Application Of Non-Destructive Testing And Machine Learning In The Study Of SO₂ Purification Capacity Of Urban Trees

Plant purification systems can effectively reduce the concentrations of major pollutants such as sulfur dioxide（SO₂）,play a significant role in absorbing SO₂,and improve air quality.The purpose of this study was to establish a rapid and non-destructive method for testing the SO₂ purification capacity of urban trees.This study focused on the application of non-destructive testing and machine learning to the SO₂ purification ability of urban trees in response to the problem of traditional plant monitoring methods that destroy plant samples.Non-destructive testing techniques,such as hyperspectral remote sensing and infrared thermal imaging,digital images,and microscopic images,were used to study the SO₂ purification capacity of urban trees in Changchun City,establish estimation models for the SO₂purification capacity of leaves of different greening trees and layers,and develop comprehensive and large-scale non-destructive testing technology for the SO₂ purification capacity of urban trees.Hence,the objective of non-destructive testing of the SO₂ purification capacity of urban trees in Changchun City at a regional scale was achieved.The main results are as follows:（1）Characterization of urban greening trees to SO₂ purification effect and non-destructive testing indicators during plant ontogenyThe ability of plants to absorb and purify SO₂ is related to the concentration of SO₂ in the atmosphere.The difference in non-destructive monitoring indicators among different tree species has reached a significant level,and different tree species have different tolerance abilities to the same type of pollutants.Regardless of the sampling date or SO₂ concentration,the green-peak reflectance and Fv/Fm of the upper leaves were significantly lower than those of the lower leaves.There is a high correlation between SO₂ purification rate and non-destructive testing indicators,and spectral reflectance is ultimately used as the optimal non-destructive testing indicator to evaluate SO₂ purification ability.（2）Non-destructive detection of vertical distribution of SO₂ purification capacity at the scale of urban green leavesThe SO₂ purification rate has varying degrees of correlation with different spectral transformation reflectivity,and its sensitive regions mainly exist in the green,red-edge and near-infrared region.Among the 16 combinations,the SG+CON+SD combination has the highest correlation coefficient between the predicted results and the measured values,and the evaluation effect is the best.Spectral transformation reflectance has great application potential in estimating the SO₂ purification ability of green plants.The sensitive bands of the optimal vegetation index and its band combination are different,but they all have a linear fitting relationship with the leaf SO₂ purification rate;By using the best index and AO-DELM to predict the scatter plot fitting results of SO₂ purification capacity,it can be seen that the overall predicted value is lower than the measured value,which can better predict the state of SO₂ purification capacity at different layers of urban green trees during critical periods.（3）Non-destructive detection of SO₂ purification capacity at the urban tree canopy scaleThe input parameter of the PROSAIL model is chlorophyll content,which is always lower in the upper leaves than in the lower leaves.The dry matter content and equivalent water thickness of the U.pumila are always lower than those of the S.oblate and P.cerasifera.From an overall perspective,the parameters with higher sensitivity indices to canopy spectral reflectance in the PROSAIL model are N,Cab,LAI,and Cw.The error between the simulated spectral values and the measured spectral values after optimizing the PROSAIL model parameters under different levels of SO₂ purification is small,and the simulated spectra are higher than the measured spectral;The optimal spectral index obtained from simulation has a good correlation with the purification ability of the corresponding green leaves,showing a negative correlation.Visualize the spatial distribution of SO₂ purification capacity of urban tree canopies in five sample areas of Changchun Industrial Park on August 6 and September21,2022.The estimated SO₂ purification capacity in the northwest region of the study area is weaker than that in the central region.（4）Non-destructive detection of SO₂ purification capacity at the urban tree and regional scalesAmong the 10 commonly used red-edge band normalization indices based on Sentinel-2multispectral red-edge bands,except for NDVIre2,NDVIre3,and NDVIre3n,there is a good correlation with SO₂ purification ability;The correlation coefficient between NDre2 and SO₂purification ability reached the highest among the 10 red-edge spectral indices,showing a significant negative correlation（-0.59）.In general,the SO₂ purification capacity of the five vegetation communities is as follows:deciduous broad-leaved forest>deciduous coniferous forest>evergreen coniferous forest>shrubland>mixed coniferous and broad-leaved forest.This study aims to address the lack of research regarding rapid,non-destructive,and low-input evaluation of plant purification capacity,and achieve non-destructive testing of plant purification capacity from point to surface,from static to dynamic,which can effectively,non-destructively,and inexpensively assess the level of vegetation air purification,providing a basis for predicting the amount of air purification in large areas in the future.The results can be extended to other types of air pollutants and regions in China,promote air pollution and management research,and have theoretical significance.