When we think about the atmosphere, we often overlook the fact that the steam It is the true engine of the hydrological cycle and one of the pillars of the planet's energy balance. It not only governs cloud formation and precipitation, but it is also the greenhouse natural gas more influential. In recent years, the scientific community has refined tools capable of tracking it with a resolution unthinkable decades ago, and GNSS has become one of the most versatile.
Signals from navigation satellites pass through the troposphere before reaching the receiver, and during this journey, they experience a delay that reveals valuable information about atmospheric humidity. This delay, when accurately modeled, allows for the derivation of products such as the Zenith Tropospheric Delay and the Precipitable Water VaporThese technologies are now integrated into numerical weather prediction, climate research, and even hydrological field applications. Best of all, their coverage is continuous, global, and updated every minute.
What does GNSS contribute to atmospheric humidity?
When a GNSS signal crosses the lower approximately 15 km of the atmosphere, it encounters a variable mixture of water vapor, temperature and pressure which slows it down and curves it slightly. For convenience, the total delay is separated into two terms: the hydrostatic, which is fairly stable and linked to pressure, and the wet, which is much more erratic and governed by moisture content. This decomposition is the starting point for extracting quantitative information about the vapor column.
Modern processing strategies employ mapping functions to project line-of-sight delays to their zenith equivalent, while random walk-type stochastic models control for the variability of the wet component and the horizontal gradientsProperly adjusting these constraints is key to capturing both smooth and significant changes. sudden spikes in humidityparticularly during intense convection.
Once the zenith delays are obtained, they become estimates of precipitable water vapor which are assimilated into numerical weather prediction models. This information provides signals about the time and location of precipitation and improves the detection of storms, tropical cyclones and wind gustsWith dense grids of stations, techniques similar to tomography can also be applied to reconstruct three-dimensional moisture fields and study convective processes with great detail.
Beyond the operational aspects, continuously measured tropospheric delays provide long series that capture gradual changes in humidityThese are extremely high temporal resolution records that complement radiosondes and satellites, and which the IPCC considers essential for understanding the variability and change of climate on a regional and global scale.
From ZTD to PWV: critical models and parameters
To transform the zenith retardation into a quantity directly interpretable as water vapor, the hydrostatic term must be separated from the wet term and the conversion factor, which depends on the average temperature of the atmospheric columnThis parameter, known as Tm, is critical in the chain and is estimated with empirical models.
Several proposals have been developed and validated. A study in west-central Argentina compared three widely used Tm models: Bevis, Mendes and YaoAnalyzing in the latter two groups of coefficients adapted to southern latitude sectors. The result showed that Mendes and Bevis better represent the spatiotemporal variability Tm in that region, while Yao, with specific coefficients, offers value in specific areas but does not generalize as well.
The calculation chain also includes the relationship between the wet delay and water vapor, where refractive coefficients are involved. The classical coefficients were compared of Thayer and Rüeger and it was concluded that their differences are small, so in practice either can be used without significant impact on the derived PWV.
Geodetic processing and rigorous validation
A key element for GNSS to deliver reliable humidity is the precise processingA strategy based on Double Phase Differences with Bernese 5.2 was applied to a network extending from Vigo to Brest. This configuration included nine main stations, reinforced with eight more to optimize the network geometry and the robustness of the solutions.
The quality of the tropospheric product was compared with the reference results of EPN REPRO2 in 13 common stations. The agreement was very high, with a mean square error around 3 mm in the zenith delays. From these, the precipitable water vapor was calculated based on the GPT3 modelcovering four full years of data and ensuring consistency throughout the period.
Independent validation of the water vapor series was carried out using radiosondes near the GNSS stations of A Coruña and SantanderThe result was once again remarkable: differences with maximum mean square error values of 3 mm, in line with international standards and consistent with other work comparing GNSS and radiosonde.
Spatial, seasonal, and daily patterns
The derived series allowed us to characterize the spatial variation of water vapor, with a clear decrease observed with increasing latitude. In temporal terms, the annual component dominated over the semi-annual component, with a marked seasonality: the maximums are concentrated in summer and the minimums in winter.
On an intraday scale, the daily anomalies They show common features across seasons, with low values ​​at night and a peak that usually appears in the afternoon. This diurnal undulation is more intense and of greater amplitude in summer, and attenuates in winter, consistent with the dynamics of convection and the availability of moisture.
The joint analysis with local meteorological variables revealed a strong correlation between temperature and water vapor, something expected from the thermodynamic link. However, no direct relationship was detected between water vapor and the recorded rainfallThis suggests that the microphysics and dynamics of each episode play a decisive role beyond the integrated moisture content.
These series were used to evaluate an index of Precipitation EfficiencyFinding low values ​​and less effective precipitation mechanisms in summer compared to winter, despite high vapor levels in the warm season. This result suggests less efficient convective processes or drier environments in middle layers during the summer season.
Pre-rain signs and windows of opportunity
tracking nine episodes of rain Data collected at different seasons allowed for the identification of a recurring pattern: water vapor tends to increase significantly in the hours leading up to precipitation and drops sharply after the event. This behavior was parameterized in quantitative indicators that facilitate its operational use.
The window with the most relevant information was concentrated in the previous 12 hours At the beginning of the rain, where the increase in vapor provides useful clues for immediate prediction. Furthermore, the strength of this signal showed a marked seasonal component, with summer events generally more expressive than winter ones.
Results in extensive networks across America
To fill gaps in South America, where there were hardly any determinations, a network of 136 GNSS stations distributed from Southern California to Antarctica. The period considered spanned seven continuous years, from 2007 to 2013, with zenith lag estimates every 30 minutes, following the most recent IERS recommendations to ensure geodetic consistency.
