An attempt for development of pavements temperature prediction models based on remote sensing data and artificial neural network

ABSTRACT

Results of Falling Weight Deflectometer (FWD) test are dependent upon pavement temperature and its changes daily or seasonally. It is crucial to have the temperature of pavement when FWD data are collected, and also to compare the FWD data for different temperatures. Conventional models for pavement temperature rely on parameters such as air temperature, solar radiation, wind speed, and humidity. However, the accuracy of these data might not be high when the pavement is far from meteorological stations. Thus, this study is investigating the applicability of using remote sensing technology for the estimation of pavement temperature. Satellite input data are used to develop three models as linear, non-linear and Artificial Neural Network (ANN) for depths of 20 and 25 cm and in different layers of asphalt pavement. The predicted temperatures are very close to the measured temperature(correlation coefficients 0.79–0.99). Moreover, the results of this model are compared with the BELLS3 model for Chosen sites in Raton Pass, Aguilar, and Schurz-Nevada, based on the data in IOWA State University in the United States. Nonetheless, the results of this model can be extended for similar conditions with hot weather, subtropical area and low vegetation.

KEYWORDS:

• Temperature prediction models
• pavement
• remote sensing
• regression
• artificial neural network

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 https://www.wunderground.com/