Studies On Age Estimation Algorithms Of Living Trees Based On The Micro-damage Measurement Of Resistograph

Age estimation of standing trees is the most important and basic common problem in forest production and research,which plays a guiding role in forest protection,forest management and ancient tree protection.The traditional methods of measuring tree age,Such as analytical wood disk method,growth cone method,are destructive.Under the background of ecological protection in China,micro-damage methods of standing tree age estimation are in desperate need,which can provide theoretical basis and scientific and technical support for natural forest protection and management of old and famous trees.Resistograph is a kind of Micro-damage tool which can be applied in measuring the age of standing trees.In view of the limitation of large error and over dependence on artificial experience in the tree age determination by Resistograph,this paper studied the algorithm of tree age estimation with the series of drilling resistance as the object,so as to provide the method and basis for the tree age determination.In this paper,Larix principis-rupprechtii,Korean larch,Pinus Koraiensis and Abies nephrolepis were used as the experimental objects to carry out in the field work,Based on the acupuncture sampling using Resistograph(4452p / s,4450 p / s)produced by RINNTECH company in Germany,the following research work had carried out:(1)An algorithm of measuring tree age by Resistograph based on stationary Kalman filter was proposed.According to the denoising principle of Kalman filter,the Kalman filter of Resistograph was designed and the stationary Kalman filter of Resistograph was proposed.The parameters of Resistograph Kalman filter were simplified,and the denoising effect was only related to parameter Rat.The new sequence was obtained from the drilling resistance value sequence which had been denoised by the Resistograph stationary Kalman filter.The tree age estimation was the peak and valley number of the new sequence,and only depended on the selection of the parameter Rat.(2)An algorithm of measuring tree age by Resistograph based on spectrum analysis was proposed.The application of discrete spectrum analysis in the drilling resistance value sequence of Resistograph was studied.After detrending the drilling resistance value sequence,the frequency spectrum was decomposed by discrete Fourier series,and the period number of harmonic with larger amplitude was determined as the estimated age of trees which only depended on the estimated value Selection of window parameter Wid.(3)An algorithm for the determination of tree age by Resistograph based on peak valley analysis was proposed,it is confirmed that there was an appropriate threshold parameter Det,which made the Resistograph to estimate tree age with high accuracy,and a method for prediction of peak valley analysis algorithm parameter Det based on BP neural network was proposed.In this paper,the above three algorithms was discussed and the following results were obtained:(1)When the appropriate parameter rat was selected according to their DBH size,the Resistograph stationary Kalman filter applied in 323 groups of drilling resistance of Larix principis-rupprechtii had better denoising effect,so that the algorithm estimating tree age closed to the actual age.The average relative error of tree age automatic interpretated by Resistograph software was-40.49%;The average relative error distribution of the algorithm was reduced to 0.75%,Through paired data t-test,it was found that t-value was-0.4682,showing the reliability of the algorithm.(2)The algorithm was applied to Larix principis-rupprechtii,the window parameters were selected according to the DBH size of living trees,and the algorithm estimating tree age was closed to the actual age.the average relative error was-0.35%.Through paired data t-test,it was found that t-value was 0.85.The results showed that the accuracy of spectrum analysis algorithm for estimating the age of Larix principis rupprechtii was better,and the accuracy of tree age estimation by Resistograph was improved.(3)Peak valley analysis algorithm was applied to 113 groups of drilling resistance sequences obtained from three tree species in natural forest including Larix olgensis,Pinus Koraiensis and Abies nephrolepis,When the appropriate threshold value(Det)was selected according to the tree real age,The tree age was estimated by the peak-valley analysis,and the results showed that it was close to the real tree age,with mean relative error as-2.69%,Paired data t-test was carried out with t-value of 1.31,showing that the algorithm was feasible to estimate the age of natural forest trees,and there was an appropriate threshold parameter Det,which made the estimation of tree age more accurate.(4)The nonlinear mapping performance of BP neural network was used to fit the threshold parameters of peak and valley analysis algorithm.Ten features that affected the selection of threshold value were found from the original resistance value sequence as the network input,and the threshold parameter Det of peak and valley analysis algorithm was used as the expected output of the network to construct the sample training set,established BP neural network,and predicted the accuracy of randomly divided test set.It can be used to predict the threshold parameters of peak valley analysis of Larix olgensis,Pinus Koraiensis and Abies nephrolepis.Above all,three kinds of algorithms for measuring the living trees age,including the stationary Kalman filter,frequency spectrum analysis and peak valley analysis,were proposed in this paper,which could be used to estimate the age of living trees by the series of drilling resistance values.The experimental results showed that: according to the tree diameter distribution,the algorithm parameters Rat and window parameters Wid were determined,and the estimation of the age of Larix principis rupprechtii by using the algorithm of stationary Kalman filter and the algorithm of spectral analysis were better,in which the error of the algorithm of spectral analysis was slightly lower than the algorithm of stationary Kalman filter;Using peak valley analysis algorithm to estimate the age of Korean larch,Pinus Koraiensis and Abies nephrolepis.was better,but BP neural network was needed to predict the threshold parameter Det.These three tree age estimation algorithms improved the accuracy of measuring the age of living trees by Resistograph,improved the shortcomings of its own software,promoted the research process of the micro-damage tree age determination.The next step was to apply the algorithm to the micro damage age estimation of more kinds of natural forest living trees.The key point of the research was to choose the algorithm parameters.