One of the major sources of error in short-term forecast of precipitation is the lack of both precise and continuous water vapour data to provide the NWP model Initial Condition (IC). Different techniques can be envisaged, but their spatial and temporal sampling resolutions are not always optimal. An improvement in monitoring the atmospheric water vapour and its assimilation in NWP models would lead to more accurate forecasts of precipitation and severe weather.
In this framework, space borne InSAR appears as a useful tool for high-resolution water vapour retrieval [Pichelli et al., 2015; Mateus et al., 2017]. The interferometric phase is in fact dependent among other things on the changes in the atmospheric path delay. Then, as soon as other sources of phase difference are properly removed (topography, surface displacement, etc) the interferograms provide the Atmospheric Phase Screen which on its turn is almost proportional to the Integrated Precipitable Water Vapour (IPWV).
A geosynchronous SAR, as the one proposed in the last call for Earth Explorer (G-CLASS mission) would strongly push this technique, as it would ensure the high temporal resolution required in the frame of the operational weather forecast. The APS maps can be converted into path delay or IPWV and they can be assimilated into limited area, cloud-resolving Numerical Weather Prediction (NWP) models for a better characterization of the local scale phenomena. This can be done with unprecedent space and time resolution with respect to other sources of path delay or water vapour (e.g., GNSS ground receivers, microwave radiometer, near infrared radiometers, ground meteorological stations). The removal of other contribution to the interferometric phase (including error in the knowledge of the baseline) and the conversion from the differential APS to absolute path delay requires additional information, i.e., available through a Global Navigation Satellite System (GNSS) network.
This work aims at investigating the capability of InSAR to monitor the atmosphere in view of the possible availability of a geosynchronous SAR mission. We compare the total delay (ZTD) predicted by the Weather Research and Forecasting Model (WRF) and retrieved by a network of GNSS stations with the APS derived from a stack of Sentinel-1 data by using the SBAS interferometric processing technique. We have reconsidered methodological approaches available from the literature to retrieve the absolute values of the path delay (Pichelli et al., 2015; Mateus et al., 2017) and introduced some novel solutions. Then we have assessed the method on real data for a couple of case studies in North Italy, were the path delay maps derived from Sentine-1 have been assimilated using the 3D variational assimilation technique implemented in WRF. The prediction results without assimilation or assimilating GNSS Zenith Total Delay or delay maps from Sentinel-1 have been compared to reference data, mainly rain gauge precipitation available on the area, to assess the improvement in the prediction skill brought by the assimilation. The results show that Sentinel-1 definitely improved the forecast still in one case, whereas in the other case the impact of the atmospheric delay assimilation is questionable, although the InSAR derived delay map improves the performances with respect to the assimilation of the GNSS network ZTD retrievals.

Pichelli, E., R. Ferretti, D. Cimini, G. Panegrossi, D. Perissin, N. Pierdicca, F. Rocca, B. Rommen, “InSAR Water Vapor Data Assimilation into Mesoscale Model MM5: Technique and Pilot Study,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 8(8), 3859–3875, 2015
Mateus P., J. Catalão and G. Nico, "Sentinel-1 Interferometric SAR Mapping of Precipitable Water Vapor Over a Country-Spanning Area," in IEEE Transactions on Geoscience and Remote Sensing, vol. 55, no. 5, pp. 2993-2999, May 2017.

Area D - Board 374

On 6 June 2017, the Generic Atmospheric Correction Online Service (GACOS, http://ceg-research.ncl.ac.uk/v2/gacos) was officially released in the ESA’s FRINGE workshop in Helsinki, Finland to assist InSAR researchers to reduce atmospheric water vapour effects on InSAR observations. GACOS has notable features including (i) global coverage, (ii) all-weather, all-time useability, (iii) correction maps available in near real-time, and (iv) indicators to assess the correction performance and feasibility. The model integrates operational high resolution ECMWF data (0.125 degree grid, 137 vertical levels, 6-hour interval) and continuous GPS tropospheric delay estimates (every 5 minutes) using an iterative tropospheric decomposition model. Indicators describing the model’s performance including (i) GPS network and ECMWF cross-RMS, (ii) phase-delay correlations, (iii) ECMWF time differences, and (iv) topography variations, are available to provide quality control for subsequent automatic processing, and provide insights of the confidence level with which the generated atmospheric correction maps may be applied. GACOS’ performance for correcting atmospheric effects on a single pair of SAR interferogram was tested using globally-distributed SAR data, encompassing both flat and mountainous topographies, mid-latitude and near polar regions, monsoon and oceanic climate systems, achieving a phase precision and displacement accuracy of approximately 1 cm for the corrected interferograms even with a coverage (e.g. 250 km x 250 km).

