(A) Adaptive Strategies and Projects

Our research tries to compare the cost and accuracy of SUMO, TAMDAR and THORPEX to the current radiosonde network. Each individual adaptation has different instrument specifications and equipment costs. This information is analyzed below.

First, the overall price for the SUMO system was 1200 EUROS (1,584 dollars) (Reduer 2009). Research showed that each airplane averaged one hundred flights; however, the maximum number of flights possible was unclear. SUMO is comparable to the current radiosonde network in several ways, as discovered in the background section. While the longevity of the sensors onboard SUMO model aircraft is unknown, the accuracy specifications are as follows.

Pressure: +/- 0.5 mb
Temperature: +/- 0.5 K (absolute)
Humidity: +/- 1.8%

In general, the accuracy of SUMO is similar to the current radiosonde network. Here, one of the main discrepancies lies with the temperature sensor. The temperature readings have a warm bias during ascent and a cold bias during descent, which alters the measurements. In addition, SUMO is limited by cloud coverage, especially when clouds are low. Besides these limitations, the main problem that we discovered with SUMO comes from the Federal Aviation Administration (FAA). Airspace restrictions limit SUMO to a ceiling height of 1000 meters for ascent (Reuder 2009) which ultimately limits the entire SUMO system altogether in the U.S.

Next, we have the TAMDAR adaptive network. Again it was very difficult to find cost reports. The information that was received from Southwest Airlines Chief Meteorologist, Rick Curtis, indicated that the sensors alone cost around 15,000 dollars, and that an additional 5,000 dollars is needed for installation onto the actually aircraft (the labor). The sensors specifications are indicated below.

Pressure: +/- 3 mb
Temperature: +/- 0.5 degrees C (absolute)
Humidity: +/- 5%

Again, like SUMO, the accuracy of the adaptive strategy seems comparable to the current radiosonde network. The TAMDAR data stream has grown over the years. However, the NWS never purchased the full data stream. We directly communicated with Dr. William Moninger about TAMDAR. He informed us that the NWS purchased a tiny subset, roughly 2-3%, of TAMDAR for an approximate 300,000 dollars per year. Beyond this, TAMDAR details are strictly confidential.

Lastly, was the research activity, THORPEX. This system's monetary figures were determined by personal contact as well. After consulting with Dr. David Parsons from the OU School of Meteorology, we received an average estimated cost of 75,000 dollars for land measurements. As THORPEX is research based, there were several conclusions on how it influences forecasts. Rabier et al (2007) found that mid-latitude targeted observations are twice as effective as random observations. Here it was also concluded that targeting alone is unlikely to significantly improve the accuracy of forecasts one to fourteen days out. While we found various results from THORPEX studies, the accuracy specifications were harder to evaluate than the costs. Due to the complexity of the project, we could not find the actual accuracy values of the pressure, temperature and relative humidity sensors that THORPEX employs. It is interesting that the costs and accuracies of THORPEX were so difficult to find because the researchers claim that the global strategy is a cost-effective one.

Overall, evaluating the cost-effectiveness for SUMO, TAMDAR and THORPEX is difficult because costs are not available to the public and the benefits are hard to judge. Different observations have different impacts and consequently, different accuracies. There is no way to make direct comparisons between SUMO, TAMDAR and THORPEX for this very reason. Therefore, the cost-efficiency for our project is one based off qualitative results instead of quantitative ones. Our project attempts to evaluate a situation that has been problematic for several decades, owing to the fact that the U.S. radiosonde network has not changed in over 50 years. With such an 'old' system in place, there certainly remains areas of improvement. In order to assess which areas of the radiosonde network should or should not be changed in conjunction with the parts that are most beneficial to the NWS and private sector, we analyzed the following survey information.


The private sector survey (Appendix A) responses had many common results. All of the companies said that they directly or indirectly used radiosonde data. Several of these companies used the data after it was processed into another forecasting product, such as a weather map. Also, they all welcomed the idea of having more data and filling the gaps in the network. This would help the companies make better decisions for their customers by supplementing their forecast accuracy. The radiosonde data helped hindcast as well to recreate a past weather events for forensic purposes. A few agreed that network budget cuts and less launches would negatively impact their business and in turn, their customers. Overall, the survey concluded that all the private sector businesses heavily rely on the radiosonde network.

The NWS survey (Appendix B) responses were similar to the private sector ones, but this survey was more detailed and catered to special soundings. The sample size of the survey included 86 NWS forecasters. The breakdown of the representation by NWS regions are shown in Figure 4. Each of the responses indicated how many special soundings were launched on an average year at each station (Figure 5). It was found that the most launches were made in the central region. This could be related to the high number of launches during severe weather.

The survey showed that severe weather was the number one reason to launch the soundings for a majority of the regions, not just the central one. The only region that had another high-impact weather event was in the eastern region due to tropical weather. In general, this result helped us to understand where the need for observations is during certain types of weather events.

The last two topics on the survey requested the thoughts of the NWS WFO employees on the creation of an adaptive network and if they had any suggestions for it. Their main thoughts included:

- It would be beneficial to fill the real-time gaps in the current radiosonde network
- An adaptive network is a good idea, yet there is not a budget to support it
- They want to see asynoptic times and no set locations
- The adaptive network would need to be integrated into their AWIPS/AWIPS II system.
- Only launch radiosondes if they are needed (high-impact weather vs. non high- impact weather)
- Possibly use alternative observation networks (i.e. GPSMet Sites provided by ESL)
- Collect the data as the radiosonde descends to gather more data from one launch

It was very interesting because these main thoughts ran almost parallel to the motivations behind this research project. These results helped guide our conclusions.