School of Science Department of Mathematics 44 Estimation of Wet Deposition Chemical Components in Northern Hemisphere by Using Deep-learning Technique Supervisor: FUNG, Jimmy Chi Hung / MATH Student: WAH, Monique Zoe / DSCT Course: UROP1100, Spring Atmospheric deposition, process of removal of gas and particulates in atmosphere, has been studied extensively. One major part of atmospheric deposition is wet deposition, which refers to the removal effect of rainfall. Raindrop collects material as it falls and deposits the material when reaching the ground (Viner, 2023). Despite its function in the reduction of particulate matter (Zanoletti et al., 2021), wet deposition of sulfate, calcium, nitrate, and ammonium can have negative impact on ecosystem, such as acidification and eutrophication (World Meteorological Organization, n.d.). Therefore, understanding the long-term trend of wet deposition concentration is crucial for environmental protection. This project aims to enhance the estimation of Northern hemisphere ammonium wet deposition using Convoluted Neural Network (CNN). Estimation of Wet Deposition Chemical Components in Northern Hemisphere by Using Deep-learning Technique Supervisor: FUNG, Jimmy Chi Hung / MATH Student: ZHANG, Jinming / DSCT Course: UROP1100, Fall UROP2100, Spring Wet deposition is an important environmental parameter as it plays a crucial role in removing harmful pollutants from the atmosphere and mitigating their impact on ecosystems and human health. However, excessive wet deposition of certain pollutants can also lead to environmental problems such as acidification of soils and bodies of water. Estimation of wet deposition compound, including non-sea-salt (nss)-sulfate (SO4 2-), nitrate (NO3 -), ammonium (NH4 +), and calcium (Ca2+), is essential for understanding the sources and impact of air pollution, and for guiding government regulations aimed at reducing its effects on the environment. Various methods have been used for estimating wet deposition, including direct sampling of precipitation, remote sensing, and modeling techniques. In this project, a convolutional neural network (CNN) model is developed to estimate the patterns of non-sea-salt (nss)-sulfate (SO4 2-), nitrate (NO3 -), ammonium (NH4 +), and calcium (Ca2+) deposition, and further studies are conducted based on the CNN model attempting to increase the model estimation accuracy.
RkJQdWJsaXNoZXIy NDk5Njg=