Sentinel-1 Mission Research Articles

Turbidity classification of the Paraopeba River using machine learning and Sentinel-2 images

The collapse of Dam I, owned by Vale S.A, in Brumadinho-MG (Brazil), among other serious socio-environmental consequences, contaminated the waters of the Paraopeba River in a stretch of hundreds of kilometers. Considering the relevance of monitoring water quality, and knowing that field evaluation is a time-consuming and costly procedure, the use of satellite images, widely available at low cost, emerges as a relevant alternative. This work proposes a systematic experimental evaluation of five machine learning methods - Extra Trees, Multilayer Perceptron, Naive Bayes, Random Forest and Support Vector Machine, under different configurations and input data treatments, to classify the turbidity of the Paraopeba River waters from Sentinel-2 mission images. In a classification setup defined from Brazilian legislation turbidity classes, all methods obtained results equal to or greater than 0.87 accuracy, with appropriate settings and data treatments. The best result was obtained using the Support Vector Machine classifier with hyperparameters adjustment by random search, input data processed by the Yeo-Johnson transformation and selection of spectral bands by collinearity analysis. In this case, the accuracy was 0.96 with two classes, and 0.91 with 3 classes, indicating the feasibility of using this method to classify turbidity.

Top-of-atmosphere hyper and multispectral signatures of submerged plastic litter with changing water clarity and depth.

The exploitation of satellite remote sensing is expected to be a critical asset in monitoring floating and submerged plastic litter in all aquatic environments. However, robust retrieval algorithms still havel to be developed based on a full understanding of light interaction with plastic litter and the other optically active constituents of the atmosphere-water system. To this end, we performed laboratory-based hyperspectral reflectance measurements of submerged macroplastics under varying water clarity conditions (clear - 0 mg/L, moderate - 75 mg/L, very turbid - 321.3 mg/L) and submersion depths. This comprehensive optical dataset was used (i) to relate the plastic-related signal to submersion depth and turbidity parameters, and (ii) to investigate the top-of-atmosphere signal through full radiative transfer calculations. Simulated TOA radiation was used to explore the nominal pixel and spectral requirements based on WorldView-3, Sentinel-2, and Sentinel-3 missions with very high to moderate geo-spatial resolutions. Results showed that plastics remained detectable when submerged in the top ∼1 m of the water column regardless of water clarity conditions. At TOA, uncertainties attached to atmospheric correction were shown to be reasonable and acceptable for plastic detection purposes in the infrared part of the spectrum (> 700 nm). The impact of aerosols on the TOA signal was found to be complex as (i) over large plastic patches. The aerosols produced little impact on satellite observations mostly due to adjacency effects and (ii) optical signature from isolated/small extent plastic patches was critically altered suggesting the atmospheric transmittance should be accurately corrected for in plastic detection algorithms. The sensitivity analyses also revealed that the narrow band widths of Sentinel-3 did not improve detection performance compared to the WorldView-3 coarser band widths. It is proposed that high spatial resolution wavebands such as the pan-chromatic could be advantageously explored for submerged plastic monitoring applications.

A bibliometric analysis on the visibility of the Sentinel-1 mission in the scientific literature

Seven years after the launch of the first Sentinel-1 satellite, its data have been widely used in the scientific community. This study provides the first quantitative analysis of the visibility of the Sentinel-1 mission to the scientific literature through a bibliometric analysis of 1628 articles published in scientific journals during the 2014–2020 period. The main findings show that the number of Sentinel-1 mission-related papers increased significantly over the years, with an annual growth rate of 83%. Remote sensing is the most popular journal where 31.75% of the publication collection has been published. China and the USA are the two most productive countries with a share of 22.30% and 16.22% in the collection. Research based on the Sentinel-1 data covered a wide range of topics in geoscience disciplines. The use of SAR interferometry, focusing on the studies of landslide, earthquake, ground deformation, and subsidence, is the most important research direction using Sentinel-1 data. Image fusion of Sentinel-1 and Sentinel-2 observations for mapping and monitoring applications is the second most important research direction. Other popular research areas are glaciology, soil moisture, agriculture, rice monitoring, and ship detection. This study uses bibliographic data derived only from the Scopus database; therefore, it might not cover all Sentinel-1 related documents. However, this paper is a good reference for researchers who want to use Sentinel-1 data in their studies. The two Sentinel-1 satellites will provide scientific data for years to come, meaning that this type of analysis should be done on a regular basis.

