• Korean Journal of Remote Sensing
• /
• v.39 no.6_1
• /
• pp.1413-1425
• /
• 2023

The increasing frequency of wildfires due to climate change is causing extreme loss of life and property. They cause loss of vegetation and affect ecosystem changes depending on their intensity and occurrence. Ecosystem changes, in turn, affect wildfire occurrence, causing secondary damage. Thus, accurate estimation of the areas affected by wildfires is fundamental. Satellite remote sensing is used for forest fire detection because it can rapidly acquire topographic and meteorological information about the affected area after forest fires. In addition, deep learning algorithms such as convolutional neural networks (CNN) and transformer models show high performance for more accurate monitoring of fire-burnt regions. To date, the application of deep learning models has been limited, and there is a scarcity of reports providing quantitative performance evaluations for practical field utilization. Hence, this study emphasizes a comparative analysis, exploring performance enhancements achieved through both model selection and data design. This study examined deep learning models for detecting wildfire-damaged areas using Landsat 8 satellite images in California. Also, we conducted a comprehensive comparison and analysis of the detection performance of multiple models, such as U-Net and High-Resolution Network-Object Contextual Representation (HRNet-OCR). Wildfire-related spectral indices such as normalized difference vegetation index (NDVI) and normalized burn ratio (NBR) were used as input channels for the deep learning models to reflect the degree of vegetation cover and surface moisture content. As a result, the mean intersection over union (mIoU) was 0.831 for U-Net and 0.848 for HRNet-OCR, showing high segmentation performance. The inclusion of spectral indices alongside the base wavelength bands resulted in increased metric values for all combinations, affirming that the augmentation of input data with spectral indices contributes to the refinement of pixels. This study can be applied to other satellite images to build a recovery strategy for fire-burnt areas.

• Economic and Environmental Geology
• /
• v.30 no.2
• /
• pp.137-142
• /
• 1997

At the tailings dam of the disused Brukunga pyrite mine in South Australia, reaction of groundwater with the tailings causes the formation and discharge of sulphuric acid. There is a need to improve remediation efforts by decreasing groundwater flow through the tailings dam. Geophysical methods have been investigated to determine whether they can be used to characterise variations in depth to watertable and map preferred groundwater flow paths. Three methods were used: transient electromagnetic (TEM) soundings, direct current (DC) soundings and profiling, and self potential (SP) profiling. The profiling methods were used to map the areal extent of a given response, while soundings was used to determine the variation in response with depth. The results of the geophysical surveys show that the voltages measured with SP profiling are small and it is hard to determine any preferred channels of groundwater flow from SP data alone. Results obtained from TEM and DC soundings, show that the DC method is useful for determining layer boundaries at shallow depths (less than about 10 m), while the TEM method can resolve deeper structures. Joint use of TEM and DC data gives a more complete and accurate geoelectric section. The TEM and DC measurements have enabled accurate determination of depth to groundwater. For soundings centred at piezometers, this depth is consistent with the measured watertable level in the corresponding piezometer. A map of the watertable level produced from all the TEM and DC soundings at the site shows that the shallowest level is at a depth of about 1 m, and occurs at the southeast of the site, while the deepest watertable level (about 17 m) occurs at the northwest part of the site. The results indicate that a possible source of groundwater occurs at the southeast area of the dam, and the aquifer thickness varies between 6 and 13 m. A map of the variation of resistivity of the aquifer has also been produced from the TEM and DC data. This map shows that the least resistive (i.e., most conductive) section of the aquifer occurs in the northeast of the site, while the most resistive part of the aquifer occurs in the southeast. These results are interpreted to indicate a source of fresh (resistive) groundwater in the southeast of the site, with a possible further source of conductive groundwater in the northeast.