GACOS has become a popular tool in the InSAR community with 14 visitors each day on average, 7 of which are new visitors. Up to now (11 Nov 2018), GACOS has processed over 28k jobs for InSAR researchers, over 80% of which were requeted for Sentinel-1A imagery. Given its near-real-time capability, GACOS has successfuly responded to several recent events including the Maoxian Landslide (China, 24 June 2017), the Xinjiang Earthquake (China, 8 August 2017) and the Agung volcanic eruptions (Indonesia, November 2017 to July 2018) by providing atmospheric correction maps so as to assist in generating near-real-time deformation fields, which have been reported and highlighted by over 20 social medias.

To further improve the performance of GACOS to better serve the InSAR community, a new generation is being developed by considering the following factors: (i) the temporal resolution of ECMWF products, (ii) the potential contributions from other water vapour products (e.g. radiosonde, MODIS and OCLI), and (iii) indicators for the regional performance of GACOS.

Area D - Board 375

Remote sensing techniques, such as GNSS and SAR observations, together with high-resolution modelling of the atmosphere with NWP models can provide a synergic framework for a number of earth observation applications. Firstly, the combination of GNSS observations with active satellite observations (InSAR), processed with advanced differential interferometry methodologies (i.e. PSI and SBAS) is capable of providing precise geodetic measurements, thus greatly enhancing the knowledge of local crustal deformations with multiple benefits for monitoring co-seismic, post-seismic, as well as aseismic discontinuities. Meteorology is an integral part of this monitoring technique, as high-resolution NWP models have the potential to accurately simulate the tropospheric state and remove a main source of error from the data which is the delay due to the atmospheric refraction of the signal. On the other hand, the same synergic approach can provide very useful information of meteorological and climatological interest. GNSS and SAR observations are already used to monitor the high temporal and spatial variability of water vapour in the troposphere. It is highly possible that in the near future, InSAR near real-time water vapour products will be assimilated in weather forecasting models for improving the localization and timing of heavy precipitation events, in the same way that GNSS data are currently being used.
Our work focuses on the interaction of the three aforementioned techniques in the area of the western Gulf of Corinth in Greece and aims to optimise their synergic output by identifying strengths, weaknesses and uncertainties with respect to the measurement of zenithal tropospheric delays. We investigate the extent to which a high-resolution WRF 1-km re-analysis can produce detailed tropospheric delay maps of the required accuracy, by coupling its output (in terms of Zenith Total Delay) with the vertical delay component in GNSS measurements. The model initially is operated with varying parameterization in order to demonstrate the best possible configuration, with GNSS measurements providing a benchmark of real atmospheric conditions. In the next phase, the two datasets (predicted and observed) are compared and statistically evaluated for a period of one year, in order to investigate the extent to which meteorological parameters that affect ZTD, can be simulated accurately by the model under different weather conditions. Finally, we compare twenty Sentinel-1A interferograms with differential delay maps at the LOS produced by WRF re-analysis. We find that WRF-derived differential meteograms correlate in various degrees with the real interferograms, with results suggesting a potential of the model to re-produce both the long-wavelength stratified atmospheric signal and the short-wave turbulent atmospheric component which is evident in the interferograms.

