH differentials. The outputs from these sensors will also be used to notify the operator when inside conditions are within tolerance.

2. The size of the inflation area was increased to enable use of properly filled 600 gram balloons which will reduce the likelihood of early flight termination.

3. The overall height of the BILS was increased to accommodate a work surface at a comfortable height so balloon fill and flight train preparation can be accomplished while standing.

4. An air tight door will be added between the inflation area and a storage room to prevent heated air from entering the inflation area immediately prior to launch while still providing a more comfortable work environment for operators.

5. The BILS controller will be modified to enable the use of gas flow meter that can be controlled by the computer. The operator will be able to preset the amount of gas the balloon is to be filled with by volume. When the balloon is filled the flow of gas will be automatically shut off.

6. In order to avoid possible damage to the balloon by the shrouds holding it in place prior to release, in the future, balloons will be held in place at the neck of the balloon. There are various methods of doing this. However, this will require the use of balloons that will hold a spherical shape. Balloons that are made from more elastic material will distend vertically about 50% more than their horizontal diameter. Because of the limited clearance inside the inflation area of the BILS, this will result in the balloon touching the hatch which could cause the balloon to burst.


There are two issues which we have been unable to find resolution to. One involves the effect a roof has on temperature and relative humidity of the air directly above its surface, especially on sunny days. The other is the difficulty in developing a method of automatically adding water to a water activated battery.

The heating of the roof surface by the sun alters the temperature and relative humidity of the air in its immediate vicinity. Even if the ventilation system was working properly, the air that the radiosonde was being exposed to would not be representative of the conditions that should exist if the building were not there. The result will be a contamination of the radiosonde temperature and relative humidity data just off the roof of the building.

There has been no mention of a fully automated flight preparation and launch capability. The NWS would prefer a fully automated launcher. However, there is a major problem in automatically preparing a water activated battery. The amount of water is critical. Too much water will add unnecessary weight and could result in damage to the electronics. Too much or too little water will degrade the performance of the battery. Batteries can not be activated in advance because they only have a two to three hour useful life after the water is added. Finally, techniques related to activating batteries are specific to the vendor proprietary designs. This problem would be much more manageable if radiosonde designs were standardized. Any system the NWS fields must be capable of use with multiple brands of radiosondes. There are no other major problems to developing a fully automated system. Perhaps, in the future some other type of battery may replace the water activated batteries currently in use.


Assessment of the NEXUS System, August, 1995. U.S. Department of Commerce, National Oceanic and Atmospheric Administration, National Weather Service, Office of Systems Operations, pp. 2-19.

shltal.jpg (20111 bytes)BILS-1.bmp (621990 bytes)




















Left photograph - typical NWS inflation shelter.  Right photographs - NCAR Prototype BILS.  Top right photograph shows hatch closed.  Bottom right photograph shows hatch lowered and balloon exiting.



Back to the Upper-air Observations Main Page