The undeniable importance of forests in the sustainability of the planet Earth and its impact on the climate change increases interest in monitoring these ecosystems. Within the parameters that support the management of forests there is the forest height (FH), which not only provides information by itself, but also can be used as a proxy of other parameters, as the biomass. Multispectral observations jointly with machine learning regressors have been used to produce FH maps, however most of them were fitted using large calibration sets covering the area being mapped. In this article, a FH mapping model optimized to be calibrated with sparse and limited calibration sets is proposed. To achieve this objective, Sentinel-2 and ancillary data is used, with robust feature generation and processing steps, together with a stacking regressor. For the calibration and posterior testing, Light Detection and Ranging (LiDAR) data acquired during Airborne Laser Scanning campaigns was used. The approach was tested in 16 study areas covered by Mediterranean forests, achieving a RMSE/rRMSE in the range 1.50-6.06m/13.58-27.15%, respectively, although it only resorted to calibration sets that only covered 25% of the study areas (decreasing the amount of LiDAR data needed for calibrating FH regressors).

The near-worldwide coverage of spaceborne LiDAR data from the GEDI offers unprecedented opportunities for mapping canopy height (CH). Notwithstanding the sensitivity of the GEDI to forests’ vertical structure, it provides sparse sampling measurements, which hinder gap-free mapping. Several machine and deep learning models that resort to optical, radar, and GEDI have been tested to produce gap-free CH maps. Not all factors affecting the accuracy and consistency of these GEDI-fused products have been explored. Specifically, the sampling fraction coverage and the geolocation correction of footprints on-orbit positions have not been deeply studied. In this article, a collection of data from 15 study areas, characterised by Mediterranean landscapes, had their CH mapped resorting to Sentinel-1/2, ALOS-2, ancillary data, and a locally fitted extreme gradient boosting regressor. The produced maps had an average %RMSE of 32.41%, outperforming the other two global products in Mediterranean regions. Additionally, the geolocation correction of the GEDI footprints was limited to 1.23 percentage points. This experiment was able to demonstrate three key points: (1) the importance of including InSAR in the optical/SAR synergy; (2) the reduced impact of collocating GEDI footprints at the track level; and (3) the benefits of increasing GEDI-footprint coverage using multi-year data are hindered by temporal variation effects.

The importance of mapping the forest height (FH) is increasing due to the more frequent impacts of climate change in the society (wildfires, droughts, and extreme weather events). Remote Sensing is often used for mapping this variable; however, it usually relies in costly and extensive field or airborne campaigns. In addition, when using synthetic aperture radar (SAR), most approaches do not use freely available data. Considering this, in this work a model is proposed that resorts to Advanced Land Observing Satellite 2 (ALOS-2), Sentinel-1 (S1), and ancillary data. Airborne laser scanning (ALS) data are used for local calibration but, with the aim of developing a more scalable model, the latter is optimized to work with small calibration datasets (representative of just 25% of the study area to be mapped). With this purpose, the model combines a featuring generation and a features' processing stage with a stacking regressor to produce estimates at the pixel level. Their impact was assessed, and an improvement of 8.11 and 2.01 pp in the relative root mean square error (rRMSE) was achieved by including the features' generation and features' processing stages, respectively. In addition, when the multifrequency dataset was used, the model achieved an rRMSE better than when using only a C-band dataset (S1) or only an L-band dataset (ALOS-2), respectively, by 4.21 and 3.05 pp. Finally, the model achieved an average ${R} <^>{2}$ /rRMSE of 0.6240%/24.30% and 0.5901%/22.64% for the validation and test study areas, respectively. The proposed approach revealed to be effective on mapping the FH resorting to multifrequency SAR and small calibration datasets acquired by ALS.

