No abstract

<p>One of the indispensable elements of high-resolution weather forecast systems is the provision of reliable initial conditions using observations. Among the methods for collecting meteorological data, besides the quality of measurements, their time and space variability play a crucial role. Hence, GNSS observations stand out as stable, bias-free alternatives for weather stations, radiosondes, or microwave satellites.</p> <p>Current studies of GNSS observations in weather forecasting give promising results. However, the observations themselves are subject to errors due to their geometry, mainly caused by insufficient vertical and horizontal resolution. Therefore, applying them in an operational forecasting model is challenging. A possible way to solve this is to integrate space and ground-based observations into one tomography model.</p> <p>The solution should be able to detect local, extreme weather phenomena with repeatable uncertainty and high numerical stability. Hence, we propose a precise 3D ray tracing solution for effective simulations of the ray path between the GNSS satellite and the GNSS receiver (Low Earth Orbiting LEO satellite), along with the ground receiver. Although, the combination of these results in one computationally efficient and stable model is a complex task.</p> <p>The following step is the 3D ray tracing simulation integration into a modified TOMO2 operator dedicated to the tomography of 3D wet refractivity fields. The ray tracing module collects information on ray points’ refractivity and distance traversed in models’ voxels along the ray path. Then delivers it to mutual observational matrices for ground- and space-based simulations. </p> <p>This study focuses on the methodology of integrated tomography modeling.  Results are compared to the ground-based only GNSS tomography solution and validated with radiosondes profiles. The case studies are based on severe weather events in Poland with RO data delivered by SPIRE company and GNSS ground-based observations produced by UPWr. Numerical Weather Model input comes from European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5.</p>

<p>The number of tropospheric observations that are assimilated in current numerical weather forecasting systems, is large. From automatic weather stations, to geostationary satellites, from radiosondes to polar orbiting  microwave satellites. Amid the currently available data sources, GNSS stands out as a bias free, self calibrating, high fidelity temperature and water vapour measurements. <br>Until recently GNSS was used weather forecasting only in two ways: as a ground-based point, high-frequency observation of integrated water vapour (IWV) or zenithal integrated observation of temperature and water vapour, or as a sparse space-based profile observation of temperature and water vapour content (provided as a refractivity or bending angle profile). In the last couple of years GNSS tomography, a 3D imaging technique, is gaining attention as a weather model data source. However, low space resolution combined with large uncertainty of the tomography reconstruction makes this technique difficult to apply in operational forecasting. <br>Therefore this technique, to be considered as a valuable data source in weather models, has to be numerically stable with known repeatable uncertainty. We believe that a way forward is to combine space-based and ground-based observations using the tomography principle. A way forward is to effectively simulate the signal trajectory between the GNSS transmitter and GNSS receiver (Low Earth Orbiting LEO satellite). 3D ray-tracing modelling of the radio occultation (RO) event based on Numerical Weather Model is performed. The challenge here is to make these ray-tracing results comparable with excess phase observations at the LEO satellite.  <br>Modelling by 3D ray-tracing is performed by the modified Atmospheric TOMography (ATOM) software with the use of the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 model. This module uses the position of the GNSS satellites as starting point and iteratively propagates the signal path by collecting information on refractive parameters along its path based on nodal points. This study is based on the ten selected RO events from the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) report “Optimising tracking strategies for Radio Occultation. Task 1 - the profile dataset.”. Modelling was performed by varying the grid resolution of the ERA5 model and the length of a propagator step size segment to obtain total excess phase delay values. Additionally, Radio Occultation Processing Package (ROPP) 2D ray-tracing multiple phase screen simulation was run to confront obtained from ATOM phase delays. The COSMIC Data Analysis and Archive Center (CDAAC) observed excess phase was used as a reference data source. </p>