Multi-technique geodetic detection of onshore and offshore subsidence along the Upper Adriatic Sea coasts

• InSAR-GNSS-Levelling data cross-validation procedure. • Unprecedented 20-years window estimation of onshore deformation along the Upper Adriatic Sea coastal areas. • Cross-validated deformation detected at an offshore platform for hydrocarbon extraction activity. We assess about 20 years of onshore and offshore subsidence along a sector of the Upper Adriatic Sea (Italy) coastal areas affected by natural soil compaction and intense anthropogenic activities such as aquifers exploitation and hydrocarbons extraction. Our approach is based on the synergistic use of independent remote sensing and in-situ geodetic data to detect and spatially characterise the deformation pattern by cross-validating the different available measurements. We collect extensive datasets from i) SAR images provided by Envisat, Cosmo-SkyMed and Sentinel-1 missions, ii) GNSS measurements from continuous stations managed by public institutions, local authorities and private companies and iii) Leveling surveys. The cross-validation analysis shows good agreement among all the independent datasets, thus providing a reliable assessment of the ongoing deformation. We detect an onshore and offshore subsidence peak of about −1/-1.5 cm/yr in the proximity of the coastline, close to Lido di Dante and Fiumi Uniti villages, and at the present offshore platform. The outcomes highlight how the integration of different remote sensing and in situ geodetic techniques is successful to retrieve deformation history in time and space in complex areas, where different natural and anthropogenic sources concur to the overall deformation pattern. Moreover, such approach provides a robust support to modelling studies for hazard assessment in both inland and shoreline areas.

Cloud Mask Intercomparison eXercise (CMIX): An evaluation of cloud masking algorithms for Landsat 8 and Sentinel-2

Cloud cover is a major limiting factor in exploiting time-series data acquired by optical spaceborne remote sensing sensors. Multiple methods have been developed to address the problem of cloud detection in satellite imagery and a number of cloud masking algorithms have been developed for optical sensors but very few studies have carried out quantitative intercomparison of state-of-the-art methods in this domain. This paper summarizes results of the first Cloud Masking Intercomparison eXercise (CMIX) conducted within the Committee Earth Observation Satellites (CEOS) Working Group on Calibration & Validation (WGCV). CEOS is the forum for space agency coordination and cooperation on Earth observations, with activities organized under working groups. CMIX, as one such activity, is an international collaborative effort aimed at intercomparing cloud detection algorithms for moderate-spatial resolution (10–30 m) spaceborne optical sensors. The focus of CMIX is on open and free imagery acquired by the Landsat 8 (NASA/USGS) and Sentinel-2 (ESA) missions. Ten algorithms developed by nine teams from fourteen different organizations representing universities, research centers and industry, as well as space agencies (CNES, ESA, DLR, and NASA), are evaluated within the CMIX. Those algorithms vary in their approach and concepts utilized which were based on various spectral properties, spatial and temporal features, as well as machine learning methods. Algorithm outputs are evaluated against existing reference cloud mask datasets. Those datasets vary in sampling methods, geographical distribution, sample unit (points, polygons, full image labels), and generation approaches (experts, machine learning, sky images). Overall, the performance of algorithms varied depending on the reference dataset, which can be attributed to differences in how the reference datasets were produced. The algorithms were in good agreement for thick cloud detection, which were opaque and had lower uncertainties in their identification, in contrast to thin/semi-transparent clouds detection. Not only did CMIX allow identification of strengths and weaknesses of existing algorithms and potential areas of improvements, but also the problems associated with the existing reference datasets. The paper concludes with recommendations on generating new reference datasets, metrics, and an analysis framework to be further exploited and additional input datasets to be considered by future CMIX activities.

