Glacier surface facies (GSF) are the visible glaciological regions which can be distinguished and mapped at the end of summer using optical satellite data. GSF maps act as visual metrics of glacier health when assessed independently. Spatial distribution of all accumulation and ablation facies are an important input to 3D mass balance models. This is principally because although the two broad categories of ablation and accumulation can be mapped without division into surface facies, it is the spatiotemporal change of specific GSF that enables better modelling. For example, the progressive increase in area and distribution of melting ice and decrease in area and distribution of glacier ice may signal potential mass loss without significant change in overall area of the ablation zone. As glaciers in Svalbard are warming at a significantly higher rate than the global average, tracking the evolution of GSF is important for predictive assessment for the cryosphere in the Arctic. This will further facilitate robust methods for monitoring GSF on a planetary scale. In this context, we present a local scale spatiotemporal analysis of GSF of Midtre Lovénbreen, Svalbard. We used openly available Landsat 8 OLI and Sentinel 2A imagery from 2017-2022, to track the occurrence and variations in GSF via machine learning (ML). Current results suggest that ablation facies such as melting ice and dirty ice are increasing over time. Sentinel 2A provides finer resolution but is limited by its temporal coverage. Although Landsat is suitable for long-term trend analysis, its coarser resolution can lead to errors such as over/underestimation of smaller patches of facies on relatively smaller glaciers. As the spectral properties of GSF are consistent over time, a robust set of spectra depicting variations in physical appearance of facies may be used to train ML algorithms, thereby improving efficacy. In forthcoming studies, our objective is to expand the temporal scope spanning decades and to trace facies evolution over longer time series. This endeavor seeks to establish a robust GSF inventory for Svalbard.

Photon transport within Earth's atmosphere is a critical phenomenon in atmospheric science, remote sensing, and climate modeling. Accurately modeling the behavior of photons as they interact with atmospheric components like gases and aerosols is crucial for quantitative analysis of the remote sensing data. However, these simulations are computationally intensive due to the complexity of the problem, especially in the case of two- and three dimensional models. A common way to tackle the problem of multi-dimensional photon transport is Monte-Carlo simulations, which require considering a huge amount of photons to obtain a reliable statistics. This problem can be efficiently parallelized to reduce the computation time and exploit the multi-core CPU and GPU devices. This research conducts a comparative analysis of three simulation techniques: traditional loops, NumPy's vectorized operations, and CuPy's GPU acceleration. The study implements a Monte Carlo simulation method in a inhomogeneous atmosphere.

Conventional approaches to simulating photon transport involve iterative loops, leading to computational bottlenecks. In this study we examine the efficiency of three approaches that can accelerate computations. The first approach is based on a multiprocessing library that essentially deals with multithreading. The second method is based on the code vectorization and using matrix notations though the NumPy library. In this case it is possible to get rid of the loop across the photons and to track an ensemble of photons in a single batch. The simulations are performed in the paradigm “single instruction – multiple data”. The third method is a natural extension of the previous method. It is based on the GPU-accelerated library CuPy. By harnessing GPU parallelism, CuPy can significantly accelerate simulations, making them feasible for large-scale scenarios. In order to consider stopping criteria for a batch of photons, a sort of masking is applied to an ensemble of photons. Metrics such as runtime, memory usage, and scalability are evaluated to quantify the performance trade-offs. It is shown that the initial structure of the python code should not be significantly changed when using aforementioned approaches. Moreover, CuPy library allows an easy transition of the model from CPU to GPU. Experiments are performed on two cores of Intel Xeon(R) CPU @ 2.30GHz and GPU Tesla T4. Three orders of magnitude performance enhancement is achieved by using vectorization. Using GPU provided additional 10x speedup as compared to the vectorized version for 10⁵ photons.

The study demonstrates the superiority of CuPy for photon transport simulations. CuPy's GPU acceleration substantially reduces simulation time if the number of photons is sufficiently large (>10⁷), otherwise the overhead of CuPy is comparable with the actual computing time).

