When using InSAR data for assimilation in weather models one of the main obstacles is the differential nature of InSAR. Interferograms only give information on the difference in atmoshperic delay between two acquisitions at two locations. InSAR data by itself will therefore give no absolute values of atmospheric delay, while most weather model assimilation techniques are based on absolute single epoch data.

Current approaches to assimilate InSAR data in weather models therefore try to isolate one of the epochs of an interferogram by diminishing the effect of other epochs using external data. This can be the model state of the model itself, GNSS information or other remote sensing techniques which measure related weather parameters like integrated water vapor. However, these measurements have either much lower resolutions or much lower precision than InSAR data, which deteriorates the quality of the final product greatly.
An alternative to these methods is to solve the atmospheric delay per epoch based on a network of interferograms combined with some assumptions on model statistics. The problem here is that often these assumptions do not hold and these network solution can become unstable.

Therefore, we propose a new method where a network of interferograms is directly combined with a series of NWP model realizations during the satellite overpasses. To do so, single epoch atmospheric delays are simulated from the NWP model based on a ray-tracing technique.
The method comprises of the following steps: 1) The mean value of every interferogram is adjusted to the mean of the corresponding differential model simulations 2) The uncertainty in the model simulations are estimated based on the deviation of the interferograms and differential model simulations. 3) The network of interferograms and model simulations is solved using a regularized least-squares on a pixel by pixel basis.
If the mean of the resulting atmospheric delay fields differs to much from the model outputs we can iterate this process using the new atmospheric delay fields instead of the simulated model data.
Finally, the retrieved absolute single epoch delay fields can be used to obtain the delay fields of new incoming SAR acquisitions without using new model realizations. These delay fields can then be used for data assimilation.

The advantage of this approach is that the precision of the InSAR information does not deteriorate because the model data is converted to atmospheric delay, which prevents any errors due to conversion of InSAR data to other variables like integrated water vapor. Further, the quality of the weather model realization can be evaluated before they are used for regularization, which prevents the propagation of errors from the weather model estimates.
The approach will be demonstrated using a series of Sentinel-1 images and Harmonie model data over the Netherlands.

Highly urbanized Tokyo metropolitan area is vulnerable to severe storms such as local torrential rainfall, thunder lightning and tornado. Recently, the number of days with thunderstorms has been increasing in Tokyo. Elucidation of the mechanism of extreme weather in Tokyo is urgent matter (Nakatani et al. 2013). To study the rolls of water vapor and wind for the rapid life-cycle (initiation, development and dissipation) of hazardous cumulus convection, we installed a mobile Raman-lidar (Sakai et al. 2018), a wind profiler, and five GNSS stations in and around Tokyo.
Using 3DVAR capability of WRF data assimilation system (WRFDA)(Barker et al. 2012) , three-dimensional atmospheric fields with 1 km horizontal resolution were produced at every 10min throughout the August 2017 and 2018. Background fields were produced by JMA nonhydrostatic model (NHM)(Saito et al. 2007). In addition to the observed data at these campaign observation instruments, Japan Meteorological Agency’s operational surface meteorological stations, wind-profiler stations and GNSS PWV derived at dense GNSS network of Geospatial Information Authority of Japan are assimilated. We confirmed that the reproduced atmospheric fields (Temperature, humidity, and wind) showed better agreement with observations of radiosonde, lidar derived specific humidity, and wind-profiler than background fields at altitudes lower than 2 km. Assimilation of surface meteorological observations (atmospheric pressure, temperature, humidity and wind) introduced better agreement mainly at altitude lower than 0.5 km. Inclusion of GNSS derived PWV resulted better agreement of humidity at altitude higher than 0.5 km and inclusion of wind field from wind-profiler led better agreement of wind field.
We have investigated roles of water vapor for several local heavy rainfall events occurred in Tokyo in August 2017 and 2018 and found that convergence of water vapor flux at 0.5 km altitude had intensified prior to the occurrence of torrential precipitation. Statistical and quantitative results will be discussed at the symposium.

Reference
Barker D., X.-Y. Huang, Z. Liu, T. Auligné, X. Zhang, S. Rugg, R. Ajjaji, A. Bourgeois, J. Bray, Y. Chen, M. Demirtas, Y.-R. Guo, T. Henderson, W. Huang, H.-C. Lin, J. Michalakes, S. Rizvi, and X. Zhang, 2012: The Weather Research and Forecasting Model's Community Variational/Ensemble Data Assimilation System: WRFDA. Bull. Amer. Meteor. Soc., 93, 831–843.
Nakatani T., Y. Shoji, R. Misumi, K. Saito, N. Seino, H. Seko, Y. Fujiyoshi and I. Nakamura, 2013: WWRP RDP Science Plan: Tokyo Metropolitan Area Convection Study for Extreme Weather Resilient Cities (TOMACS). WWRP report for Joint Scientific Committee, 26pp. (Available online at http://www.wmo.int/pages/prog/arep/wwrp/new/documents/Doc4_6_TOMACS_RDP_proposal_20130704.pdf)
Saito, K., J. Ishida, K. Aranami, T. Hara, T. Segawa, M. Narita, and Y. Honda, 2007: Nonhydrostatic atmospheric models and operational development at JMA. J. Meteor. Soc. Japan, 85B, 271‒304.
Sakai T., T. Nagai, T. Izumi, S. Yoshida, Y. Shoji, 2018: Mobile water vapor Raman lidar for heavy rain forecasting: system description and validation, Atmospheric Measurement Techniques (in revision).