Field-level crop yield estimation with PRISMA and Sentinel-2

Satellite image data deliver consistent and frequent information for crop yield estimation over large areas. Hyperspectral narrowbands are more sensitive spectrally to changes in crop growth than multispectral broadbands but few studies quantified the gains in the former over the later. The PRecursore IperSpettrale della Missione Applicativa (PRISMA) mission offers narrow (≤10 nm) band capability across the full optical range. The multispectral broadband Sentinel-2 mission carries four experimental red-edge and near infrared (NIR) hyperspectral narrow (≤20 nm) bands. We compared the performance of PRISMA and Sentinel-2 spectral bands at important phases of crop development (vegetative, reproductive, maturity) in estimating field-level biomass and yield for corn, rice, soybean, and wheat. We selected three data-driven methods: two-band vegetation indices (TBVIs), partial least squares regression (PLSR), and random forest (RF). The PRISMA and Sentinel-2 models on average explained approximately 20% more variability in biomass and yield with RF than TBVIs and PLSR. The mean RMSE of the PRISMA RF models was 0.42 and 0.17 kg m−2, which was lower than the Sentinel-2 RF models (0.48 and 0.18 kg m−2). Multidate image (seasonal) model performance was generally higher than single-date image model performance. PRISMA shortwave infrared narrowbands and Sentinel-2 red-edge and near infrared bands were among the top-performing spectral regions. The results highlight potential complementarity between the PRISMA and Sentinel-2 missions for predicting crop biomass and yield. The results also show the benefits, limitations, and pitfalls of hyperspectral imaging in agricultural monitoring, which is important for upcoming operational hyperspectral missions, such as ESA CHIME and NASA Surface Biology and Geology (formerly HysPIRI).

Uncertainties and Perspectives on Forest Height Estimates by Sentinel-1 Interferometry

Forest height is a key parameter in forestry. SAR interferometry (InSAR) techniques have been extensively adopted to retrieve digital elevation models (DEM) to give a representation of the continuous variation of the Earth’s topography, including forests. Unfortunately, InSAR has been proven to fail over vegetation due to low coherence values; therefore, all phase unwrapping algorithms tend to avoid these areas, making InSAR-derived DEM over vegetation unreliable. In this work, a sensitivity analysis was performed with the aim of properly initializing the relevant operational parameters (baseline and multilooking factor) to maximize the theoretical accuracy of the height difference between the forest and reference point. Some scenarios were proposed to test the resulting “optimal values”, as estimated at the previous step. A simple model was additionally proposed and calibrated, aimed at predicting the optimal baseline value (and therefore image pair selection) for height uncertainty minimization. All our analyses were conducted using free available data from the Copernicus Sentinel-1 mission to support the operational transfer into the forest sector. Finally, the potential uncertainty affecting resulting height measures was quantified, showing that a value lower than 5 m can be expected once all user-dependent parameters (i.e., baseline, multilooking factor, temporal baseline) are properly tuned.

Interference-sensitive coastal SAR altimetry retracking strategy for measuring significant wave height

Satellite altimetry is a radar remote sensing technology for the precise observation of the ocean surface and its changes over time. Its measurements allow the determination of geometric and physical parameters such as sea level, significant wave height or wind speed. This work presents a novel coastal retracking algorithm for SAR altimetry to estimate the significant wave height. The concept includes an adaptive interference masking scheme to sense and mitigate spurious interfering signals that typically arise from strongly reflective targets in the coastal zone. The described procedure aims at increasing the number of valid records in the coastal zone. The effectiveness of the novel retracking algorithm is validated using the methodology recently developed in the framework of the European Space Agency Sea State Climate Change Initiative project. Several different metrics were extracted as functions of sea state and distance to the nearest coast: outliers, number of valid records, intrinsic noise, power spectral density, and correlation statistics for the comparison with wave model and in-situ data. Two coastal case study scenarios complement the validation. The results show that with the presented novel retracking algorithm, the number of valid 20-Hz records in the near coastal zone of less than 5 km off the coast is increased by more than 25% compared to the best competing coastal retracking algorithm with no degradation of quality of the estimated records. We emphasise the importance of the correct choice of the quality flag that is provided together with the significant wave height. Our findings suggest that the strategy for the significant wave height quality flag of the official baseline Level-2 product of the Sentinel-3 mission can be redefined to obtain more robust significant wave height estimates in the coastal zone.

