A neural network implementation for near real time tropospheric water vapour profiling over Nairobi using ground-based gps receiver

Abstract

A remote sensing tool employing an Artificial Neural Networks algorithm was proposed for near real time determination of the relative humidity profile using Global Positioning System (GPS) data recorded by a ground-based GPS receiver. The GPS data was processed to obtain the Integrated Water Vapour. This Integrated Water Vapour in conjunction with ground level information for temperature, pressure and relative humidity were fed as inputs to the developed neural network which in turn generated the instantaneous relative humidity profile as output. GPS and radiosonde data for the years 2009 and 2010 were used to train the system while the same data for 2011 were used to validate the system. The RH profile results for 2011 generated using GPS data and the neural network, upon comparison with recorded in situ radiosonde relative humidity profile measurements for the same days and times in the year 2011, had Root Mean Square Error of less than 4%, which fell within the margin of error of the Vaisala RS92 Radiosonde’s humidity measurement regime.