Estimating Forest biomass stocks is crucial to monitor carbon cycle dynamics, which supports a better understanding and reporting on the climate change effects. Despite the work already done on mapping the aboveground biomass, other biomass pools have been neglected. This article proposes methodology capable of mapping the Belowground Biomass Density (BGBD) Litter Biomass Density (LBD) and Shrub Biomass Density (SBD) combining multispectral, Synthetic Aperture Radar (SAR), and ancillary products and resorting to an extreme gradient boosting regressor. The regressor was locally calibrated using datasets covering less than 4% of the areas and wall-wall ALS-based BGDB, SBD and LBD maps for training. The proposed mapped methodology achieved an %RMSE of 56.13% for BGBD, 27.07% for LBD, and 37.40% for SBD. Additionally, the maps were produced for six study areas with an average %RMSE of 71.50%, 31.09% and 38.65% when compared to the ALS-based maps, respectively for the BGBD, LBD and SBD.

Biomass is a relevant indicator of how much carbon can be retained by the forest, and it can be divided into four different pools: Aboveground Biomass Density (AGBD), Belowground Biomass Density (BGBD), Litter Biomass Density (LBD), and Shrub Biomass Density (SBD). Besides the AGBD, which is extensively studied, the other biomass pools are often overlooked, and that may result in evaluations of forest carbon storage that are either insufficient or erroneous. This article presents a methodology for estimating AGBD, BGBD, LBD, and SBD using various machine learning (ML) regression techniques, leveraging data from the Global Ecosystem Dynamics Investigation (GEDI) mission. Biomass was estimated across multiple study areas in Portugal and Spain. In Portugal, the average %RMSE values for AGBD, BGBD, SBD, and LBD were 61.95%, 60.84%, 36.93%, and 35.14%, respectively. For Spain the corresponding values were 63.07%, 53.24%, 28.30%, and 37.67%.

Using images from the Solar Dynamics Observatory (SDO)/Atmospheric Imaging Assembly (AIA) instrument, this project employs computer vision techniques to study solar differential rotation. Coronal Bright Points (CBPs), serve as reliable tracers across multiple images due to their widespread distribution across latitudes, independence from the solar cycle, and distinct shape. Using an image matching algorithm, these CBPs were detected and tracked, demonstrating consistent performance throughout various solar cycle stages. Solar rotation values were calculated based on the movements of CBPs during their lifetime as a result of the performed tracking. A dedicated website was created in order to make the results available to anyone interested in this topic, providing near real-time information. Performance metrics of the tool, including its effectiveness during both low and high solar activity phases are showcased, demonstrating its utility across different solar conditions, as well as the impact of the detection performance on the speed rotation results and a comparison between the proposed detection method and Gradient Path Labelling (GPL).

In the current domain of image super-resolution (SR), particularly concerning satellite imagery processing based on mainstream deep learning methodologies, most algorithms typically employ loss functions such as L 1 loss (Mean Absolute Error) or L 2 loss (Mean Squared Error) based on pixel value differences when training network models. However, when magnifying the details of the resulting images, there is often a blurring effect at the edges where different ground conditions, such as city roads, buildings, and various terrains, intersect, making it difficult to distinguish the edges of these different objects. On the other hand, in our previous experiments, using the Perceptual Loss function (based on calculating perceptual error of images) for training models yielded images with improved visual quality, allowing for better object distinction. Nevertheless, although the object edges did not appear blurred, the transitions at the edges were somewhat abrupt and distorted. Therefore, in this paper, a composite loss function that combines L 1 loss and Perceptual Loss is proposed, aiming to leverage their advantages to enhance the visual quality of objects in satellite images while avoiding edge blurring and achieving higher object discrimination. Additionally, we continue to explore the optimization of the SGNET (Sentinel-2 Google-Earth-Pro Network) architecture to improve the image super-resolution results.

One of the most important forests’ characteristics is its height, which can give valuable knowledge for different purposes, such as its management, wildfire prevention, carbon stock estimation, or even helping in obtaining other indicators. Since forests are normally spread over extensive areas, it can be expensive and time-consuming to map all areas with precise measurements taken directly on the ground or airborne. A solution that has been studied over the last years is using satellite imagery to help map forests’ height. The purpose of this paper is to present how Synthetic Aperture Radar (SAR) data, particularly from Sentinel-1, can improve the forest height estimation with Machine Learning (ML) regressors. First, an analysis of SAR’s Single Look Complex (SLC) data was performed to test how this data can provide similar results to Ground Range Detected (GRD) data. Then an analysis of how the results with GRD data can be improved with the use of Land Cover Land Use (LCLU) maps, specifically on the speckle filters usually applied. Finally, a combination of SLC and GRD data was tested, showing an improvement in the overall results of ML regressors, with a Stacking Regressor averaging a R 2 of 70.39% and a relative RMSE of 21.44%. All the tests were performed on six regions of Portugal (and a few tests on ten more similar regions, from Spain and California), with data from six months in 2020.