Multi-Temporal SAR Interferometry for Vertical Displacement Monitoring from Space of Tengiz Oil Reservoir Using SENTINEL-1 and COSMO-SKYMED Satellite Missions

This study focused on the quantitative assessment of the vertical displacement velocities retrieved using Sentinel-1 and Cosmo-SkyMed synthetic aperture radar images for the Tengiz oilfield. Tengiz oilfield was selected as a study area because of its historically reported continuous subsidence and limited up-to-date studies during recent years. The small baseline subset time-series technique was used for the interferometric processing of radar images acquired for the period of 2018–2020. The geospatial and statistical analyses allowed to determine the existing hotspots of the subsidence processes induced by oil extraction in the study area. Ground deformation measurements derived from the Sentinel-1 and COSMO-SkyMed satellite missions showed that the Tengiz oilfield continuously subsided during 2018–2020 with the maximum annual vertical displacement velocity around −77.4 mm/y and −71.5 mm/y, respectively. The vertical displacement velocities derived from the Sentinel-1 and the COSMO-SkyMed images showed a good statistical relationship with R2≥0.73 and RMSE ≤3.68 mm. The cumulative vertical displacement derived from both satellites for the most subsiding location also showed a good statistical relationship with R2 equal to 0.97 and RMSE = ± 4.69. The observed relative differences of measurements by both satellites were acceptable to determine the ongoing vertical surface displacement processes in the study area. These studies demonstrated a practical novelty for the petroleum industry in terms of the comparative assessment of surface displacement measurements using time-series of medium-resolution Sentinel-1 and high-resolution COSMO-SkyMed radar images.

Detecting Soil pH from Open-Source Remote Sensing Data: A Case Study of Angul and Balangir Districts, Odisha State

Soil sampling, collection, and analysis are a costly and labor-intensive activity that cannot cover the entire farmlands; hence, it was conceived to use high-speed open-source platforms like Google Earth Engine in this research to estimate soil characteristics remotely using high-resolution open-source satellite data. The objective of this research was to estimate soil pH from Sentinel-1, Sentinel-2, and Landsat-8 satellite-derived indices; data from Sentinel-1, Sentinel-2, and Landsat-8 satellite missions were used to generate indices and as proxies in a statistical model to estimate soil pH. Step-wise multiple regression (SWMR), artificial neural networks (ANN), and random forest (RF) regression were used to develop predictive models for soil pH, SWMR, ANN, and RF regression models. The SWMR greedy method of variable selection was used to select the appropriate independent variables that were highly correlated with soil pH. Variables that were retained in the SWMR are B2, B11, Brightness index, Salinity index 2, Salinity index 5 of Sentinel-2 data; VH/VV index of Sentinel 1 and TIR1 (thermal infrared band1) Landsat-8 with p-value < 0.05. Among the four statistical models developed, the class-wise RF model performed better than other models with a cumulative correlation coefficient of 0.87 and RMSE of 0.35. The better performance of class-wise RF models can be attributed to different spectral characteristics of different soil pH groups. More than 70% of the soils in Angul and Balangir districts are acidic soils, and therefore, the training of the dataset was affected by that leading to misclassification of neutral and alkaline soils hindering the performance of single class models. Our results showed that the spectral bands and indices can be used as proxies to soil pH with individual classes of acidic, neutral, and alkaline soils. This study has shown the potential in using big data analytics to predict soil pH leading to the accurate mapping of soils and help in decision support.

Bathymetric and Capacity Relationships Based on Sentinel-3 Mission Data for Aswan High Dam Lake, Egypt

Aswan High Dam Lake (AHDL) is one of the most relevant hot spots at both local and global levels after construction of the Grand Ethiopian Renaissance Dam (GERD) was announced. The management of AHDL is a vital task, which requires the input of reliable information such as the lake bathymetry, water level, and the water surface area. Traditional, bathymetric methods are still very expensive and difficult to operate. Nowadays, satellite data and remote sensing techniques are easily accessible. In particular, datasets produced by operational missions are freely and globally available, and may provide efficient and inexpensive solutions for the retrieval of quantitative parameters concerning strategic water bodies, such as AHDL. This work identifies the performance of Sentinel-3A optical imagery data in the visible and NIR bands from the two optical instruments SLSTR and OLCI, and proposes the integration with Sentinel-3A radar altimetry from SRAL instrument applied to AHDL. This preliminary and first study investigated the relationship between the reflectance data and in situ data for water depth after a bathymetric campaign in the deep-water region using statistical regression models. These statistical models showed promising results in terms of correlation value (R2 &gt; 0.8) and normalized root mean square errors (NRMSE &lt; 0.4). Also, Heron’s formula was applied to combine optical imagery and Sentinel-3 altimetry water level datasets to estimate water storage variations in AHDL. In addition, equations governing the relationship between water level, water surface area, and water volume were analyzed. The work is very useful for all authorities and stakeholders dealing with large water bodies.

