Response of dust particle pollution and construction of a leaf dust deposition prediction model based on leaf reflection spectrum characteristics

Abstract

Urban plants can improve several environmental pollution problems in cities, especially dust prevention, noise reduction, purification of the atmosphere, etc. To explore the influence of dust deposition on the spectral characteristics of the leaf, a foliar dust deposition prediction model based on high-spectrum data was established. Taking Euonymus japonicus L., the common greening tree species in Beijing, as the research object, high (T1), medium (T2), and low (T3) dust pollution gradients were set and hyperspectral data were collected. Results showed that: (1) in the dust-contaminated environment with different concentrations, the trend of the reflectance curve of the leaves of Euonymus japonicus L. was generally consistent. The spectral reflectance of the leaf surface was positively correlated with the amount of leaf dust. (2) There were five obvious reflection peaks and five main absorption valleys with the same positions and ranges in the 350–2500 nm range. (3) The spectral reflectance of leaf flour dust particles of Euonymus japonicus L. was significantly different before and after dusting, and its size was generally clean leaves > dust-depositing leaves. The sensitive range of its spectral response was 695–1400 nm. (4) The overall trend of the first derivative spectrum was basically the same. The red edge slope and the blue edge slope appeared as T3 > T2 > T1, the red edge position and the blue edge position appeared as T1 < T2 < T3. The red edge position of the leaf surface after dust deposition had an obvious blueshift , and the moving distance increases with the increase of dust retention on leaf surface. (5) The leaf water index (y = − 1.18x2 + 0.5424x + 0.9917, R2 = 0.8030, RMSE = 0.187) had the highest accuracy in the regression model of leaf surface dust deposition using spectral parameters. The test showed that the R2 reached 0.9019, which indicated that the model has a good fitting effect. This prediction model can effectively estimate the dust deposition of the leaf surface of Euonymus japonicus L.