Galina Dick, working group leader of GNSS meteorology at GFZ, enables with her work the use of satellite navigation signals in meteorology and climate research. We talk to her about her work and how the deflection of radio signals by water vapour in the atmosphere turned from an interfering signal to the most important measurement system.
You are head of the working group GNSS Meteorology. What is your research about?
GNSS stands for 'Global Navigation Satellite System'. This is the generic term for systems like the American GPS. Other such systems are the European Galileo, the Russian GLONASS, and the Chinese Beidou.
We process radio signals from all these GNSS satellites, i.e. signals that are actually sent out for satellite navigation, for atmospheric research. The signals allow us to make statements about water vapour content in the atmosphere.
GNSS meteorology is still a relatively new field, but has been established for more than ten years now. Recently, for use in climate research, there has been rapid development.
Why are the navigation signals so well suited for determining water vapour content?
There are now large numbers of GNSS receivers installed in many countries around the world, some of them in very remote regions. The navigation signals are therefore available globally with very high resolution, i.e. with a large number of data points per area and time unit, and can be easily combined with other meteorological data. With GNSS, we therefore obtain supra-regional measurement data for the entire atmosphere.
In summer, we can reach up to 50 milimetres of water within an imaginary column from the ground to the upper atmosphere in Central Europe. We can determine this with GNSS with an accuracy of less than one millimetre.
We can measure in all weather conditions, even when the sky is cloudy. A conventional meteorological instrument, such as the water vapour radiometer, for example, does not measure in rain.
What makes the water vapour content so important?
Along with temperature and pressure, water vapor content is THE key parameter for the state of the atmosphere. It is of enormous importance for meteorology and climate research. Water vapor is, for example, the most important natural greenhouse gas.
It is very difficult to measure it in the conventional way and it is highly variable. Temperature and pressure can be determined quite easily with thermometers and manometers. Conventional instruments for determining water vapor, such as radiosondes, i.e. weather balloons, and radiometers, are, however, complicated to use and require extensive calibration. They are also very expensive.
A water vapor radiometer cost more than 100,000 euros. Very good geodetic GNSS receivers are also relatively expensive, at around 15,000 euros, but we also have low-cost receivers in use that cost less than 1,000 euros.
But the fact that water vapor in the atmosphere can be determined via radio signals of navigation satellites must mean that it does deflect navigation signals. Doesn’t that disturb positioning?
Yes, water vapour is an interfering signal for positioning, but this is what makes it usable in meteorology in the first place. We speak of "from error to signal". The measured signals must be subjected to calculations to provide information on either positioning or water vapour content.
You said that GNSS signals have been used in meteorology for over ten years now. Was there an "aha moment" when it was realized that navigation data can also be used to determine water vapor content?
This was already experimentally established in the 1990s. At that time, first scientific articles with proofs of concept were published. GNSS data were compared with those of radiosondes or water vapour radiometers and it was found that similar results were gained.
Then, from 2001 to 2004, there was a major EU-funded 'COST' action in Europe - 'COST stands for 'European Cooperation in Science and Technology' - with the participation of 15 countries. Its aim was to operationalise the method, i.e. to find out how radio signals can be used to determine water vapour in meteorology. As a result, we at GFZ deliver GNSS water vapour values to weather services continously, at hourly intervals. And since 2006 these have been used for weather forecasting.
How established is the use of GNSS signals in meteorology now?
Since 2006 they are used in weather forecasting. After a rather long development phase, also the German Weather Service DWD has now, since 21 April, started to use the data from 900 stations worldwide for daily global forecasts.
Convincing meteorologists to use GNSS data was not easy. At first, they were very sceptical whether navigation data were really better than their proven methods. And, of course, the data here cannot simply be read off a measuring instrument, but must be elaborately calculated to gain water vapour data.
It was very exciting for me, being involved from the beginning, to see how the perception changed "from zero to hero" and how meteorologists today are totally enthusiastic about GNSS water vapor data. This goes so far thatif a radiosonde shows different values than GNSS data, the values of the radiosonde are more likely to be questioned. This is how much trust GNSS data has gained.
Tell us about the new climate reference network GRUAN, which is currently being established with your participation.
GRUAN, for 'Global Climate Observing System Reference Upper Air Network', is a network for certain climate-relevant parameters that are determined globally with highest accuracy and quality. The aim is to create a reference network to whose data other measurements can refer.
The initiative for GRUAN was launched in 2007 by the Global Climate Observing System of the World Meteorological Organization of UNESCO.
The main data are temperature, pressure, and water vapour. The network must provide these three parameters without gaps, and with highest quality. The GNSS data have the highest priority in determining water vapour content.
