COMPARISON OF ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM, ARTIFICIAL NEURAL NETWORKS AND NON-LINEAR REGRESSION FOR BARK VOLUME ESTIMATION IN BRUTIAN PINE (PINUS BRUTIA TEN.)


Catal Y. , SAPLIOĞLU K.

APPLIED ECOLOGY AND ENVIRONMENTAL RESEARCH, cilt.16, ss.2015-2027, 2018 (SCI İndekslerine Giren Dergi) identifier identifier

  • Cilt numarası: 16 Konu: 2
  • Basım Tarihi: 2018
  • Doi Numarası: 10.15666/aeer/1602_20152027
  • Dergi Adı: APPLIED ECOLOGY AND ENVIRONMENTAL RESEARCH
  • Sayfa Sayıları: ss.2015-2027

Özet

The bark is required to determine the volume of bark while identifying tree volume in forest planning. Since the bark volume of brutian pine (Pinus brutia Ten.) is considerably more when compared to other tree species, a real-like estimation should be made for the amount of bark. In this study, the Adaptive Neuro-Fuzzy Inference System (ANFIS) method was implemented in the bark estimation. The results obtained with the ANFIS method were compared with the results obtained with the Non-Linear Regression (NLR) and Artificial Neural Networks (ANN) methods. Among the eight models that can be used to determine the bark volume in the NLR method, the Morgan-Mercer-Flodin (MMF) model was determined to be the model giving the best results. Brutian pine bark volume model with the smallest values of the indicators used (MAE = 0.01630; RMSE = 0.02345; FI = 0.955363 and Bias = 0.00151) is the MMF nonlinear model. The amount of bark obtained with the ANFIS method provided better results when compared to ANN and to NLR. The slope graphs for the values estimated with the real value and method for the ANFIS, ANN and NLR methods were found to be 44.01 degrees, 44.60 degrees and 44.83 degrees, respectively. In conclusion, the bark estimation with the ANFIS method provided better results when compared to the ANN and LNR methods.