The impacts of the climate change in the society make forest monitoring increasingly important. Consequently, there is a growing interest in mapping variables as the Forest Height (FH). The direct measurement of the FH through field campaigns is expensive and difficult to scale. Alternatively, Airborne Laser Scanning (ALS) campaigns can be used to map it, however they share the same disadvantages of the previous approach. Therefore, Remote Sensing (RS) data have been used for local and large-scale mapping of the FH. In this paper a Regression Methodology (RM) that combines GEDI, ALOS-2, Sentinel-1/2, and ancillary data is proposed for mapping the FH in Mediterranean forests. The proposed RM, tested for the 15 regions of interest, achieves a RMSE/rRMSE of 4.95m/33.93%, when evaluated with GEDI data, and 5.11m/41.70%, when evaluated with ALS data.

Remote Sensing has been widely used for monitoring forests, namely for the retrieval of structural parameters such as the Forest Height (FH). The reason behind the use of remote sensing is the fact that measuring the FH through field campaigns is expensive and non-scalable. The resort to Airborne Laser Scanning campaigns, despite its high accuracy, have the same limitations. Therefore, Synthetic Aperture Radar (SAR) and Multispectral sensors carried by spaceborne platforms are widely used to address this problem. This paper evaluates the effects of combining a dataset that includes multifrequency backscatter (L and C bands) and multispectral variables, with Interferometric SAR (InSAR) variables (Coherence and Phase) for FH mapping resorting to a locally calibrated regression methodology. To make it more suitable for operational scenarios, only free access data is used, and the calibration sets are small. The scope of this study is the Mediterranean forests, and it has achieved a R2/RMSE ranging from 50.33–72.01

Recent advancements in satellite image super-resolution (SR) have been driven by neural network-based methods, emphasizing the reconstruction of high-frequency information such as object edges and texture details. These high-frequency elements are critical for enhancing image clarity and overall visual quality. Mainstream approaches typically involve deep neural network architectures and altering loss function formulations to improve high-frequency detail retention. However, as resolution magnification increases substantially (e.g., by a factor of five or more), reconstructing the high-frequency information of object edges becomes increasingly challenging, often resulting in edge blurring. To address this issue and to enhance edge definition, this study integrates an edge detection block into the SGNET network. It uses a Laplacian of Gaussian operator to detect initial SR image edges, process these through a streamlined convolutional network to reconstruct high-frequency edge information, and subsequently add these enhanced edges to the initial SR image, boosting edge contrast and clarity in the final output. This method effectively mitigates edge blurring, demonstrating significant improvements in visual metrics such as SSIM (Structural Similarity Index Measure) and LPIPS (Learned Perceptual Image Patch Similarity), which collectively enhance the overall visual quality of the SR images.

Remote Sensing has been widely used for monitoring forests, namely for the retrieval of structural parameters such as the Forest Height (FH). The reason behind the use of remote sensing is the fact that measuring the FH through field campaigns is expensive and non-scalable. The resort to Airborne Laser Scanning campaigns, despite its high accuracy, have the same limitations. Therefore, Synthetic Aperture Radar (SAR) sensors carried by spaceborne platforms are widely used to address this problem. The aim of this paper is to assess the effects of combining multiband backscatter datasets (L and C bands) with Interferometric SAR (InSAR) variables (Coherence and Phase) for FH mapping resorting to a locally calibrated regression methodology. To make it more suitable for operational scenarios, only free access data is used, and the calibration sets are small. The scope of this study is the Mediterranean forests, and it has achieved a R2/RMSE ranging from 42.78-65.61%/1.68-2.68m in the validation and 47.23-64.38%/0.90-2.56m for the operational scenarios. The addition of the InSAR variables leads to an improvement of 3.44% in the R2 and 0.07m in the RMSE.

