Satellite-based prediction of surface dust mass concentration in southeastern Iran using an intelligent approach

Abstract

The southeastern section of Iran, especially the province of Khuzestan, experience severe air pollution levels, such as high values of Surface Dust Mass Concentration (SDMC). The province lacks accurate and well-placed ground observational stations, therefore the only viable approach for evaluating SDMC is via remote sensing. In this study, meteorological, hydrological and geological data on 11 input variables from Modern-Era Retrospective analysis for Research and Applications Version (MERRA-2), Global Precipitation Measurement (GPM) and Global Land Data Assimilation System (GLDAS) for the year 2018 are used for prediction of the SDMC values, also obtained from MERRA-2. For real-time prediction, Pearson's Correlation Coefficient (PCC) analysis shows that wind-related variablessurface wind speed, surface aerodynamic conductivity and surface pressureare those with the highest correlation with SDMC. Using the Gradient Boosting Regression (GBR) algorithm, these three variables can simulate SDMC with good accuracy (𝐶𝐶 = 0.815 , 𝑁 − 𝑅𝑀𝑆𝐸 = 0.605). Future forecasting of SDMC requires knowledge of both wind-related and heat-related variables. However, SDMC predictions can be obtained with the GBR algorithm with adequate accuracy (𝐶𝐶 = 0.640 . 𝑁𝑅𝑀𝑆𝐸 = 0.781) by just considering the surface pressure value observed four days before the forecasted day. This study shows that robust predictions of SDMC can be obtained using exclusively remote sensing data, without ground-based observations.

Long-Term Water Quality Monitoring: Using Satellite Images for Temporal and Spatial Monitoring of Thermal Pollution in Water Resources

“…The considered remote sensing sources have been demonstrated to be useful tools to detect and quantify temperature changes in rivers, and the same goes for numerous water and environmental bodies (such as aquifers, wetlands, agricultural plots, etc.) that usually depict the same limitation to apply remote sensing data: the reduced spatial size of these elements, which encompass low pixels from the mentioned satellite sources [56][57][58][59][60]. Therefore, studying, investigating, and presenting appropriate solutions in this area can be very useful and practical to the scientific community and to local authorities.…”

Groundwater quality assessment for sustainable human consumption in arid areas based on GIS and water quality index in the watershed of Ain Sefra (SW of Algeria)

“…spatial variation maps were generated (Achilleos 2011;Maleika 2020). The IDW has employed a deterministic model method, which calculates unknown values based on the closeness of adjacent points (Asadollah et al 2023). This IDW approach is used to characterise pollutant distribution and to find physicochemical patterns in space and time (Kanagaraj and Elango 2016).…”

“…This coefficient underlines how much each parameter matters and its effect on the hydrochemical process in the case of the Ain Sefra watershed. If (R) values of the Pearson's correlation matrix are close to + 1 or − 1, they are strong correlation coefficients and represent total correlation, or functional dependence, between two variables (Asadollah et al 2023;de Vente et al 2011). If the values are close to zero, no significant interaction exists between the two variables.…”

Intelligence framework dust forecasting using regression algorithms models

“…The program fits a decision tree-based weak regression model to the training data. The residuals, the disparities between actual dust levels and current ensemble predictions, are then calculated [20]. New weak regression models are trained to forecast residuals in subsequent iterations.…”