No abstract

No abstract

This study aims to develop an integrated approach of deep learning convolutional neural network (DL-CNN) and Google Earth Engine (GEE) platform for salt storm (ST) modeling and monitoring. First, we selected three ST’s predisposing factors, including Land Surface Temperature (LST), Soil Salinity (SS), and Normalized Difference Vegetation Index (NDVI) to train models. We then collected 957 Ground Control Points (GCPs) from the study area, which randomly divided into training (70%) and validation (30%) datasets. Finally, ReLu, Cross-Entropy, and Adam employed as activation function, loss function and optimizer, respectively. Our findings demonstrate the efficiency of an integrated DL-CNN and GEE for monitoring ST (Overall Accuracy (OA) = 0.93.02, 0.92.99, 0.93.88, and 0.92.01 for years 2002, 2010, 2015 and 2021, respectively). The results also show an increase in the volume of ST in the study area from 2002 to 2021. Such approach is a promising step toward understanding, controlling, and managing ST and recommend for ST spatial monitoring in other favored areas with similar environmental conditions.

Forest fire in recent years has given much attention to climate change and ecosystem. Detection of fire in Near real time is necessary to prevent large-scale casualties. Remote sensing is a quick and inexpensive way to detect and monitor forest fires on a large scale. The purpose of this research is to identify forest and rangeland fire using MODIS and AVHRR sensors in Kayamaki Wildlife Refuge. In order to carry out research, the dates of the fire occurred at MODIS products were recorded. Then the images of both sensors were prepared based on the date of the fires. After preprocessing the images, different fire detection algorithms (i.e. IGBP, Giglio, Extended, Dynamic for NOAA/AVHRR images and Giglio, Extended for MODIS /AQUA &TERRA images) for fire detection were investigated. The results of fire detection algorithms with MODIS products were there by evaluated. The results showed that among different algorithms, the dynamic algorithm on NOAA/AVHRR images is more suitable than the other ones with a low error rate of 28% for fire detection. Although the IGBP algorithm has a lower error rate relative to the dynamic algorithm, based on the characteristics (using NDVI, the use of two images (D1, D2) to detect fire in Near real time, the thresholds for removal of wrong alarms, etc.) has a dynamic algorithm, but the IGBP algorithm shows incorrect alarms despite the low error rate and the Giglio algorithm on MODIS/TERRA images were determined as favorable algorithm with 28% error. This is confirmed in the fires that were detected in terms of location relative to the fire locations found in the MODIS products. Also, MODIS and AVHRR sensors were compared in terms of Near real-time fire detection ability. AVHRR detected the highest number of fires (68.25%) with low error rate (31.75%) and MODIS discovered lower fire numbers (46.5%) and high error rate (53.5%).