Semantic Segmentation of Metoceanic Processes Using SAR Observations and Deep Learning

Through the Synthetic Aperture Radar (SAR) embarked on the satellites Sentinel-1A and Sentinel-1B of the Copernicus program, a large quantity of observations is routinely acquired over the oceans. A wide range of features from both oceanic (e.g., biological slicks, icebergs, etc.) and meteorologic origin (e.g., rain cells, wind streaks, etc.) are distinguishable on these acquisitions. This paper studies the semantic segmentation of ten metoceanic processes either in the context of a large quantity of image-level groundtruths (i.e., weakly-supervised framework) or of scarce pixel-level groundtruths (i.e., fully-supervised framework). Our main result is that a fully-supervised model outperforms any tested weakly-supervised algorithm. Adding more segmentation examples in the training set would further increase the precision of the predictions. Trained on 20 × 20 km imagettes acquired from the WV acquisition mode of the Sentinel-1 mission, the model is shown to generalize, under some assumptions, to wide-swath SAR data, which further extents its application domain to coastal areas.

Evaluation of the Urban Microclimate in Catania Using Multispectral Remote Sensing and GIS Technology

This study focuses on the determination and examination of both the Land Surface Temperature (LST) and the atmospheric temperature in the city of Catania Sicily (Italy), through freely available satellite remote sensing images from the Sentinel-2 and MODIS missions. Satellite images were processed as raster data in free and open-source GIS environments. The GIS software allows the retrieval, processing of the satellite images for the estimation of the LST and the atmospheric temperature with a very coarse spatial resolution. In particular, the proposed procedure allows increasing the spatial resolution of satellite images, from 250 m (LRES) to 10 m (HRES) through the principle of “Disaggregation of thermal images”. The analysis provided georeferenced maps which show the LST, as well as the atmospheric temperature within the investigated area with a very fine resolution, 10 m. Such spatial resolution reveals evident correlations between areas with different urban densities and their microclimate. An important result of this study is that significant LST differences can be observed during both day (15–17 °C) and night (2–3 °C) between green and built-up areas. The outcomes of this study highlight the effectiveness of the combined use of satellite remote sensing and GIS for analyzing the thermal response of urbanized areas with different built density.

Radiometric calibration stability assessment of Sentinel-1B using point targets at Surat Basin, Australia

AbstractThe launch of the Sentinel-1B satellite in April 2016 completed the two-satellite synthetic aperture radar (SAR) constellation of the European Copernicus Sentinel-1 mission. The European Space Agency executed the calibration of this sensor during the commissioning phase and an independent calibration by the German Aerospace Center (DLR) in 2016. The calibration parameters must be monitored to assess the stability of the instrument. This study reports the temporal stability assessment of radiometric calibration and image quality parameters of Sentinel-1B SAR data using the corner reflector (CR) array, Surat Basin, Australia. Impulse response functions generated from the CRs in the satellite images were used to derive the image quality parameters. The average radar cross-section difference between estimated and theoretical values (38.40 dB m2) was 0.53 dB m2 for 1.5 m CRs, which is accordant with the absolute radiometric accuracy specified for the Sentinel-1 SAR system. Derived image quality parameters viz. the mean peak-to-side lobe ratio, mean integrated side lobe ratio, and spatial resolutions in the range and azimuth directions were found to be accordant with the specified value for the Sentinel-1 SAR system. The results indicate the excellent quality of the Sentinel-1B data.

Exploiting time series of Sentinel-1 and Sentinel-2 to detect grassland mowing events using deep learning with reject region

Governments pay agencies to control the activities of farmers who receive governmental support. Field visits are costly and highly time-consuming; hence remote sensing is widely used for monitoring farmers’ activities. Nowadays, a vast amount of available data from the Sentinel mission significantly boosted research in agriculture. Estonia is among the first countries to take advantage of this data source to automate mowing and ploughing events detection across the country. Although techniques that rely on optical data for monitoring agriculture events are favourable, the availability of such data during the growing season is limited. Thus, alternative data sources have to be evaluated. In this paper, we developed a deep learning model with an integrated reject option for detecting grassland mowing events using time series of Sentinel-1 and Sentinel-2 optical images acquired from 2000 fields in Estonia in 2018 during the vegetative season. The rejection mechanism is based on a threshold over the prediction confidence of the proposed model. The proposed model significantly outperforms the state-of-the-art technique and achieves event accuracy of 73.3% and end of season accuracy of 94.8%.