The calculation of the R-factor (Rainfall erosivity) for implementation in soil erosion models such as USLE (Universal Soil Loss Equation) and RUSLE (Revised Universal Soil Loss Equation) encounters substantial difficulties due to the scarcity of spatial databases in adequate resolution for actions of territorial planning at the local level. Otherwise, there is a spatial database available with a coarse resolution of themes that can be used to calculate the R-factor. We apply the spatial downscaling, based on regression models: linear (LN), general additive model (GAM), random forest (RF), cubist (CU), on erosivity data (target variable) prepared for the State of São Paulo, Brazil, with a spatial resolution of 2,500 m. We used DEM and slope data with 30 m fine-resolution from the Atibaia watershed, located between the metropolitan regions of São Paulo (RMSP) and Campinas (RMC) to apply the downscaling. This framework improved the spatial resolution of the R-factor, necessary to calculate soil loss in the USLE and RUSLE equations in a territory where the scarcity of data with the fine resolution is still limited to the development of territorial planning projects at the local level. The RF model was better with R² 0.93.

Permafrost, defined as perpetually frozen soil, characterizes most of the Arctic terrestrial surface and it is overlayed by an active layer that melts seasonally. The global warming is triggering an earlier snow melting and a related reduction in average snow cover extent in the spring with a consequent lengthening of the Snow-Off period and exposure of land surface to the effect of solar radiation. The spatial pattern of the Land Surface Temperature (LST) and its general increases, change with a different rate according to the vegetation surface types and the associated mid-summer albedo. The immediate consequence can be observed in the annual variability of permafrost thawing and in the direct impact on the tundra biome in terms of greening process and above all, on the permafrost itself, which stores tons of carbon. The permafrost Active Layer Thickness (ALT) has reached an ever-increasing annual average. The geospatial model here presented estimates the permafrost ALT over the entire Arctic in the last 20 years and it is based on the spatial and temporal oscillations measured by satellite based Essential Variables (EVs) associated with the Thermal State of Permafrost (TSP). The model integrates the climate components such as the LST and the mid-summer Albedo, with the structural and functional descriptors of Arctic tundra biome such as the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR). The Arctic tundra region has been divided in two subzones: the first characterized by less structured vegetation, mainly mosses and lichens, the second by grass and shrubs. The ALT biogeographical variability is obtained by building a linear regression model between the EVs and the ALT data provided by the Circumpolar Active Layer Monitoring (CALM) field measurements for each subzone. For the mosses and lichens subzone, ALT average has been estimated to increase of 5 cm in twenty years, instead for the grass and shrubs subzone, the estimated ALT average has been increased of 2 cm. Although a general average ALT increase has been estimated over the whole tundra region, with rates up to 2 cm/year, the areas which encounter the highest thickening rates, partially overlap with the areas where a vegetation persistence and a potential greening phenomenon have been estimated to occur, along the boreal tree line.

The current stage of development of geological exploration and reproduction of the mineral resource in Russia is focused on the discovery of new ore deposits. In mountainous regions in a variety of landscapes, secondary geochemical halos and dispersion flows of ore deposits are clearly manifested on the modern earth surface. An industrial deposit lying in the thickness of ore-bearing rocks, or the smallest manifestation of a mineral of any mineral raw material, being a geochemical anomaly in itself, forms lithochemical anomalies in loose weathering products. Taking into account one of the main limitations of the use of remote sensing methods in geology, namely the depth of surface exploration, the proposed method is reduced to revealing the manifestation of secondary scattering halos on the day surface, i.e. mapping zones of hydrothermally altered rocks and products of hypergenesis, identification of iron oxides/ hydroxides in soils and rocks, in particular.

The geological information content of the WorldView-2 data lies in the high spectral resolution and the presence of VNIR bands, which allow displaying detailed spectral characteristics of surface objects, in particular, minerals of the iron oxide/hydroxide group containing transitional iron ions (Fe³⁺ and Fe³⁺/Fe²⁺) that are part of the zones wall-ore hydrothermally altered rocks. In accordance with the absorption spectral features of a group of minerals (hematite, magnetite, goethite, ilmenite, jarosite, limonite) containing iron oxides and hydroxides, the spectral band ratio technology (mineralogical indices) was used for WorldView-2 VNIR channels. The mineralogical index (b3*b4)/(b2*1000) (Segal, 1982) was used to map Fe³⁺/Fe²⁺; for ferric iron mapping, the mineralogical index (b4 + b2)/b3 (Pour, 2019) was used; mineralogical index (b6 + b8)/b7 (Pour, 2019) was used to map ferrous iron.

Then a pseudo-color RGB composite, which displays the classes of geological materials that have spectral characteristics associated with iron oxides/hydroxides was created. The R band corresponds to Fe³⁺, the G band corresponds to Fe³⁺/Fe²⁺, and the B band corresponds to Fe²⁺. Thus, this color combination well emphasizes the geostructural characteristics of rocks associated with hydrothermal changes. According to the authors, the selected pseudo-color RGB composite is the most informative for mineralogical mapping of the study area.

