Fire is a natural disruption that affects the structure and function of forest systems by changing the vegetation composition, climatic situation, carbon cycle, wildlife habitat, and many other major properties. The measure of these changes’ degree is known as fire severity, and it can be assessed using remote sensing data (i.e., satellite images, aerial images, etc.) and various biophysical indices (such as Normalized Burn Ratio (NBR), Char Soil Index (CSI), Burn Area Index (BAI), etc.), in addition to the measurement of Land Surface Temperature (LST). This research aims to assess the response of NBR and LST in pre- and post-forest fire, taking as a study area, a Mediterranean forest located in the northern part of Morocco (35.1167° N, 5.7754° W), which burned in the summer of 2022. We used seven Landsat-8 images spanning three years: three images from 2021 (i.e., pre-fire), one image from the summer of 2022 (i.e., fire period), and three images from 2023 (i.e., post-fire). Results demonstrated a negative correlation between LST and NBR in the pre-fire period; when the temperature rises, the NBR drops. Same for the fire period in summer 2022, LST reached its peak at 50°C, while NBR decreased to its lowest point at -0.2. Whereas, in the recovery time (i.e., 2023), LST and NBR changed their fluctuation patterns; the first one variated normally according to seasons, dropping from the 50°C to 12°C in winter and reaching 37°C in summer, and the second one increased over time, going from the -0.2 to -0.04 in winter rising to 0.03 in summer, which indicates the gradual restoration of vegetation in the study area. The study concludes that in the post-fire period when the forest is recovering, NBR is unaffected by seasonal changes in temperature and is more reflective of the vegetation it projects more the vegetation situation in the area, unlike LST. Thus, relying only on LST to measure fire severity can give biased results due to changes in seasons.

The Recovery Ice Stream is one of the longest ice streams in Antarctica, annually discharging a significant mass of ice into the Southern Ocean. Beneath the Recovery Ice Stream are numerous active subglacial lakes whose drainage and storage directly impact the flow velocity of the entire ice stream. This, in turn, has a considerable influence on ice dynamics, grounding line stability, and the mass balance of the East Antarctic Ice Sheet. Approximately twenty years ago, scientists discovered that the water transfer movements within subglacial lakes caused surface deformation on the ice sheet. The latest NASA Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2), utilizing laser altimetry technology, can capture more dense and precise spatial details, helping us better understand this hydrological process. Based on this, we use all the ICESat-2 data from September 2018 to July 2022 to reconstruct and analyze the activity of subglacial lakes under the Recovery Ice Stream. To investigate water transfer between the subglacial lakes and the latest subglacial lake outlines, we calculate the differential between measurement points and the reference Digital Elevation Model (DEM) to depict the surface elevation changes of each active subglacial lake in monthly time steps. The new lake outlines are defined as contour lines representing the average elevation changes of the static ice sheet. After obtaining the lake outlines, we further analyze the crossover tracks to generate higher temporal resolution elevation change time series for the regions of interest. We have observed differences in the location and volume of the subglacial lake signals compared to the previously published inventory. Firstly, we discover that Lake REC1, originally considered as one lake, is composed of two distinct lakes during the study period, displaying opposing elevation change trends in repeated orbits. While the left area of the REC1 lake experienced a rise of 1m, the right area showed a decrease of approximately 0.5m. Additionally, through calculations of temporal elevation changes, REC1, REC2, and REC3 exhibited characteristics of cascading responses from upstream to downstream. The upstream lake, REC6, initially drained and has been continuously refilling since late 2019, resulting in a surface elevation change of approximately 4m and consuming nearly 0.4 km³ of subglacial water. This substantial water supply has effectively lubricated the ice-bedrock interface, facilitating the fast flow of the Recovery ice stream. Finally, we estimated the hydraulic head of the lakes and predicted water flow paths that align with the sequence of lake activity depicted in the time series plot. In conclusion, the subglacial lakes within the Recovery ice stream constitute a well-connected hydrological system, and the hydrological dynamics in this region are closely associated with the unique subglacial topography of the Recovery area.

Acid sulfate soil mapping is the first step to avoid possible environmental damages created by one of the most problematic soils existing in nature: the acid sulfate soils. This type of soil is especially hazardous when it is drained by agricultural or forestry land use. Nowadays, more objective and precise maps are possible thanks to the application of machine learning. The use of a supervised machine learning technique in acid sulfate soil mapping requires two different types of data: the soil samples and the environmental covariates created by remote sensing data. One of the problems in acid sulfate soil mapping is the lack of soil samples in some regions since the collection of soil samples and their analysis is a long process. This prevents the creation of acid sulfate soils occurrence maps. For a first recognition of these regions, in addition to using the remote sensing data of the area, a possible solution could be the use of soil samples from other areas with similar characteristics for training the model. The question is whether a machine learning model could correctly classify acid sulfate soils in an area where it has not been trained. If this were possible, this first prediction could be used to design an efficient sampling plan for the region. In previous works, Random Forest has shown high abilities for the correct prediction of acid sulfate soils. In this work, we analyze if Random Forest is able to correctly classify the soil samples in an area where it has not been trained. For this, two different regions located in southern Finland with a similar composition of their soils are considered. It is known that remote sensing data play a fundamental role in the detection of acid sulfate soils. In this study, the remote sensing data used are LiDAR and geophysics, which arise from airborne surveys. The raster data of both areas consist of 17 environmental covariates of different types: Quaternary geology, digital elevation model, terrain layers and aerogeophysics layers. Digital elevation model is made using LiDAR data, and the terrain layers are derived from the digital elevation model. In this work, we show that Random Forest is able to classify the acid sulfate soils of an area where it has not been trained. The precision of the model is above 60%. These results are very good for a model that has not been trained in the area of the prediction. Training the model in the same area improves the results by up to 10-13%. Therefore, training the model in a different region can be used for a first recognition of regions with limited soil samples as well as for the creation of the sampling plan design in those regions.

