Detection of Root Rot in Living Trees

The goal of the project is to detect root rot in living trees. Today only subjective methods exist for quality control of forest areas and individually growing trees. The proposed method analyses the multichannel response of a blow of a hammer. Frequencies and mode shapes of the damped part have been estimated using a multi-channel Prony method. The resonance frequencies from the multi-channel Prony method and the propagation velocity of a surface wave are used in a detector for evaluating 93 trees. An optimal method based on the 2-dimensional ESPRIT has also been invented for estimation of the resonance frequencies and mode shapes. The coherence or the cross-spectrum is a measure of the signal dependence between two signals, ideally for each time point and each frequency value, where the mutual time- as well as frequency information of the signal performance is combined, see example in the figure to the left, where the individual time-frequency spectra from the existing data-base where a channel close to the excitation point at 45 degrees and a channel far from the excitation point at 225 degrees are depicted. All these different types of spectra call for robust estimation methods with appropriate time-frequency resolution. In this case Peak Matched Multiple Windows (PM MW) are used.