Sentinel-1 snow depth retrieval at sub-kilometer resolution over the European Alps

Abstract. Seasonal snow is an essential water resource in many mountain regions. However, the spatio-temporal variability in mountain snow depth or snow water equivalent (SWE) at regional to global scales is not well understood due to the lack of high-resolution satellite observations and robust retrieval algorithms. We investigate the ability of the Sentinel-1 mission to monitor snow depth at sub-kilometer (100 m, 500 m, and 1 km) resolutions over the European Alps for 2017–2019. The Sentinel-1 backscatter observations, especially in cross-polarization, show a high correlation with regional model simulations of snow depth over Austria and Switzerland. The observed changes in radar backscatter with the accumulation or ablation of snow are used in an empirical change detection algorithm to retrieve snow depth. The algorithm includes the detection of dry and wet snow conditions. Compared to in situ measurements at 743 sites in the European Alps, dry snow depth retrievals at 500 m and 1 km resolution have a spatio-temporal correlation of 0.89. The mean absolute error equals 20 %–30 % of the measured values for snow depths between 1.5 and 3 m. The performance slightly degrades for retrievals at the finer 100 m spatial resolution as well as for retrievals of shallower and deeper snow. The results demonstrate the ability of Sentinel-1 to provide snow estimates in mountainous regions where satellite-based estimates of snow mass are currently lacking. The retrievals can improve our knowledge of seasonal snow mass in areas with complex topography and benefit a number of applications, such as water resource management, flood forecasting, and numerical weather prediction. However, future research is recommended to further investigate the physical basis of the sensitivity of Sentinel-1 backscatter observations to snow accumulation.

Simultaneous retrieval of selected optical water quality indicators from Landsat-8, Sentinel-2, and Sentinel-3

Constructing multi-source satellite-derived water quality (WQ) products in inland and nearshore coastal waters from the past, present, and future missions is a long-standing challenge. Despite inherent differences in sensors’ spectral capability, spatial sampling, and radiometric performance, research efforts focused on formulating, implementing, and validating universal WQ algorithms continue to evolve. This research extends a recently developed machine-learning (ML) model, i.e., Mixture Density Networks (MDNs) (Pahlevan et al., 2020; Smith et al., 2021), to the inverse problem of simultaneously retrieving WQ indicators, including chlorophyll-a (Chla), Total Suspended Solids (TSS), and the absorption by Colored Dissolved Organic Matter at 440 nm (acdom(440)), across a wide array of aquatic ecosystems. We use a database of in situ measurements to train and optimize MDN models developed for the relevant spectral measurements (400–800 nm) of the Operational Land Imager (OLI), MultiSpectral Instrument (MSI), and Ocean and Land Color Instrument (OLCI) aboard the Landsat-8, Sentinel-2, and Sentinel-3 missions, respectively. Our two performance assessment approaches, namely hold-out and leave-one-out, suggest significant, albeit varying degrees of improvements with respect to second-best algorithms, depending on the sensor and WQ indicator (e.g., 68%, 75%, 117% improvements based on the hold-out method for Chla, TSS, and acdom(440), respectively from MSI-like spectra). Using these two assessment methods, we provide theoretical upper and lower bounds on model performance when evaluating similar and/or out-of-sample datasets. To evaluate multi-mission product consistency across broad spatial scales, map products are demonstrated for three near-concurrent OLI, MSI, and OLCI acquisitions. Overall, estimated TSS and acdom(440) from these three missions are consistent within the uncertainty of the model, but Chla maps from MSI and OLCI achieve greater accuracy than those from OLI. By applying two different atmospheric correction processors to OLI and MSI images, we also conduct matchup analyses to quantify the sensitivity of the MDN model and best-practice algorithms to uncertainties in reflectance products. Our model is less or equally sensitive to these uncertainties compared to other algorithms. Recognizing their uncertainties, MDN models can be applied as a global algorithm to enable harmonized retrievals of Chla, TSS, and acdom(440) in various aquatic ecosystems from multi-source satellite imagery. Local and/or regional ML models tuned with an apt data distribution (e.g., a subset of our dataset) should nevertheless be expected to outperform our global model.