To effectively manage the terrestrial firefighting fleet in a forest fire scenario, namely, to optimize its displacement in the field, it is crucial to have a well-structured and accurate mapping of rural roads. The landscape’s complexity, mainly due to severe shadows cast by the wild vegetation and trees, makes it challenging to extract rural roads based on processing aerial or satellite images, leading to heterogeneous results. This article proposes a method to improve the automatic detection of rural roads and the extraction of their centerlines from aerial images. This method has two main stages: (i) the use of a deep learning model (DeepLabV3+) for predicting rural road segments; (ii) an optimization strategy to improve the connections between predicted rural road segments, followed by a morphological approach to extract the rural road centerlines using thinning algorithms, such as those proposed by Zhang–Suen and Guo–Hall. After completing these two stages, the proposed method automatically detected and extracted rural road centerlines from complex rural environments. This is useful for developing real-time mapping applications.

Forest monitoring is gaining new importance with the increasing number of events related to climate change (as wildfires). Therefore, mapping the Forest Height (FH) becomes an important activity in the forest management when preparing for the fire seasons and for an improved understanding of climate change. The FH can be used directly, or as a proxy of other variables, as the aboveground biomass. The most accurate way to measure this variable is through field campaigns or airborne laser scanning, however both approaches are expensive and have limitations in terms of spatial and temporal scalability. As an alternative, other Remote Sensing sensors can be used, such as Synthetic Aperture Radar (SAR) or Multispectral scanner. When using SAR data, the commonest approach is to estimate the FH through SAR interferometry, a technique that usually relies in data that is not freely available, making it less suitable for operational scenarios. Also, most of the approaches based on SAR or Multispectral data need large datasets for calibrating the algorithms. In this paper, a Regression Methodology (RM) that resorts to multifrequency SAR, from Sentinel-1 and ALOS-2, and Multispectral data, from Sentinel-2, is proposed for the generation of FH maps of Mediterranean forests. The RM uses a Stacking Regressor, that can generate FH maps, calibrated with data covering only 25% of the study area being mapped. A R2 between 50.79-78.01% and a RMSE between 0.76-3.68m were achieved on a total of 17 study areas across Portugal, Spain, and USA.

In recent years, the application of neural networks in the field of satellite image super-resolution has become increasingly widespread. This paper aims to explore how to utilize images from two completely different data sources (Sentinel-2 and Google Earth Pro) as training data to surpass the spatial resolution limitation of 10 meters provided by the Sentinel-2 satellite and obtain higher resolution image. The Sentinel-2 satellite offers open data with four bands with spatial resolution of 10 meters and a resampling cycle every 5 days. This provides abundant data that can be used as Low-Resolution (LR) images for research. Adopting a multi-image super-resolution processing approach allows for the full utilization of geographical information contained within different images of the same location captured at shorter time intervals, while also eliminating concerns regarding information loss due to weather-affected low-quality images. Google Earth Pro provides images with a resolution of up to 0.15 meters, allowing the High-Resolution (HR) image resolution to be adjusted according to the needs. In our experiments, we proposed Sentinel-2 Google Earth Pro Network (SGNET). We used HR images with a resolution of 2 meters, and through SGNET implementation, a five-fold increase in spatial resolution for Sentinel-2 images was achieved, yielding very satisfactory results.

Canopy Height (CH) is an important variable in any forest inventory, not only by its own information, but also as a proxy variable to estimate other parameters as the above-ground biomass. The CH information can also be helpful to understand the climate change trends, for forest management, and in decision support systems related to wildfires. The growing availability of Remote Sensing observations acquired from different sensors, create an alternative for the CH mapping to field campaigns and Airborne Laser Scanning (ALS) missions. Here a comparison between using Multispectral and Synthetic Aperture Radar sensors for CH estimation is presented. Both used the same Regression Methodology, being achieved a R 2 /RMSE between 43.71%-72.85%/0.85-4.03m for Multispectral and 42.12%-62.62%/0.96m-4.49m for SAR, for a total of 17 regions of interest. It is concluded that Multispectral data revealed to be more suitable for the CH mapping.

