103(3)_str42

ISSN 1392-3196 / e-ISSN 2335-8947
Zemdirbyste-Agriculture, vol. 103, No. 3 (2016), p. 327–334
DOI  10.13080/z-a.2016.103.042

Measurements and modelling of wind erosion rate in different tillage practices using a portable wind erosion tunnel

Kazım ÇARMAN, Tamer MARAKOĞLU, Alper TANER, Fariz MIKAILSOY

Abstract

Artificial intelligence systems are widely accepted as a technology providing an alternative method to solve complex and ill-defined problems. Artificial neural network (ANN) is a technique with a flexible mathematical structure, which is capable of identifying a complex nonlinear relationship between the input and output data. The objective of this study was to investigate the relationship between dust concentration and wind erosion rate, and to illustrate how ANN might play an important role in the prediction of wind erosion rate. Data were recorded via field experiments by using a portable field wind tunnel. The experiments were carried out for eight different tillage applications that include the conventional, six different reduced tillage and the direct seeding practices. Particulate matter (PM) concentration generally decreased with a decrease in number or intensity of tillage operations. Direct seeding resulted in the lowest PM10 concentration. After tillage applications, wind erosion rate varied between 113 and 1365 g m-2 h-1. Results showed that wind erosion rate was lower in direct seeding than in conventional and reduced tillage applications. In this paper, a sophisticated intelligent model, based on a 1-(8-5)-1 ANN model with a back-propagation learning algorithm, was developed to predict the changes in the wind erosion rate due to dust concentration occurring during tillage. In addition, the prediction of the model was made according to traditional methods of wind erosion rate by using the programme Statistica, version 5. The verification of the proposed model was carried out by applying various numerical error criteria. The ANN model consistently provided better predictions compared with the nonlinear regression-based model. The relative error of the predicted values was found to be less than the acceptable limits (10%). Based on the results of this study, ANN appears to be a promising technique for predicting wind erosion rate.

Key words: artificial neural network, conservation tillage, dust concentration, soil erodibility by wind.

Full text: 103_3_str42.pdf