Search

Effective monitoring and detection of changes in forest cover are paramount for environmental stewardship and sustainable management practices. However, existing methodologies face challenges in obtaining consistent and high-quality satellite imagery, predicting deforestation-prone areas, optimizing deep learning architectures for change detection, and establishing robust evaluation metrics. This paper makes a comprehensive literature review to examine the progress in deep learning segmentation techniques and their application in change detection using satellite imagery. The key questions guide the investigation, focusing on optimization of satellite image collection, development of predictive models for deforestation risk assessment, enhancement of U-Net architectures for change detection, and establishment of evaluation metrics. Through the review, significant advancements are identified, particularly in integration of attention mechanisms and modified U-Net architectures. Contrastingly, traditional image differencing methods are surpassed by convolutional neural network (CNN) approaches, showcasing superior feature extraction and change detection accuracy. Noteworthy innovations include residual learning, attention gates and spatiotemporal encoders, which enhance segmentation performance. The findings underscore the potential of advanced methodologies to enhance detection and monitoring of changes in forest cover. By drawing insights from related fields such as urban planning and disaster monitoring, the study advocates for a holistic approach to environmental management. Ultimately, this review contributes to the evolution of deep learning techniques for change detection, with a specific focus on advancing forest cover monitoring practices. [ABSTRACT FROM AUTHOR]

Urban forests globally face severe degradation due to human activities and natural disasters, making deforestation an urgent environmental challenge. Remote sensing technology and very-high-resolution (VHR) bitemporal satellite imagery enable change detection (CD) for monitoring forest changes. However, deep learning techniques for forest CD concatenate bitemporal images into a single input, limiting the extraction of informative deep features from individual raw images. Furthermore, they are developed for middle to low-resolution images focused on specific forests such as the Amazon or a single element in the urban environment. Therefore, in this study, we propose deep learning-based urban forest CD along with overall changes in the urban environment by using VHR bitemporal images. Two networks are used independently: DeepLabv3+ for generating binary forest cover masks, and a deeply supervised image fusion network (DSIFN) for the generation of a binary change mask. The results are concatenated for semantic CD focusing on forest cover changes. To carry out the experiments, full scene tests were performed using the VHR bitemporal imagery of three urban cities acquired via three different satellites. The findings reveal significant changes in forest covers alongside urban environmental changes. Based on the accuracy assessment, the networks used in the proposed study achieved the highest F1-score, kappa, IoU, and accuracy values compared with those using other techniques. This study contributes to monitoring the impacts of climate change, rapid urbanization, and natural disasters on urban environments especially urban forests, as well as relations between changes in urban environment and urban forests. [ABSTRACT FROM AUTHOR]

Abstract Spatial information on the timing of forest cover loss is important to identify and map stand age, which is a key factor driving the recovery of carbon pools and can also be used to estimate aboveground biomass (AGB) based on its relationship with stand age. Here, we estimated the spatial distribution of stand age and AGB of young forest (

Urban forests globally face severe degradation due to human activities and natural disasters, making deforestation an urgent environmental challenge. Remote sensing technology and very-high-resolution (VHR) bitemporal satellite imagery enable change detection (CD) for monitoring forest changes. However, deep learning techniques for forest CD concatenate bitemporal images into a single input, limiting the extraction of informative deep features from individual raw images. Furthermore, they are developed for middle to low-resolution images focused on specific forests such as the Amazon or a single element in the urban environment. Therefore, in this study, we propose deep learning-based urban forest CD along with overall changes in the urban environment by using VHR bitemporal images. Two networks are used independently: DeepLabv3+ for generating binary forest cover masks, and a deeply supervised image fusion network (DSIFN) for the generation of a binary change mask. The results are concatenated for semantic CD focusing on forest cover changes. To carry out the experiments, full scene tests were performed using the VHR bitemporal imagery of three urban cities acquired via three different satellites. The findings reveal significant changes in forest covers alongside urban environmental changes. Based on the accuracy assessment, the networks used in the proposed study achieved the highest F1-score, kappa, IoU, and accuracy values compared with those using other techniques. This study contributes to monitoring the impacts of climate change, rapid urbanization, and natural disasters on urban environments especially urban forests, as well as relations between changes in urban environment and urban forests.

Spatial information on the timing of forest cover loss is important to identify and map stand age, which is a key factor driving the recovery of carbon pools and can also be used to estimate aboveground biomass (AGB) based on its relationship with stand age. Here, we estimated the spatial distribution of stand age and AGB of young forest (<20 years) in three types of tropical dry forest in the Yucatan peninsula using Landsat NDVI (normalized difference vegetation index) time series from 2000 to 2020. We estimated AGB based on chronosequence data and compared these results to reference field data and estimations obtained from remote‐sensing studies. The overall and user accuracy of the age map was high (95.7–99.9% and 87.35–98.5% respectively). However, lower producer accuracy values (from 31.2 to 67.2%) suggest an underestimation of the extension of young forests. We found a greater extent of young forests in the semi‐deciduous and deciduous forests compared to the semi‐evergreen ones. Mean AGB estimated from stand age (53.1 Mg ha−1) was lower than that estimated from remote‐sensing studies (67.5 to 95.2 Mg ha−1). These results indicate that spatial information of forest age can be accurately assessed from Landsat time series, and that the combination of stand age with chronosequence data can reduce the overestimation of AGB of recovering forests commonly found in remotely sensed data. [ABSTRACT FROM AUTHOR]

This paper examines the possibilities, advantages and limitations of the use of high resolution archive satellite images in evaluation of forest and other land cover changes, based on research in the sample area of the Fruška Gora mountain (Serbia). Satellite images available from the declassified archives of CORONA program were used to assess the extent of forest cover in the past. By comparing the resulting datasets with newer images, changes in the forest coverage can be detected and reconstructed for a period of more than 40 years. Despite many limiting factors, the images provided valuable information about the state of the forest cover in the past. The methodology used can be utilized in other similar areas, where no other, more precise source is available about forest coverage, as a means of reasonably inexpensive and time efficient assessment of forest cover and other land use changes in the past five decades.