Please use this identifier to cite or link to this item:
Title: Prediction of annual and seasonal soil temperature variation using artificial neural network
Authors: Jebamalar, A S
Raja, S Abraham Thambi
Bai, S Jeslin Sunitha
Keywords: Artificial Neural Network (ANN);Back Propagation Algorithm;Multi-layer Perceptron;Soil Temperature Prediction.
Issue Date: Feb-2012
Publisher: NISCAIR-CSIR, India
PACS No.: 92.40.Lg; 84.35.+i
Abstract: The back propagation algorithm, an artificial neural network (ANN) training algorithm, is a widely applied mathematical implementation for spatial monitoring and is used in the present study for the analysis and prediction of soil temperature. The soil temperature data at 10 and 20 cm soil depths were collected from the Agricultural College and Research Institute, Killikulam, Tuticorin District of Tamil Nadu. The observed values during the year 2004 at 10 and 20 cm soil depths were plotted to understand the annual and seasonal behaviour of the temperature wave. The wave characteristics such as range of soil temperature and rate of change of temperature/week were estimated and tabulated. Data for 1993 – 1997 (5 years) and 1993 – 2002 (10 years) were separately used as inputs for the prediction of soil temperature in 2004 using ANN. The predicted values were compared with the observed values and statistically validated. The characteristics of predicted annual and seasonal wave were also compared with observed values. It was found that the predicted values of annual wave fitted well with observed ones with little variation for the seasonal waves. The range of soil temperature for predicted values coincided almost with the observed ones with regard to the annual and the seasonal waves for both 10 and 20 cm soil depths. The rate of change of temperature/week of the predicted values coincided well with the observed ones for 10 cm soil depth. For 20 cm soil depth, the predicted values deviated from the observed ones for the winter season while the annual and pre-monsoon seasonal waves coincided well with the observed values. The surface temperature was also predicted independently from 10 and 20 cm soil temperature and error validation was done. From these, it may be convincingly stated that the ANN can be used as a good mathematical model for the prediction of soil temperature.
Page(s): 48-57
ISSN: 0975-105X (Online); 0367-8393 (Print)
Appears in Collections:IJRSP Vol.41(1) [February 2012]

Files in This Item:
File Description SizeFormat 
IJRSP 41(1) 48-57.pdf159.48 kBAdobe PDFView/Open

Items in NOPR are protected by copyright, with all rights reserved, unless otherwise indicated.