The extent and impact of wildfires are expected to increase as a consequence of climate changes, pushing forest and firefighting management decision-making process to be supported by Cyber-Physical-Systems (CPS) that are capable to promote collaborative decision processes. New IoT tools and frameworks fuse scientific knowledge and diversity of sensor data will contribute to improve strategic resources allocation, with the aim of protecting lives, assets, and the environment. The presented IoT CPS which is tunned as a decision support system (DSS) is adapted for the Portuguese wildfire context. It is composed by a geographic information system (GIS) online framework, a mobile client application and a set of portable multi-sensor devices.

Mediterranean Europe is strongly affected by wildfires. In Portugal, the Portuguese Institute for Nature Conservation and Forests (ICNF) implemented the national fuel break (FB) network responsible for fire control and suppression. FBs are regions where vegetation is reduced to break up the fuel continuity and create pathways for the firefighting vehicles. The efficiency of this strategy relies on the correct implementation of FBs and on periodic fuel treatments. Multispectral imagery from Sentinel-2 (with high temporal and spatial resolution) facilitates the monitoring of FBs and the implementation of methodologies for their management. In this paper a two stages methodology is proposed for monitoring FBs. The first stage consists in detecting fuel treatments in FBs, to understand if those were correctly executed. This is done through a change detection methodology with resource to an Artificial Neural Network. The second stage monitors the vegetation recovery after a fuel treatment, to aid the scheduling of new treatments, ensuring the efficiency of FBs during the fire season. Both methodologies resort to reflectance bands and spectral indices from Sentinel-2; and timeseries and objects, exploiting the temporal and spatial information. The two stages were tested in different regions across the Portuguese territory, demonstrating their usability for all the national fuel break network. The detection of treatments achieved a relative error lower than 4%, and the vegetation recovery cycle estimated by the second stage match the expectations from ICNF.

The increase of the world population has had an impact in the agricultural field. As a consequence, this implies an increase in food production. To address this demand, farmers had to boost crop yields and land sizes. The latter one led to ineffectiveness of traditional methods for crop monitoring. For this reason, farmers began to adopt and adapt technological breakthroughs into agriculture by applying the concept of remote sensing. Remote sensing aims to collect information at a distance through the use of cameras and, in some cases, aerial platforms. Furthermore, multispectral cameras allowed farmers to better understand the management of crops and crop's health. Also, through study of literature, it was found that thermal imaging could be an important tool to measure the crop's condition. However, due to complexity of thermal images, an open-source tool that integrated this functionality was not found. Therefore, this paper proposes an open-source method that addresses the complexities of thermal images and is able to produce maps by exploiting them.

The increased development of camera resolution, processing power, and aerial platforms helped to create more cost-efficient approaches to capture and generate point clouds to assist in scientific fields. The continuous development of methods to produce three-dimensional models based on two-dimensional images such as Structure from Motion (SfM) and Multi-View Stereopsis (MVS) allowed to improve the resolution of the produced models by a significant amount. By taking inspiration from the free and accessible workflow made available by OpenDroneMap, a detailed analysis of the processes is displayed in this paper. As of the writing of this paper, no literature was found that described in detail the necessary steps and processes that would allow the creation of digital models in two or three dimensions based on aerial images. With this, and based on the workflow of OpenDroneMap, a detailed study was performed. The digital model reconstruction process takes the initial aerial images obtained from the field survey and passes them through a series of stages. From each stage, a product is acquired and used for the following stage, for example, at the end of the initial stage a sparse reconstruction is produced, obtained by extracting features of the images and matching them, which is used in the following step, to increase its resolution. Additionally, from the analysis of the workflow, adaptations were made to the standard workflow in order to increase the compatibility of the developed system to different types of image sets. Particularly, adaptations focused on thermal imagery were made. Due to the low presence of strong features and therefore difficulty to match features across thermal images, a modification was implemented, so thermal models could be produced alongside the already implemented processes for multispectral and RGB image sets.