Combining LiDAR Metrics and Sentinel-2 Imagery to Estimate Basal Area and Wood Volume in Complex Forest Environment via Neural Networks

Forest ecosystems play a fundamental role in natural balances and climate mechanisms through their contribution to global carbon storage. Their sustainable management and conservation is crucial in the current context of global warming and biodiversity conservation. To tackle such challenges, Earth observation data has been identified as a valuable source of information capable to provide stakeholders with informative indicators to support the decision making process related to forest ecosystems management. While Earth observation data constitutes an unprecedented opportunity to monitor forest ecosystems, its effective exploitation still poses serious challenges since multi-modal (i.e. multi-scale and multi-source) information need to be combined in order to describe complex natural phenomena. To deal with this particular issue in the context of structure and biophysical variables estimation for forest characterization, we propose a new deep learning based fusion strategy to combine together high density 3D-point clouds acquired by airborne laser scanning (ALS) with high resolution optical imagery freely accessible via the Sentinel-2 mission. In order to manage and fully exploit the available multi-modal information, we implement a two-branch late fusion deep learning architecture taking advantage of the specificity of each modality: on the one hand, a 2D-CNN branch is devoted to the analysis of Sentinel-2 time series data and, on the other hand, a Multi-Layer Perceptron branch is dedicated to the processing of LiDAR-derived information. The whole framework is learnt end-to-end in order to effectively exploit the complementarity between the two sources of information.The performance of our framework is evaluated on two forest variables of interest: total volume and basal area at stand level. The obtained results underline that the availability of multi-modal remote sensing data is not a direct synonym of performance improvements but, the way in which they are combined together is of paramount importance to leverage the complex interplay among the different input sources.

Deformation Fringes Detection in SAR interferograms Using Deep Learning

The success achieved by using SAR data in the study of the Earth led to a firm commitment from space agencies to develop more and better space-borne SAR sensors. This involvement of the space agencies makes us believe that it is possible to increase the potential of SAR interferometry (InSAR) to near real-time monitoring. Among this ever-increasing number of sensors, the ESA’s Sentinel-1 (C-band) mission stands out and appears to be disruptive. This mission is acquiring vast volumes of data making current analyzing approaches inviable. This amount of data can no longer be analyzed and studied using classic methods raising the need to use and create new techniques. We believe that Machine Learning techniques can be the solution to overcome this issue since they allow to train Deep Learning models to automate human processes for a vast volume of data. In this paper, we use deep learning models to automatically find and locate deformation areas in InSAR interferograms without atmospheric correction. We train three state-of-the-art classification models for detection deformation areas, achieving an AUC of 0.864 for the best model (VGG19 for wrapped interferograms). Additionally, we use the same models as encoders to train U-net models, achieving a Dice score of 0.54 for InceptionV3. It is necessary more data to achieve better results in segmentation.

Mapping Floods in Urban Areas From Dual-Polarization InSAR Coherence Data

Previous studies have shown that the decrease of temporal interferometric synthetic aperture radar (InSAR) coherence could be exploited to detect the appearance of floodwater in urban areas. However, as of today, approaches based on this principle only make use of single co-polarization images for identifying the presence of floodwater in the double-bounce feature. In this study, we take advantage of both co- and cross-polarization images to detect significant decreases of the multitemporal InSAR coherence in order to enhance the mapping of floodwater in urban areas. We consider that not only double-bounce scattering, but also multiple-bounce may occur in urban areas depending on how the building facades are oriented with respect to the synthetic aperture radar (SAR) sensor’s line of sight. The Sentinel-1 (S-1) mission is particularly well suited for applying and testing this kind of approach due to the systematic availability of dual-polarization data. Using as a test case, the widespread flooding in the city of Houston, USA, caused by Hurricane Harvey in 2017, we demonstrate that the proposed methodology leads to an increase of the accuracy of the urban flood maps from 75.2% when only using the VV polarization, to 82.9% when using the dual polarization information.