Acknowledgements
This study is supported by JSPS KAKENHI Grants 17H00852. We used RTKLIB ver. 2.4.2 (http://www.rtklib.com/) for GNSS analysis with IGS final ephemeris (available at ftp://cddis.gsfc.nasa.gov/gnss/products/).

In September 2018, the G-Class candidate mission was chosen by ESA to enter the pre-feasibility study and compete to be the tenth Earth Explorer mission within the Earth Explorer 10 Program. G-Class would carry a synthetic aperture radar and would be placed in a geosynchronous orbit to provide a constant view of Africa and the Mediterranean regions. The mission aims to make observations of diurnal water cycle processes to improve the prediction capability of rainfall, water availability, flooding, and landslides. This would enable meteorologists to more accurately forecast how much precipitation will fall in violent storms, improving flood development prediction.
The ESA Sentinel-1 SAR mission has demonstrated its capability to map the atmosphere’s Precipitable Water Vapor (PWV) over large regions. In particular, merging sequential PWV maps and combining different orbits can reduce the temporal sampling of PWV over a given region to 1 to 2 days depending on the extension of the region. The assimilation of those maps in a Numerical Weather Model has shown a great advantage relative to the PWV estimated by GNSS receivers. In this work, we study the possible advantages that could bring by PWV maps generate by geosynchronous SAR missions. The questions we try to reply in this work are: Can those proposed SAR acquisition modes be really useful for mapping atmosphere PWV? Are they adding more information with respect to that provided by the current network of permanent GNSS receivers? Is it more effective to slightly increase the density of current GNSS networks? In this work, we show PWV maps obtained over different regions. Furthermore, we study the problem of assimilating PWV measurements as provided by GNSS stations and Sentinel-1 in a Numerical Weather Model. The result of the assimilation of the PWV measurements with the characteristics of spatial resolution and temporal sampling that would have if provided by geosynchronous systems are also shown. The aim of this work is to study, from the point of view of atmospheric scientists, the advantages/disadvantages that PWV maps, provided by Sentinel-1, have with respect to GNSS and the advantages/disadvantages that geosynchronous SAR could have with respect to both sun-synchronous SAR and GNSS systems.

ACKNOWLEDGMENTS
Fundação para a Ciência e Tecnologia (FCT) – project UID/GEO/50019/2013 and SFRH/BPD/96069/2013.

A single Interferometric Synthetic Aperture Radar (InSAR) interferogram provides observations of ground movements with centimetric accuracy, and therefore can only detect large ground motion such as those caused by the co-seismic slip or volcano eruptions. To detect small-amplitude and long-term surface displacements such as post/inter-seismic motion or ground subsidence, a time series of interferograms is often used to overcome the errors resulting from the atmosphere, DEM and orbit. In most of the currently available InSAR time series analysis packages, two fundamental assumptions are made, namely that (i) deformation signals are correlated in time, and (ii) atmospheric effects are correlated in space but not in time. Unfortunately, since atmospheric effects can be highly correlated with topography, the second assumption does not hold in most cases. The temporal correlation of atmospheric delays may completely mask or bias the geophysical signals and introduce unpredictable uncertainties in the velocity estimates.

To overcome this, we proposed a new method which (i) employs the Generic Atmospheric Correction Online Service for InSAR (GACOS, global and all-time useable in near real-time) to correct atmospheric effects for each interferogram; (ii) has proper quality control by internal cross tests to reject the date(s) with poor correction performance; (iii) estimate and corrects for the atmospheric delays for those interferograms identified in step (ii) using a sub-network approach; and (iv) applies the conventional time series tool to extract the mean velocity and time-dependent displacements.

