Extended Abstract
Introduction
The traditional method of chemical analysis has high accuracy and precision. However, it is time-consuming and laborious, and it is not possible to obtain continuous information about the pollutant status over a large area. Therefore, there is an urgent need for a reliable and environmentally friendly method to quickly identify and investigate the distribution of heavy metals in soil and thus identify suspected contaminated areas (Scheuber & Köhl, 2003:33). Remote sensing is one of the ways that can provide a cost-effective and quick solution to investigate the distribution of heavy metals on a large scale using spectroscopic techniques (Bi et al., 2009:16). Habibi et al. (2023:4) also measured and evaluated the concentration of heavy metals in the aerial parts and soil of the tree species of Bandar Abbas city and also identified the species that has the highest potential for absorbing heavy metals. The results showed that the pattern of heavy metals in soil and leaves of tree species was Mn>Zn>Pb>Cd. (Nikolaevich, 2023:30) they addressed the modeling of heavy metal pollution in Central Russia based on satellite images and machine learning. Al, Fe, and Sb contamination were predicted for 3000 and 12100 grid nodes in an area of 500 km2 for the Central Russian region for 2019 and 2020. Estimating the amount of this pollution requires time and high cost. Considering the traffic on the Aleshtar -Khorramabad highway near Kakareza forests and the effect of heavy metal concentration in the soil and leaves of the oak species which can be caused by natural and human pollution, the accumulation of heavy metals in the species Iranian oak is a serious threat to this forest. Therefore, it is necessary to study and discuss pollutants and their effects on the environmental cycle. In this regard, considering the cost and time-consuming nature of traditional methods and since remote sensing methods are a suitable complement to traditional methods; the aim of the present research is to use remote sensing techniques and spectral analyses to evaluate and model the accumulation of heavy metals in Iranian oak species.
Materials and Methods
The present study is located on the road of Aleshtar -Khorramabad, 20 kilometers northwest of Khorramabad. For this purpose, five transects were created at distances adjacent to the road, 500 and 1000 meters on both sides of the road, and 10 x 10 m sample pieces were planted. Inside the sample plots, 30 soil samples were randomly collected and 30 leaf samples were collected from trees in all directions of the crown. To extract heavy metals from soil samples and plant samples, the acid digestion method was used and the physical characteristics of the soil were measured using standard methods. After preparing the samples, the concentration of Pb, Cu, and zinc heavy metals in soil and leaves was measured and the index of biological concentration of heavy metals from soil to leaves was calculated. Then the relationship between the concentration of heavy elements measured and the reflectance in different bands or band ratios at the corresponding sampling points was obtained. Non-parametric methods and generalized multiple linear regression models were used in order to model quantitative variables and spectral values corresponding to sample parts in satellite data. ArcGIS software was used to implement sample parts on the image, ENVI software was used for image processing, and STATISTICA software was used for modeling.
Results and Discussion
Cu and Pb in Iranian oak leaves had significant differences at different distances at the 0.05 level, but Cu did not have significant differences at different distances at the 0.05 level. Cu and Pb did not have significant differences in different soil intervals at the 0.05 level, but Cu had significant differences in different soil intervals at the 0.05 level. The bioconcentration factor was obtained as (0.2, 0.5, 0.2) mg/kg. The study of modeling of non-parametric methods using Sentinel-2 satellite data showed that the highest explanatory coefficient values (0.85, 0.88, and 0.97) were obtained for the three metals Cu, Pb, and Cu, respectively. The artificial neural network (ANN) algorithm obtained the highest accuracy. Also, according to the results of the random forest algorithm, for the three mentioned metals, PSRI, HMSSI, and PSRI indices are the most important in modeling.
Based on the findings, the concentration values of Cu and zinc were significantly different at different distances, but the Cu values were not significantly different at different distances. In this regard, Mansour concluded in 2014 that there is a significant difference between the concentration of Cu and zinc in the leaves of the species, which can be attributed to traffic density and human activities, and the high amount of zinc metal in this study is the wear of car tires؛ and stated that the concentration of Cu is caused by the production of greenhouse gases and the use of vehicles using Cu gasoline. Based on the findings, the values of Cu and zinc concentrations at different distances did not have significant differences, but the Cu values had significant differences at different distances. Sources of input of Cu element to the soil are urban, industrial, and agricultural waste, fertilizers, and chemicals that add it to the soil through liquid, solid, or mineral fertilizers. These findings are with the results of some researchers including Wu and colleagues (2010:38), Botsou et al. (2016:17) are consistent. Based on the findings obtained from the calculation of the bioconcentration index and their comparison with the classification proposed by Ma et al. (2001:25) for Iranian oak species plants in relation to Cu, zinc, and Cu metals from soil to leaves, it acts as an accumulating plant. In accordance with the results of this research, in the study of Khodakarmi et al. (2009:15), Iranian oak was included in the category of superabsorbent plants in relation to the accumulation of Cu pollutants, which has a high capacity in terms of root absorption. Also, Madejón et al. (2006:25) stated that oak leaves are more resistant than olive leaves. The concentrations of elements in leaves and fruits decrease with time and the risk of toxicity in the food web is reduced. The review and comparison of five algorithms showed that (ANN) the highest explanatory coefficient values (0.85, 0.88, and 0.97) were obtained for three metals, Cu, Zn, and Cu, respectively. Considering the importance of the PSRI synthetic band in increasing the accuracy of modeling with satellite images and the influence of the visible and near-infrared bands, the amount of reflection measured by the spectroscopic method showed that with the increase in the concentration of heavy elements, the amount of reflection in the visible and infrared range decreases (Liu et al., 2011:24).
Conclusion
The results showed that Sentinel-2 images along with artificial intelligence techniques have a relatively good ability to model the level of biological pollution index in the region. In line with the obtained results, it is suggested that the Iranian oak species is used to reduce pollution on highways because it accumulates heavy metals.