High levels of heavy metal contamination in soil present substantial threats to human health and the environment, leading to severe health problems such as neurotoxicity, cancer, kidney issues, chronic obstructive pulmonary disease, and reduced life expectancy. This research aims to identify and analyze heavy metals in soil samples collected from Superfund sites in North Birmingham, Alabama, specifically in affected areas with Zip codes 35207, 35217, and control area 35214. These affected areas were previously used for mining, coal-fired power plants, coke furnaces, smelting, and other potential sources of heavy metal pollution. Laser-Induced Breakdown Spectroscopy (LIBS) was employed to study 60 soil samples systematically collected from affected and control areas. We found that using LIBS, we could detect arsenic (As), lead (Pb), and manganese (Mn) in all soil samples from the affected areas. The limit of detection (LoD) was 16 ppm for Pb, 43 ppm for As, and 33 ppm for Mn using specific parameters of the detection system and/or argon gas purging at atmospheric pressure. The results were compared with ICP-MS measurements to validate the accuracy of the LIBS findings. Data showed a good linearity for all calibration data at relatively low concentrations and a good correlation between ICP-MS measurements.

Timely detection of surface damages on wind turbine blades is imperative for minimising downtime and avoiding possible catastrophic structural failures. With recent advances in drone technology, a large number of high-resolution images of wind turbines are routinely acquired and subsequently analysed by experts to identify imminent damages. Automated analysis of these inspection images with the help of machine learning algorithms can reduce the inspection cost, thereby reducing the overall maintenance cost arising from the manual labour involved. In this work, we develop a deep learning based automated damage suggestion system for subsequent analysis of drone inspection images. Experimental results demonstrate that the proposed approach could achieve almost human level precision in terms of suggested damage location and types on wind turbine blades. We further demonstrate that for relatively small training sets advanced data augmentation during deep learning training can better generalise the trained model providing a significant gain in precision.