Area D - Board 376

Water vapor plays a crucial role for the development of hazardous cumulus convection. Via the condensation process, moisture releases latent heat which becomes an energy source for the rapid development of cumulus convection. Because most of atmospheric water vapor concentrates in lower troposphere, monitoring of water vapor in low-level has received a lot of attention. For the island country like Japan, low-level moisture often come from ocean. Several studies of hazardous weather events have revealed that low-level moisture from the ocean plays an essential role in such weather cases (Kato and Aranami 2005, Shoji et al. 2009).
However, currently, there is no continuous water vapor sensor which can observe water vapor amount over the ocean.
Unlike ground-based fixed GNSS stations, ocean-platform (ship and buoy) GNSS measurements face difficulties in analyzing the variable antenna position simultaneously with the atmospheric delay. However, recent advancement of kinematic precise point positioning technology is beginning to overcome these difficulties. Shoji et al (2017) installed two GNSS antennas on a research vessel, the RYOFU MARU of the Japan Meteorological Agency (JMA), and conducted experimental observations to assess the GNSS derived precipitable water vapor (PWV) from October 19, 2016, to August 6, 2017. The GNSS derived PWVs showed good agreement with the radiosonde observations on the vessel (1.7 mm root mean square difference, −0.7 mm bias, and 3.6% rejection rate).
Kato et al. (2018) introduced a GNSS buoy system for a synthetic geohazard monitoring. The buoy is located about 40 km south of Cape Ashizuri, west of Shikoku, Japan. We succeeded continuous PWV retrieval from June to September 2018. Retrieved PWVs vividly captured water vapor variations associated with north-south shifting of the “BAIU” seasonal rain front and passages of heavy rainfalls and typhoons. The GNSS buoy PWVs also suggest that, compared to over land, uncertainty of water vapor field in JMA’s objective analysis is larger over the ocean.
Taking these results, in the year of 2018, we started a new research project to monitor PWV over the ocean west of Kyushu, Japan, getting supports from eight vessels (six regular line cargo vessels, one fisheries research vessel, and one JMA’s research vessel).

Reference
Kato, T., Aranami, K. (2005) Formation Factors of 2004 Niigata-Fukushima and Fukui heavy rainfalls and problems in the predictions using a cloud-resolving Model, SOLA 1: 1‒4. doi:10.2151/sola.2005-001
Kato T., Y. Terada, K. Tadokoro, N. Kinugasa,A. Futamura, M. Toyoshima, S. Yamamoto, M. Ishii, T. Tsugawa, M. Nishioka, K. Takizawa, Y. Shoji, and H. Seko, 2018:Development of GNSS Buoy for a Synthetic Geohazard Monitoring System, Journal of Disaster Research Vol.13 No.3, 460-471.
Shoji, Y., Kunii, M., Saito, K. (2009) Assimilation of Nationwide and Global GPS PWV Data fora Heavy Rain Event on 28 July 2008 in Hokuriku and Kinki, Japan, SOLA 5: 045–048.
Shoji, Y., K. Sato, M. Yabuki, and T. Tsuda, 2017: Comparison of shipborne GNSS-derived precipitable water vapor with radiosonde in the western North Pacific and in the seas adjacent to Japan. Earth Planets Space. 69, 153.

Acknowledgements
The shipborne GNSS campaign observation was funded by a research grant for Mission Research for Sustainable Humanosphere from the Research Institute for Sustainable Humanosphere (RISH) at Kyoto University, “Study on accuracy improvement of shipborne GNSS water vapor measurement.”
GNSS observation on top of a floating buoy is funded by JSPS KAKENHI Grants 16H06310. A part of this study is supported by JSPS KAKENHI Grants 17H00852.

Area D - Board 377

The current meteorological structure highlights the onset of a growing number of meteorological events of intense nature. In this situation it is evident that the use of any technology able to support meteorological analysis can be useful.
Considering the state-of-the-art and the new technological availabilities, the goal of this test has been the analysis of the water vapour retrieval from a GNSS low-cost permanent network; this kind of receivers have turned up more and more in marketing for their very low cost with respect to high level of performances. Nevertheless, a still open problem concerns their use for PPP applications, since standard algorithms are based on dual frequency observations to remove ionospheric delay through ionosphere-free combination. The aim of this work is to introduce a new ground based augmentation strategy to obtain PPP at centimetric accuracy level working with single frequency low-cost receivers. The developed algorithm, starting from the observations of a unique dual frequency “reference” receiver, is able to reconstruct L2 synthetic observations for any single frequency receiver located in its surroundings (15-20 km).
In this specific case we used a GNSS low-cost permanent network located in Lazio Region, composed by three receivers present in the territory of Rome, to analyse the extreme hailstorm that surprised the Capital on October 21st, 2018. We opted for studying the month of October which was characterised by violent weather patterns and we obtained Zenith Total Delay data referred to the whole period.
Data were managed by the software RTKLIB: An Open Source Program Package for GNSS Positioning. RINEX were provided by our permanent network and to reconstruct L2 synthetic observations was used a unique dual frequency “reference” receiver, M0SE, located on the roof of Civil and Industrial Engineering Faculty at Sapienza University of Rome.
GPS data were compared to meteorological data and interesting results were provided.