According to the proposed WorldView-2 data processing method and the selected RGB pseudo-color composite, the resulting image is presented as a minerals map of the probability distribution of oxides/hydroxides containing transitional iron ions (Fe²⁺, Fe³⁺ and Fe³⁺/Fe²⁺), which assigns a mineral to each pixel, conditional the probability of occurrence of which at a given point is maximum. The revealed spectral anomaly corresponds to the supposed dispersion haloes of the products of metasomatism and hypergenesis.

The results obtained using modern methods of processing remote sensing data allow us to consider the spectral anomalies of the zones marking near-ore changes in rocks as an indicator for substantiating the choice of areas for detailed exploratory studies within ore clusters. The results of such studies can significantly reduce the cost at different stages of geological exploration.

ACKNOWLEDGMENTS

The authors are grateful to PROXIMA (www.gisproxima.ru) for providing WorldView-2 images.

SOURCE OF FINANCING

The work was carried out within the framework of the state task of IGEM RAS.

REFERENCES

Pour A.B., Hashim M., Hong J.K., Park Y. Lithological and alteration mineral mapping in poorly exposed lithologies using Landsat-8 and ASTER satellite data: North-eastern Graham Land, Antarctic Peninsula // Ore Geol. Rev. 2019. 108. 112–133. https://doi.org/10.1016/j.oregeorev.2017.07.018

Pour A.B., Park Y., Crispini L., Laufer A., Kuk Hong J., Park T.-Y.S., Zoheir B., Pradhan B., Muslim A.M., Hossain M.S. et al. Mapping Listvenite Occurrences in the Damage Zones of Northern Victoria Land, Antarctica Using ASTER Satellite Remote Sensing Data // Remote Sens. 2019. 11. 1408. https://doi.org/10.3390/rs11121408

Segal D. Theoretical Basis for Differentiation of Ferric-Iron Bearing Minerals, Using Landsat MSS Data / Proceedings of Symposium for Remote Sensing of Environment, 2nd Thematic Conference on Remote Sensing for Exploratory Geology, Fort Worth, TX (1982). 949–951.

Satellite remote sensing has become an important tool in the monitoring and assessing the impacts of drought. Drought is a widespread slow-onset natural hazard that causes significant damage to agricultural production and natural ecosystems. Estimating drought onset precisely is challenging, however, its impacts manifest in vegetation greenness and crop water requirements over time scales. Drought affects water and energy cycle components, shifting the balance between water supply and water demand. Evapotranspiration (ET) is one of the most significant parts of the hydrologic budget that leads to water loss from the processes of evaporation and transpiration and plays a critical role in drought assessment. According to climate models, global temperature is expected to rise, influencing trends, intensity, duration, and impacts of drought. Mediterranean agriculture is vulnerable to droughts as crop production is largely water-limited and projected to accelerate with climate change. Hence, it has become critical to monitor and evaluate drought in time series where water is limited. Measuring ET directly for a specific point and time is expensive and time consuming. However, remote sensing data can be used to overcome the limitations by providing us with continuous data over extended periods. In this study, an optimized modelling framework to estimate time series of ET at regional scale in the Iberian Peninsula using MODIS data from 2000 to 2022 and a Two-Source Energy Balance model to estimate surface energy balance of the soil-canopy-atmosphere continuum is presented. This modelling framework, based on the SEN-ET scheme, synergistically uses Terra and Aqua MODIS data (LST, LAI, water vapor, aerosols and reflectance products) and ERA5 atmospheric reanalysis dataset to estimate ET at 1 km spatial resolution. Although model evaluation with several flux towers and forested watersheds is still ongoing, preliminary results at a pistachio orchard in Lleida (NE Iberian Peninsula) show a RMSE ranging from 50 to 80 W·m^-2 for summer 2022.

The use of drones to gather remote data and soil sensors to collect ground information has become a powerful method for agricultural monitoring and analysis. However, integrating data from drone remote sensing and soil sensors in agricultural contexts can be problematic due to variations in spatial and temporal resolutions. Ensuring precise synchronization and calibration is crucial for accurate comparative analysis. The objective of this study was to investigate the strengths and limitations of both approaches and explore the potential for data fusion. Through a series of field trials, data from drone-based remote sensing and ground-based soil sensing were collected in parallel. This data encompassed a range of factors, including vegetation health (vegetation indices), soil properties such as EC, pH, and optical measurements. The study delves into the challenges of data synchronization, calibration, and validation between the two methodologies. We discuss the potential for synergy in building a more holistic understanding of agriculture by fusing data from drones and in situ soil sensors. The findings of this research have implications for environmental monitoring, agriculture, and ecosystem management, suggesting that the combination of aerial and ground sensing offers a multi-dimensional perspective that can enhance decision-making processes and our grasp of intricate environmental processes.