As urban areas continue to expand, there is an increasing focus on enhancing the comfort of outdoor thermal conditions.

In this study, summer discomfort index (SDI) maps were created for Seville, Barcelona, in Spain and Tetuan in Morocco .
Both temperature and humidity, which are crucial components in determining thermal comfort and discomfort, are taken into account by SDI.

The calculations used substituted air temperature with land surface temperature data from the Moderate Resolution Imaging Spectroradiometer (MODIS) and humidity from weather stations, which were then compared to thermal sensation votes (TSV) gathered from surveys given to the residents.
Land surface temperatures are normally higher than air temperatures, but the resulting maps offered a close representation of thermal comfort.

The goal was to evaluate the thermal comfort levels in the chosen cities and investigate the relationship between the remotely sensed (SDI) and the reported thermal perception by the residents. We aimed to gain insights into urban thermal environments and their effects on human perception by integrating remote sensing data and subjective (TSV).

The visual maps offer an easily readable representation of thermal comfort and discomfort and can assist designers in creating better outdoor spaces that are tailored to the needs and comfort levels of residents in each unique city.

The method involved gathering and examining MODIS land surface temperature data, processing it, and calculating each city's (SDI) values. Votes on thermal comfort (TSV) were collected through a seven scale questionnaire-based survey and represented residents' individual experiences and perceptions.

The findings provide valuable insights into the thermal conditions and comfort levels experienced by residents during summer in Seville, Barcelona, and Tetuan. Remote sensing data enabled the creation of spatially explicit (SDI) maps, facilitating a detailed assessment of thermal comfort variations within and between the studied cities. Comparing the remotely sensed (SDI) with subjective (TSV) contributes to a comprehensive understanding of agreement or divergence between objective measurements and human perception.

By highlighting the importance of integrating remote sensing techniques and subjective assessments for evaluating thermal comfort in urban areas, this research advances the field of urban climate studies and its results have implications for urban planning, design, and the development of strategies to enhance thermal conditions and well-being of city residents.

Parameter Efficient Fine Tuning (PEFT) techniques have recently experienced significant growth and have been extensively employed to adapt large vision and language models to various domains, enabling satisfactory model performance with minimal computational needs. However, despite these advancements, more research has yet to delve into potential PEFT applications in real-life scenarios, particularly in the critical domains of remote sensing and crop monitoring.

In the realm of crop monitoring, a key challenge persists in addressing the intricacies of cross-region and cross-year crop type recognition. The diversity of climates across different regions and the need for comprehensive, large-scale datasets have posed significant obstacles in accurately identifying crop types across varying geographic locations and changing growing seasons. This study seeks to bridge this gap by comprehensively exploring the feasibility of cross-area and cross-year out-of-distribution generalization using the State-of-the-Art (SOTA) wheat crop monitoring model.

This research mainly focuses on adapting the SOTA TSViT model, recently proposed in CVPR 2023, to address winter-wheat field segmentation, a critical task for crop monitoring and food security, especially following the Ukrainian conflict, given the economic importance of wheat as a staple and cash crop in various regions. This adaptation process involves integrating different PEFT techniques, including BigFit, LoRA, Adaptformer, and prompt tuning, each designed to streamline the fine-tuning process and ensure efficient parameter utilization.
By employing PEFT techniques, we achieved notable results comparable to those attained through Full Fine-Tuning methods while training only a mere 0.7% parameters of the whole TSViT architecture. More importantly, we accomplished the claimed performance using a limited subset of remotely labeled data. The in-house labeled dataset, referred to as the Lebanese Wheat dataset, comprises high-quality annotated polygons for wheat and non-wheat classes for the study area in Bekaa, Lebanon, with a total surface of 170 km², over five consecutive years from 2016 until 2020. Using a time series of multi-spectral Sentinel-2 images, our model achieved an 84% F1-score when evaluated on the test set, shedding light on the capacity of PEFT to drive accurate and efficient crop monitoring, tailored mainly for developing countries characterized by limited data availability.

We intend to publicly release the Lebanese winter wheat dataset, code repository, and model weights.