Area D - Board 378

The purpose of the data assimilation is to optimally use all the available information, to determine as accurately as possible the state of the atmosphere.
We show the assimilation of the GPS-ZTD (Global Positioning System - Zenith Total Delay) by a 3D-Var data assimilation system that can be used in cycling mode with the Regional Atmospheric Modeling System (RAMS; Federico, 2013).
The water vapour mixing ratio (q) and temperature (T) given by the RAMS model (background) are modified according to the assimilation of the GPS-ZTD with the purpose of improving the representation of the humidity and temperature fields and eventually decrease the model spin-up time.
To verify the impact of the GPS-ZTD data assimilation on the representation of the atmospheric humidity field, especially at the local scale, a numerical experiment is performed over an italian region.
Processing was carried out in PPP (Precise Point Positioning) using RTKLIB, an Open Source Program Package for GNSS Positioning (http://www.rtklib.com/), and validated using goGPS open source software (http://www.gogps-project.org/).
As regards GNSS single frequency receivers, they have turned up more and more in marketing for their very low cost with respect to high level of performances. Nevertheless, a still open problem concerns their use for PPP applications, since standard algorithms are based on dual frequency observations to remove ionospheric delay through ionosphere-free combination. The aim of this work is to introduce a new ground based augmentation strategy to obtain PPP at centimetric accuracy level working with single frequency low-cost receivers. The developed algorithm, starting from the observations of a unique dual frequency “reference” receiver, is able to reconstruct L2 synthetic observations for any single frequency receiver located in its surroundings (15-20 km).
For the verification of 3D-Var system, the whole dataset is divided in two parts: a set of receivers is used for the data assimilation, while the remaining stations are used for verification. Results show a substantial decrease of the RMSE of the GPS-ZTD and IWV for the stations used for the verification of the methodology; also and the absolute value of the bias is decreased, showing the potential of the assimilation of the GPS-ZTD and paving the way for the assessment of the impact on precipitation.
The results of this study are promising and will be further developed in future work with the aim to evaluate the impact of the GPS-ZTD data assimilation on the precipitation forecast.


References
Bock, O., Bosser, P., Pacione, R., Nuret, M., Fourrié, N., and Parracho, A.: A high-quality reprocessed ground-based GPS dataset for atmospheric process studies, radiosonde and model evaluation, and reanalysis of HyMeX Special Observing Period, Q. J. Roy. Meteor. Soc., 142, 56–71, https://doi.org/10.1002/qj.2701, 2016.
Federico, S.: Implementation of a 3D-Var system for atmospheric profiling data assimilation into the RAMS model: initial results, Atmos. Meas. Tech., 6, 3563-3576, doi:10.5194/amt-6-3563-2013, 2013.
Parrish, D. F. and Derber, J. C.: The National Meteorological Center’s Spectral Statistical Interpolation analysis system, Mon. Weather Rev., 120, 1747–1763, 1992.

Area D - Board 379

The Spanish satellite PAZ, owned and operated by Hisdesat, was successfully launched on February 22 2018. This satellite carries a GNSS Radio Occultation (RO) payload called the “Radio-Occultation and Heavy Precipitation” experiment (ROHP). An innovative capability of this experiment is its ability to acquire the GNSS signals at two linear polarisations to monitor precipitation (polarimetric RO, PRO). The measurement principle is based on the polarimetric phase shift between H- and V-polarised signals, received after forward scattering off rain droplets, acting as a bistatic ‘radar of opportunity’. This is done through a dual-polarization RO antenna pointing to the limb of the Earth in the anti-velocity direction of the satellite for capturing setting occultations, and a modified GNSS RO receiver to track the signals of the horizontal (H) and vertical (V) antenna ports in an independent yet synchronous way.

