Abstract
In this study different approaches based on multilayer perceptron neural networks are proposed and evaluated with the aim to retrieve tropospheric profiles by using GPS radio occultation data. We employed a data set of 445 occultations covering the land surface within the Tropics, split into desert and vegetation zone. The neural networks were trained with refractivity profiles as input computed from geometrical occultation parameters provided by the FORMOSAT-3/COSMIC satellites, while the targets were the dry and wet refractivity profiles and the dry pressure profiles obtained from the contemporary European Centre for Medium-Range Weather Forecast data. Such a new retrieval algorithm was chosen to solve the atmospheric profiling problem without the constraint of an independent knowledge of one atmospheric parameter at each GPS occultation.
Highlights
Global Positioning System (GPS) radio occultation (RO) is a global sounding technique for the atmospheric profiling useful for numerical weather models and climate studies
We have proposed a new retrieval algorithm based on multilayer perceptron neural networks to derive profile of atmospheric parameters from RO refractivity profiles overcoming the requirement for temperature profile availability at each GPS occultation
The vertically averaged Root Mean Square (RMS) error of N computed from Abel transformation, i.e. the input of the networks, and the mean standard deviation of the entire European Centre for Medium-Range Weather Forecast (ECMWF) database are reported in Table 3 and in Table 4, for desert and vegetation zone respectively
Summary
Global Positioning System (GPS) radio occultation (RO) is a global sounding technique for the atmospheric profiling useful for numerical weather models and climate studies. The proposed algorithm shows the possibility to estimate tropospheric parameters included the wet ones only from RO refractivity, after the settlement of the training phase of neural networks, and the possibility to increase the atmospheric observations, thanks to a wide spatial coverage of RO soundings. For this purpose, the employment of neural networks proved useful and different training approaches were tested and evaluated from the standpoint of the retrieval accuracy, and in terms of computational cost such as time and memory requirements
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call