The proposed method was applied on Sentinel-1 InSAR measurements, with strong atmospheric contaminations at coastal areas, to identify the triggered afterslip on southern Hikurangi Subduction Slab Following the 2016 Kaikoura earthquake. The resultant time series fill the data gap west and southwest of Marlborough Fault System (MFS) after correcting for atmospheric effects and therefore enable us to show clear evidence of a triggered Hikurangi subduction slab at its southern part. The afterslips suggest the Hikurangi subduction interface had an existing but low co-seismic moment release and accommodated up to 2 m plate motion during one year of post-seismic distribution. We also found shallow slip sources on the northern extensions of the Needles fault, which might induce large offshore surface displacement. The 2016 Kaikoura earthquake turns out to be a giant engine that triggered not only over 12 adjacent crustal faults, but also nearly the entire Hikurangi subduction slab, including a slow slip event on the active northern part, and afterslips on the less active central offshore part plus, most significantly, on the previously inactive southwestern end beneath the MFS.

Different techniques exist to estimate the amount of Precipitable Water Vapor (PWV) in the atmosphere. Traditionally, radiosondes are released which offer precise measurements but are very limited in space and time and their ascension trajectory depends on the wind. Satellite-based measurements (e.g. Interferometric Synthetic Aperture Radar) are characterized by a good spatial but poor temporal resolution. Another possibility to estimate PWV is to exploit the signal delays from Global Navigation Satellite System (GNSS) satellites. Recent research demonstrates that assimilating PWV from GNSS receivers on the ground can be beneficial for extreme weather forecasting. However, existing GNSS monitoring networks are not sufficiently dense to capture small-scale water vapor fluctuations. A densification with geodetic-grade receivers and antennas is economically not feasible. Instead, consumer-grade single-frequency receivers offer an alternative. In this study, we co-located two single-frequency stations to an IGS station and present verification results. In another experiment, the GNSS data from a geodetic-grade dual-frequency antenna was split to a geodetic receiver and two low-cost receivers. We applied the Satellite-specific Epoch-differenced Ionospheric Delay (SEID) model using two different reference network configurations of 50-80 kilometers and 200-300 kilometers mean station distance to account for the ionospheric delay. Our experimental setup enabled us to distinguish between the antenna, ionospheric interpolation and software related impacts on the quality of PWV retrievals. Modeling of the ionosphere and antenna type are the main sources influencing the Zenith Tropospheric Delay (ZTD) accuracy. It is demonstrated that the single-frequency units indicate a root mean square error (RMSE) comparable to the synthesized single-frequency dataset extracted from the original IGS dual-frequency data. The precision degrades from about 5 mm ZTD RMSE to 10 mm using longer station distances. Using a cheap patch antenna, a bias of approximately 5 mm was found. We conclude that single-frequency GNSS receivers can serve as a promising complement to existing GNSS networks. However, especially with respect to near-real-time processing, other techniques (e.g. network adjustment) have to be taken into account. This research is part of the project BRIGAID (BRIdging the GAp for Innovations in Disaster resilience), funded by EU-Horizon2020.

Recently, is has been recognized as SAR interferometry can also become a tool for the high resolution mapping of water vapor in atmosphere giving rise the development of the so called SAR meteorology. This recent concept has mainly benefited from the availability of Sentinel-1 images characterized by a mean temporal baseline of six days, depending on the geographical location, a spatial resolution of about twenty five meters and an even increased spatial coverage up to the scale of country size. In all above InSAR applications of C-band SAR images, the propagation delay in ionosphere was neglected due to both smaller delay in the C-band with respect to the L-band. However, the larger spatial coverage of Sentinel-1 pointed out the need to correct for ionospheric propagation delay even in C-band. Even if delay can be considered a second order correction in SAR interferograms, it becomes useful when providing precise maps of water vapor in atmosphere. The Total Electrom Content (TEC) maps made available by ionosphere community were used to compute the propagation delays at the acquisition times of the two SAR images used to generate the interferogram.
The ionosphere is, as the outer layer of the Earth, permanently exposed by many different influences coming from the outer space. For this reason many astrophysical phenomena can significantly disturb the ionospheric plasma and, because of the increase in the intensity of the ionizing radiation, induce rise of the local electron density that can reaches several orders of magnitude.
Although the sudden variations often have not enough intensity to significantly affect SAR signals due to their high frequency, the non-negligible increase in signal delay can be expected during occurrences of some types of phenomena. The importance of the influence of the particular phenomenon depends on its nature and the geographical location of the considered ionospheric part. There are several parameters which are the most important in the relevant analyses: localization of the event induced radiations, period of the day when it occur, radiation properties, geomagnetic field characteristics of the area between the event location and the considered medium, latitude and altitude of the analyzed ionospheric part etc. All of these parameters can significant influence on the variation of the electron density and, consequently, the TEC for the medium within the SAR signals propagate.
In this study we present analysis of influences of different astro and geophysical phenomena on propagation of SAR signals with different frequencies. We pay attention on variations in contribution of various ionospheric regions during perturbed conditions and indicate need for analysis of adequacy of using usual models in the TEC calculations in presence of intensive pertubations in a particular altitude domains.