The first 2-polarised GNSS RO signals observation from space were acquired in May 10, 2018, when ROHP-PAZ was switched on. The system has continued to acquire PRO signals since then, and it is planned to continue operations until the end of the satellites lifetime (five years nominal life). These signals are now being used to study the potential detection and quantification of heavy rain from the differences between the phase-delays of GNSS signals propagating through large rain droplets in their vertical and horizontal components. This new measurement concept is being proven in the ROHP experiment aboard PAZ. If successful, it will represent the first sensor with capabilities to resolve coincident thermodynamic (standard GNSS RO products) and precipitation profiles in extreme rain events, with potential for improving our understanding of these type of phenomena. Figure 1 shows an example of ROHP-PAZ acquisition over Leslie hurricane.

As part of the commissioning phase and cal/val activities, extensive co-locations are being searched between ROHP-PAZ profiles, GPM measurements, IMERG products and other precipitation and cloud sensors. The studies are intended to refine the retrieval algorithms as well as to assess the complementarity between these type of measurements (including the thermodynamic profiling) and dedicated precipitation radars. The advances and results of the study will be presented, showing evidence of the link between the forward scattered polarimetric phase shift measured by this 'radar of opportunity' and hydrometeors.

Figure 1: ROHP-PAZ observation through Leslie hurricane. Left: infrared image of the hurricane overlapped with GPM precipitation information and the projection of the ROHP-PAZ observation (parallel near-vertical straight black lines). Top-right: interpolation of the GPM reflectivity along some sample ROHP-PAZ observation rays. Bottom-right: vertical profile of the ROHP-PAZ polarimetric observable and the temperature and humidity vertical profiles obtained with the same ROHP-PAZ instrument and GNSS signals (standard RO products).

Area D - Board 380

Main objective of the “BalkanMed real time severe weather service” (BeRTISS, 2017-2019) project is to establish a pilot transnational severe weather service by exploiting Global Navigation Satellite Systems (GNSS) tropospheric products to enhance the safety, the quality of life and environmental protection in the Balkan-Mediterranean region. In Bulgaria severe weather events (intense precipitation, hail and thunderstorms) are common in the summer months and are associated with large economic losses for example in agriculture. The Bulgarian Hail Suppression Agency and the Sofia University are partners in the BeRTISS project with final aim to develop the Bulgarian Integrated NowCAsting tool (BINCA). BINCA will use data from the recently deployed ground-based GNSS network of 12 stations in Bulgaria. The GNSS, weather radar as well as surface atmospheric observation will be combined with the Weather Research and Forecast (WRF) model simulations covering Bulgaria. BINCA will provide products in near-real time on a publicly accessible web platform to facilitate the operational tasks in hail suppression in Bulgaria but also other operational and public services. In this work the BeRTISS progress and the future work in Bulgaria are discussed.

