Analysis of the influence of forestry environments on the accuracy of GPS measurements by means of recurrent neural networks

7Citations
Citations of this article
27Readers
Mendeley users who have this article in their library.

Abstract

The present paper analyses the accuracy of the measurements performed by a global positioning system (GPS) receiver located in forested environments. A large set of observations were taken with a GPS receiver at intervals of one second during a total time of an hour at twelve different points placed in forest areas. Each of these areas was characterized by a set of forest stand variables (tree density, volume of wood, Hart-Becking index, etc.) The influence on the accuracy of the measurements of other variables related to the GPS signal, such as the position dilution of precision (PDOP), the signal-to-noise ratio and the number of satellites, was also studied. Recurrent neural networks (RNNs) were applied to build a mathematical model that associates the observation errors and the GPS signal and forest stand variables. A recurrent neural network is a type of neural network whose topology allows it to exhibit dynamic temporal behaviour. This property, and its utility for discovering patterns in non-linear and chaotic systems, make the RNN a suitable tool for the study of our problem. Two kinds of models with different numbers of input variables were built. The results obtained are in line with those achieved by the authors in previous research using different techniques; they showed that the variables with the highest influence on the accuracy of the GPS measurements are those related to the forest canopy, that is, the forest variables. The performance of the models of the RNN improved on previous results obtained with other techniques. © 2012 Elsevier Ltd.

Cite

CITATION STYLE

APA

Ordóñez Galán, C., Rodríguez Pérez, J. R., García Cortés, S., & Bernardo Sánchez, A. (2013). Analysis of the influence of forestry environments on the accuracy of GPS measurements by means of recurrent neural networks. Mathematical and Computer Modelling, 57(7–8), 2016–2023. https://doi.org/10.1016/j.mcm.2012.03.006

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free