Improved Troposphere Blind Models Based on Numerical Weather Data

G. Möller, R. Weber, J. Böhm

Abstract:	The main effect while GNSS signals transit the troposphere (the lowest part of the atmosphere) is the path delay of the signals by variation of the refraction index due to temperature, pressure and water content. Dependent on this variation the tropospheric delay can reach up to a few tens of metres for low elevation angles - hence it is a limiting factor for most GNSS applications. Several troposphere models have been developed to alleviate the effect of the troposphere on GNSS signals. All these models can be allocated to one of the three categories: “profile-based”, “site-based” or “blind” models. The highest accuracy can be reached by applying profile-based models like the model derived from Boehm J. et al. together with the Vienna Mapping Function (VMF1). For this kind of models Ray-Tracing is applied to refractivity profiles from a numerical weather model (NWM), e.g. of the European Centre for Medium-Range Weather Forecasts (ECMWF). Nafisi et al. have developed the Vienna Ray-Tracer (2D and 3D) which allows determining the delay along the trajectory of a signal which is travelling though the atmosphere. Due to high data transmission rates and high computational efforts it is challenging to operate a valid Ray-Tracing software at user site. One solution to overcome this problem is to run the Ray-Tracing software in the operating centre and to forward either troposphere residuals or troposphere delays with consistent coefficients of a high quality mapping function to the user. In both cases users must be able to receive these data, e.g. via mobile internet. For users without access to information about the actual state of the atmosphere the so called “blind” models are developed. These models are based on physical principles and its input parameters are mostly derived by statistical analysis of NWM time series and are stored in the receiver database. The accuracy which can be reached with blind models is limited. Nevertheless these models are used as standard in most GNSS receivers because blind models can be applied at any time and any position on Earth and its implementation is rather convenient. Dependent on availability meteorological data can be used as input as well to improve the accuracy of blind models. The currently available blind model RTCA-MOPS - derived from the USA Standard Atmospheres is adopted as SBAS standard for WAAS and EGNOS. It is based on 10 obsolete atmospheric input parameters and the troposphere delay is modelled as a periodic function of the day of the year without distinction between Northern and Southern hemisphere. More advanced blind models are using climate maps on a global grid but most of these maps are outdated. An improvement in accuracy can be expected if available blind models would use as input data derived from actual NWM, like ERA40 or ECMWF-INTERIM. In addition it seems attractive if a Ground Based Augmentation System (GBAS) or a Space Based Augmentation System (SBAS) like EGNOS would provide actual troposphere correction parameters. Especially in combination with an error model for extreme events and conditions, such kind of troposphere model could be used in critical systems like Safety-of-Life or traffic control navigation systems. To obtain the tropospheric correction parameters, an analysis of GNSS observations from ground networks or NWM data might be feasible. The use of real-time NWM data seems promising because of continuous improvements of operational NWM in the last years. The main advantages of using NWM data is that no additional ground infrastructure is necessary and meteorological observations can be derived for any position on earth. If the signal delay in the neutral atmosphere has to be known with highest accuracy, still it should be derived from GNSS observations instead from NWM. Hence we will concentrate on both techniques and will show the potential for the derivation of tropospheric correction parameters. At first we give an overview about state-of-the-art troposphere correction models which are commonly used in blind mode. This comprises the RTCA-MOPS model, the more advanced ESA reference model and a blind model derived from actual ECMWF data. We calculate tropospheric delays for a global grid and for selected sites over a period of one month. Based on these delays we compare the different models and validate the results by independent data sources like the profile-based model derived from actual ECMWF data and globally distributed IGS tropospheric zenith path delays. In addition the independent data sources will be used to detect possible error sources and extreme weather events in order to evaluate the maximum expectable tropospheric range error. We will show the potential of a new troposphere model and present the data which are relevant for the development of such a new model. Finally we will give an outlook on the potential of the 3D ground based GNSS Atmospheric Tomography technique to derive range corrections and to reduce the tropospheric residual error for precise positioning in Local Area Augmentation Systems (LAAS).
Published in:	Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013) September 16 - 20, 2013 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	2489 - 2495
Cite this article:	Updated citation: Published in NAVIGATION: Journal of the Institute of Navigation
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In