Area D - Board 381

In this study a three-dimensional Global Navigation Satellite System (GNSS) tomography system is implemented in order to evaluate the tropospheric water vapour content in the Lisbon area, Portugal. This technique allows to monitor the water vapour dynamics at a local scale. Water vapour distribution plays a very important role in the atmospheric dynamics being nowadays still difficult to quantify. Numerical weather prediction models require precise three-dimensional water vapour measurements in space and time, in order to improve their forecast. This is particularly important for the prediction of extreme weather events that can potentially cause anthropogenic damage such as cyclones or flash floods caused by thunderstorms. Therefore, an accurate estimation of the three-dimensional water vapour state in near real time rate is desirable to provide information to the numerical models in order to predict in time these potential devastating events.
Currently the radiosonde technique provides precise water vapour atmospheric profiles. However, this method lacks a good geographical coverage and its temporal sampling is usually reduced to a few radiosonde launches per day (1 per day in Lisbon for example). On the other hand, satellite meteorological data, such as Atmospheric Infra-Red Sounder (AIRS) or Infrared Atmospheric Sounding Interferometer (IASI), acquire data in a large area footprint providing good spatial resolution but with the drawback of having an even poorer temporal resolution. In recent years GNSS has proved its capacity to measure the integrated water vapour content from a single receiver with a precision comparable to other meteorological techniques, i.e. radiosonde. It has the ability to provide measurements in all weather conditions with a continuous temporal sampling and with a spatial resolution depending on the density of the local GNSS network of stations. However these measurements lack discretization of the water vapour thought-out the vertical direction.
Integrated water vapour measurements obtained from single GNSS stations are a mean value of the original slant path observations defined by the line of site of the observations from a single receiver to group satellites visible in the horizon. The GNSS tomography implementation requires the reconstruction of these slant water vapour GNSS observations, which are performed using a group of GNSS stations forming a network. The establishment of a three-dimensional grid of voxels (volumetric pixels) along the troposphere over the GNSS network area, enables the ray tracing of the slant water vapour observations within that region, allowing the application of the tomography technique. Each slant water vapour observation from each satellite and GNSS station is parametrized as an equation, resulting in a system of equations following the typical inverse problem formulation. The inversion of this system provides a three-dimensional water vapour solution estimation of the tropospheric area represented by the grid of voxels. Nevertheless, the GNSS tomography technique is affected by two main problems. One is due to the GNSS non optimal geometry, which causes the lower grid levels to be less traversed by observations leading to a sparser distribution of the water vapour data. This causes an ill-posed problem to the system inversion that usually can only be overcome with the inclusion of constraints to the system of equations. The second issue is that the GNSS tomography system requires a good initial state of the three-dimensional water vapour state in order to provide an adequate result.
Following this second major issue, the aim of this work is to evaluate the dependence of the GNSS tomography solution and its temporal evaluation using different initial solutions. In this work we gather radiosonde measurements obtained within the study area and also IASI profile measurements acquired also inside the tomographic grid. A continuous hourly three-dimensional water vapour solution is obtained from the GNSS tomography processing. Different setups are tested using IASI, radiosonde and combining both, computing different water vapour states to initialize and updated the tomography series of continuous solutions. The dependence of the GNSS tomography solution on the initial and updated conditions obtained from different datasets (Radiosonde, IASI or Radiosonde plus IASI) and the solution stability over time is accessed comparing the various series of tomography results with the radiosonde water vapour profiles. It is found that the most feasible continuous GNSS tomography is obtained when combining both IASI and radiosonde measurements, following what is usually observed in the numerical weather models, were the assimilation of more data provides best results.
This study was funded by the Portuguese Science Foundation FCT, under the grants UID/GEO/50019/2013.

Area D - Board 382

In addition to their primary purpose of positioning, GNSS signals have being used for years also in meteorology for the estimation of atmospheric water vapor (WV), which slightly decreases the phase velocity of the signal in the air. The main limitation of the current GNSS-WV retrieval techniques is that they provide only integrated values, without any information about the vertical distribution of WV, unless properly designed and combined GNSS observation networks are available.
The present work applies the basic principles of bayesian inference to get useful information on vertical profiles of WV, and possibly temperature, from ground measurements of GNSS signals delay and atmospheric pressure (P), temperature (T) and relative humidity (RH).
The method requires a set of numerically computed atmospheric states, to which the prior probabilities are assigned. Accordingly such a set is considered exhaustively representative for all the possible conditions we are concerned in. In our case we use MERRA reanalysis at synoptic times, all over a period of ten years. For the calibration of the method, time series of concomitant ground observations of GNSS zenith tropospheric delay (ZTD), P, T and RH, on
a number of sites distributed in the region of interest are also needed. We use the International GNSS Service (IGS) sites data for this purpose.
Essentially the calibration consists in applying a best fit procedure to compute the value of the parameters entering the injective relationship y(x) between the numerical model state variable x and the observable y. Once these parameters are determined, a simple gaussian assumption for the residuals distribution, allows the computation of the likelihood p(x) for any possible state x, given the observation y. Thus the prior probability for the possible atmospheric states is changed in accordance to their posterior likelihood, as computed in the
light of the observations. In many situations the information content of the ZTD observation added to the P, T and RH measurements increase significantly the posterior probability of some vertical WV profiles with respect to all the others (as represented by the numerical model states x).
Finally the presented bayesian method of retrieval is tested on locations where balloon data are available. The comparison with directly measured vertical profiles allows to evaluate the extent the calibration remains valid, even outside the period and the sites on which it was carried out - that if this was not the case, the retrieval method would be of little use.

Area D - Board 383