The accurate estimation of Land Surface Temperature (LST) plays a vital role in various fields, including hydrology, meteorology, and surface energy balance analysis. This study focuses on the estimation of LST using data acquired from the Joint Polar-orbiting Satellite System (JPSS) missions, specifically JPSS-1/NOAA-20 and JPSS-2/NOAA-21. The methodology for this research centers on the utilization of the split window algorithm, a well-established and recognized technique renowned for its proficiency in extracting accurate Land Surface Temperature (LST) values from remotely sensed data. This algorithm leverages the differential behavior of thermal infrared (TIR) radiance measured in two adjacent spectral channels to estimate LST, effectively mitigating the influence of atmospheric distortions on the acquired measurements.

To establish the accuracy of the proposed approach, the coefficients of the split window algorithm were determined through linear regression analysis, utilizing a dataset generated via extensive radiative transfer modeling. The calculated LST values were subsequently compared with LST products provided by the National Oceanic and Atmospheric Administration (NOAA). The evaluation process encompassed the computation of root mean square error (RMSE) values, offering insights into the performance of the algorithm for both JPSS-1/NOAA-20 and JPSS-2/NOAA-21 missions.

The obtained results demonstrate the potential of the split window algorithm to effectively estimate LST from JPSS satellite data. The RMSE values, 1.4 and 1.5 for JPSS-1/NOAA-20 and JPSS-2/NOAA-21, respectively, highlight the algorithm's capability to provide accurate LST estimates for different mission datasets. This research contributes to enhancing our understanding of land surface temperature dynamics using remote sensing technology and showcases the valuable insights that can be gained from JPSS missions in monitoring and studying Earth's surface processes.

This paper presents a meticulous exploration of advanced machine learning techniques for precise forest type classification using multi-temporal remote sensing data within a woodland environment. The study comprehensively evaluates a diverse range of models, spanning from advanced ensemble learning methods to several finely tuned support vector machine (SVM) variants, with a specific focus on Bayesian-optimized SVM with radial basis function (RBF) kernel. Our findings highlight the robust performance of the Bayesian-optimized SVM, achieving a high accuracy of up to 94.27%, and average precision and recall of 94.46% and 94.27% respectively. Notably, this accuracy aligns with the levels attained by acclaimed ensemble techniques such as Random Forest and CatBoost while also surpassing those of XGBoost and LightGBM. These results highlight the potential of these methodologies to significantly enhance forest type mapping accuracy compared to tradition (Linear) SVM and black-box neural networks. This, in turn, can enable reliable identification and quantification of key services, including carbon storage and erosion protection, intrinsic to the forest ecosystem. Finding of our comparative study emphasizes the profound impact of employing and fine-tuning advanced machine learning approaches in the realm of remote sensing-based environmental analysis.

Landslides are natural disasters with a high socio-economic impact on human societies due to the considerable number of fatalities and the destruction of infrastructure that they cause. A comprehensive landslides inventory is vital for reducing this impact, as it can be used in landslide susceptibility studies for the identification of the most critical to landslide subregions of an area, for the evaluation of the landslide precipitation activation thresholds, and subsequently for the determination of the most suitable precautionary measures. Nowadays, remote sensing techniques are widely used by scientists for creating landslides inventories, as they can be rapidly applied to identify landslides along with their spatial characteristics. Nevertheless, besides these characteristics, a comprehensive inventory must also include the time of their activation and the factors that led to their activation. These elements can be quite difficult to specify, especially in areas where official landslide data do not exist, such as in countries that do not have a published national landslides inventory. The objective of this research study is to provide a framework for the creation of a comprehensive landslides inventory by combining open access or publicly available data with remote-sensing data and techniques. The Chania regional unit in the western part of Crete Island, Greece, was selected as the study area. Our study presents how a complete landslides inventory, consisting of more than 150 landslides, was established based on differential interferometry synthetic aperture radar (DInSAR) techniques and open access or publicly available data. This framework can significantly contribute to scientific research on landslide susceptibility in countries that lack a comprehensive landslides inventory. Moreover, it highlights the potential of remote-sensing techniques and open access data in improving our understanding of landslide activation mechanism.