The delays were compared to operational products of IGS and with results from the second global reprocessing. Compatibility was complete: the mean value of the differences at any station remained the same. below 5 mmThe largest discrepancy, of 5 mm, was observed compared to operational products in high latitudes, consistent with the additional modeling challenges in those regions.
In 15 locations, total water vapor derived from GNSS was contrasted with radiosondes, reaching absolute mean difference values less than 0,7 mm and standard deviations less than 3 mm. As with other authors, a slight dry bias in Vaisala radiosondes with respect to GNSS estimates, an important nuance for data fusions.
The performance of the estimated zenith delay was analyzed model GPT2w in blind mode. The mean absolute difference was less than 3 cm at any location. The model accurately represents the mean value and the annual and semi-annual variationsHowever, it does not accurately capture any atmospheric state, showing patterns dependent on local weather. A linear variation of the mean delay value with height was also verified by weather type, modulated more by the effect of altitude due to the climate regime itself.
Finally, the calculations were performed water vapor trends explicitly incorporating the autocovariance function to obtain realistic errors. During 2007 to 2013, a consistent regional pattern was observed: tropical zones tended to become more humid and temperate zones to dry up, a climate signal with direct implications for water resources and extreme weather events.
The multi-constellation era and high frequency
The availability of several constellations, including GPS, GLONASS, Galileo, BeiDou, QZSS and IRNSSIt has improved spatial coverage and sampling rate, refining the resolution of atmospheric estimates. This added geometry allows for the detection of local phenomena, such as convective cells or sea breeze fronts, which could go unnoticed with conventional instrumentation.
In operation, the flows of almost real time They prioritize punctuality and reliability: data is collected from regional or global networks in minutes, and solutions are generated every 30 or 60 minutes, with delay estimates in five- to fifteen-minute windows. In rapidly evolving scenarios, such as air navigation or storm tracking, atmospheric products derived from GNSS can be updated. in real time, from seconds to minutes, to feed the immediate prediction.
The GGE group has demonstrated established capability in processing GNSS observations to recover atmospheric water vapor for operational assimilation purposes, studies of severe phenomena, and long-term monitoring. This combination of operation and consistency allows it to serve both tactical forecasts and climate records homogeneous.
GNSS and hydrology: much more than humidity
GNSS doesn't just measure steam. Its millimeter-level accuracy in distance makes it possible to detect subtle deformations of the surface linked to water. Thus, subsidence due to aquifer extraction or ground rise associated with glacial meltingThese vertical displacements provide information on mass balances and geotechnical risks.
The reflections of the GNSS signal off water or snow surfaces, known as GNSS-RThey provide estimates of soil moisture, sea level, snow depth, and lake volume. This aspect broadens the hydrological scope, connecting the atmosphere and surface with a large-scale deployable sensor and contained cost.
Among the benefits for hydrology, the following stand out: three vectorsA more detailed understanding of the water cycle, better resource management, and greater sustainability. In summary, here are some practical contributions:
- KnowledgeHigh-resolution spatiotemporal data that complement stations, radars and satellites, useful from the local to the global level.
- Managerial AccountingNear real-time monitoring and evaluation of droughts, floods and erosion, supporting decisions on water infrastructure.
- Sustainability: support for climate adaptation, efficient water use and environmental education with objective and continuous indicators.
Signal, modeling and standardization challenges
Signal availability and quality may be reduced by terrain, vegetation, adverse weather, or interferencesAugmentation services like EGNOS in Europe improve the integrity and performance of GPS and Galileo, and are a useful tool for applications time-sensitive.
Interoperability is another challenge: each GNSS system offers its own frequencies and features. Taking full advantage of them requires receivers and software capable of using them. multiple constellations and signage. Accessibility also matters; tailored solutions, shared services, and incentives can lower barriers to entry for small and medium-sized users.
In tropospheric modeling, the stochastic models Random walk-type measurements do not always reflect true variability. They can underestimate sudden humidity spikes or overestimate variation in stable situations. This is being investigated in adaptive constraints that are adjusted with near real-time weather indicators to better capture steep gradients.
To build reliable climate records, consistency in reference frames, orbit and bias products, and processing strategy is vital. Changes in these elements can introduce artificial ruptures in the series. Careful homogenization, as with other climatological data, avoids erroneous interpretations and supports trend analyses.
Good practices and integration opportunities
Methodologically, it is advisable to combine robust mapping functions with explicit estimation of gradients and careful selection of stochastic constraints. Multiconstellation fusion and network densification open the door to tomography-like 3D reconstructions, especially valuable in convective storms.
Integration with radiosondes and remote sensing satellites provides clear synergies: the radiosonde provides vertical profiles, the satellites map large areas, and the GNSS fill The temporal gap with continuous observations. Joint assimilation in NWP models improves the representation of humidity, which is a decisive ingredient in the prediction of local precipitation and wind.
Operationally, cadences of 30 to 60 minutes with windows of 5 to 15 minutes are a balanced standard for NRT, while flows in seconds to minutes serve the immediate predictionFor climate, the priority is homogeneity over years or decades, minimizing changes in hardware and software or documenting them for later corrections.
The evidence accumulated in Europe and America, from regional networks such as the Vigo to Brest even across continental areas, it demonstrates that GNSS delivers millimeter-level accuracy in delays and millimeter-level accuracy in water vapor when validated against EPN, IGS and radiosondesFurthermore, it detects precursor signals of rain, characterizes seasonality, and quantifies climate trends consistent with tropical and temperate regimes.
GNSS has become established as a tool that, with a single infrastructure, simultaneously serves the real time navigation long-term climate change monitoring. Its strengths grow with the multi-constellation era, and its value multiplies when combined with complementary models and observations to improve early warnings. severe weather, optimize water resources and better understand how water moves on our planet.