How is GFZ involved here?
Within GRUAN, there are several working groups, one for each measurement system. Together with Jens Wickert from GFZ, we are actively involved in shaping GNSS measurements. We determine, for example, the accuracy with which GNSS data must be available at a receiving station and how they must be processed to meet the high quality requirements for climate research.
Together with other institutes, we monitor data quality and compare the values again and again with data from radiosondes, radiometers, and weather models. If, for example, maintenance work is carried out at any station or an antenna is replaced, it must of course be documented why there is no data from there for a short time.
Since 2013, GFZ is the official centre for processing GNSS data of GRUAN. Using software developed by us, we calculate the end product, i.e. the water vapour content, from the GNSS data received from all stations. I am responsible for this together with my colleagues. The German Weather Service, as the control centre, collects all data and makes them available to climate researchers.
Our water vapour products were proposed for certification by GRUAN in May 2019. This means that the quality of our data meets the highest accuracy requirements for climate research. In November this year we will probably be officially certified.
Why was GFZ given this "job"?
We have a high level of expertise due to many years of experience and can rely on an infrastructure with many ground stations. Since 2004, we have been active in the Helmholtz research programme "Atmosphere and Climate", in which we use GNSS satellite data for atmospheric sounding, especially in cooperation with the German Weather Service. We are, it must be said, quite unique in the world.
Rain forecasts were very inaccurate, previously, because there was no data as highly accurate as GNSS data available now. Will this be improved with GRUAN?
No, not directly. At least not via new infrastructure. There is only one GRUAN station in Germany in Lindenberg. This does not improve the rain forecast in Germany.
But within GRUAN we use about 300 ground stations in Germany, which were originally installed for the national surveys. This very dense network of stations enables us to determine the spatial distribution of the water vapour content over Germany very quickly. In comparison with the temperature and pressure conditions, the temporal development of precipitation can then also be predicted.
And the data exist globally?
Only in those areas where there are ground-based GNSS receiving stations. So, not over the oceans for example. Most of the ground stations in use were originally installed for land surveying. In Germany, there are about 300 of these stations by the land surveying network of the surveying offices.
In addition to the German receiver network, we are working on more than 350 additional stations distributed all over the world. We will then make the water vapour values available just as quickly as for the German network. They are used by various international weather services to improve their forecasts.
Data from ships equipped with GNSS receiving antennas could also be used, but they do, obviously, not form a permanent measuring network.
Do you have the servers in the basement for processing?
Exactly, we run our own servers and have them maintained by our own IT specialists. Our software is also permanently maintained and developed. Only with this comprehensive control can we guarantee that we deliver the data permanently. We could not do this if we were dependent on external infrastructure.
We also archive raw data on our servers and thus make them available in the long term.
What makes archiving so important after you have provided the data for the weather services?
Archiving the raw data is important because it is possible that processing will be improved in the future. You might then want to apply new calculation methods with a new quality standard to all previous data.
For whom else are the data of interest?
The data become more and more valuable over time because data series become longer and longer. Some receiving stations have now been measuring for 30 years, already. This is a relevant period of time for climate research because long-term trends can be identified.
That is why climate researchers are very interested in our data. Previous climate models are usually based on less accurate data from radiosondes with regard to water vapour. Now, with the new GNSS water vapour measurements, reference data with higher accuracy and quality are available.
With the new data sets it is much easier than before to check the accuracy of the climate prediction models and the derived forecasts.. They do this by having their models "predict" the values of the past 30 years, which we now have available, and then see how well they match actual developments.
Do you continue research in this area?
Yes, we recently completed a four-year European COST action, involving 26 countries: 'GNSS for Severe Weather Events and Climate'. This involved the use of GNSS data for forecasting extreme weather events and for climate research.
GFZ was one of the initiators of the action, I was deputy chair of one of the working groups.
For everything that is initiated in this field, it must be guaranteed that the data can be processed, collected, and, above all, made available for decades. But, of course, we are proud that we as GFZ can support this.
You are also involved in the MOSAiC project, a major Arctic expedition by ship that freezes in pack ice, under the lead of Alfred Wegener Institute?
Yes, we work very closely with AWI within GRUAN, anyway. For radiosonde measurements, especially in the Arctic, there are no good reference-calculations available so far. During the whole expedition, which lasts one year, radiosondes are ascending four times a day. We have installed a GNSS receiver on the ship and compare its measurements with those of the radiosondes.
For me, this is new territory, because the ship is drifting over the North Pole, i.e. it is in permanent, albeit slow, motion. We will certainly have to adapt the software in order to achieve millimetre accuracy of the data. That is our aim.
Interview: Ariane Wiesmann (geb. Kujau)