Snow is a vital environmental parameter that holds significance across various disciplines, such as hydrology, meteorology, and natural disaster management. With the increasing accessibility of snow products derived from Synthetic Aperture Radar (SAR) and optical data, like Sentinel-1 wet snow and Sentinel-2 total snow, users have benefited from improved snow mapping and monitoring. However, snow mapping in the mountainous areas remains challenging due to the difficulty of obtaining reliable ground truth data on steep mountain terrain. In this study, we introduce a deep semantic segmentation framework, SACUNet, specifically designed for wet snow detection from SAR image time series in mountainous environments. To address the lack of ground truth, we constructed a high-confidence training and validation database through a rigorous decision-fusion process combining multi-temporal Sentinel-1 wet snow detections with Sentinel-2 total snow maps. We also propose two complementary metrics, the Conditional Agreement Rate (CAR) and the Wet Snow Intersection over Union (WSIoU), to quantify the robustness and consistency of the fusion procedure, therefore ensuring the reliability of training labels in the absence of in-situ data. SACUNet integrates advanced techniques like: (i) Depthwise Separable Convolution, which captures cross-channel dependencies and adapts feature representations, and (ii) Atrous Separable Convolution, which further refines and consolidates the learned features, into the U-Net architecture. The proposed framework has been successfully employed to monitor wet snow in the Mont-Blanc massif, using a time series of 69 Sentinel-1 images acquired from 05 July 2020, to 29 August 2021. SACUNet demonstrates remarkable accuracy in wet snow detection, with an Overall Accuracy of 97%, Precision of 94%, Recall of 97%, Intersection over Union at 92%, and an F1-Score reaching 96%. Validation against meteorological records from four alpine stations confirmed that SACUNet effectively tracks seasonal wet snow dynamics, suppresses false detections during cold periods, and captures realistic high-altitude melt events. Moreover, the model trained in Mont-Blanc generalized successfully to the Vanoise massif, demonstrating its transferability to other alpine regions. Beyond quantitative accuracy, SACUNet enables the spatio-temporal analysis of wet snow evolution, offering insights into its extent, frequency, and seasonal progression across elevation bands. These findings highlight the framework’s potential as an operational tool for large-scale wet snow monitoring in mountainous environments.

In this study, we develop an original method for segmenting Sentinel-1 SAR images as a discrete-time dynamical system for estimating the extent of wet snow. The system considers different variables and parameters, including radar amplitude images and terrain information (altitudes, slopes, orientations). The dynamical variable is the SAR amplitude ratio (ratio between a SAR and reference image without snow), whose evolution follows a law described by a function applied to a set of connected pixels. The set of corresponding pixels and the strength of their connection are determined from a digital terrain model. The application of the dynamical system is nonlinear and is inspired by models of interacting particles from biology (neural network models). As the iterations increase, the dynamics leads to a segmentation of the image taken as an initial condition. A significant advantage of our system is that it considers the physical information of the terrain to determine couplings between neighbouring pixels. This algorithm has been tested over the season 2017-2018 and provides estimates of wet snow maps that are perfectly consistent with reference products, opening up new prospects for their use in different contexts (for example, for improving existing segmentation methods and for enhancing the quality of gaps filling approaches in case of missing data, etc.).

Monitoring the state of seasonal snow and its evolution over time is essential for various applications, including meteorology, risk and water resource management, biodiversity, ecosystems, and climate change studies. The use of Sentinel-1/-2 satellites has revolutionised mountain snow cover observation, providing unprecedented spatial resolutions and revisit times. Sentinel-1 SAR images are used to identify wet snow through active remote sensing in the C-band, while Sentinel-2 data monitor snow extent, whether wet or dry. This work aims to enhance the spatial and temporal variability of snowpack simulations from the Crocus model by assimilating wet snow products from Sentinel-1. This is achieved by relying on the ensemble assimilation chain at Centre d’Etudes de la Neige, which uses a particle filter applied to SURFEX/Crocus model snowpack simulations at a resolution of 250 meters. An observation operator has been implemented to allow the snow model to assimilate meltlines from Sentinel-1, which provides the probability of wet snow by classes of altitudes and orientations. First assimilation results over a vast alpine area of steep relief (Grandes-Rousses massif) following the assimilation of several Sentinel-1-based products will be presented. The impact of the assimilation on key snow parameters such as snow depth and snow water content will be discussed, along with an analysis of the assimilation performance according to independent data and according to topography, time, and date of observation.

Snow and glaciers play a crucial role in various environmental services in hydrology and climate, and also regarding hazards assessment related to snow and ice avalanches. SAR imagery is particularly useful for studying snow and glacier-related issues: it is insensitive to the cloud cover and is sensitive to some snow/glacier properties such as liquid water content. In this study, we focus on the snowmelt dynamics in the catchment of the Saint-Sorlin Glacier in the French Alps using SAR images in C-bands from Sentinel-1. Our objective is to gain a better understanding of the snow and ice melt processes in a glacierized catchment by monitoring the spatial and temporal variability of the SAR signal on the glacier (ablation and accumulation areas) and outside the glacier, and comparing it to fractional snow and ice maps from Sentinel-2 spectral unmixing. We are able to identify key melting phases using SAR backscatter time series. The set of melting phases are compared with the ones estimated from liquid water content and snow water equivalent simulations made with the snowpack model Crocus.

We present a novel iterative method for segmenting Sentinel-1 SAR images to estimate the extent of wet snow in the mountains. The algorithm consists of a discrete-time dynamical system fed by various variables, including radar amplitude images and terrain information (elevations, slopes, orientations), and controlled by different parameters. The dynamical system uses the SAR amplitude ratio (ratio between an SAR and a reference image without snow) as an initial condition and makes it evolve iteratively toward a segmented image. A digital terrain model modulates the connection between pixels in the discrete-time dynamical system, thereby ensuring a physical consistency in the iterative algorithm. This algorithm is tested over the 2017–2018 season and its outputs are compared with the Copernicus state-of-the-art snow/wet snow products and with expert estimates of snow elevations at the scale of the Grandes-Rousses alpine French massif. We show that the dynamical system can be used to derive wet snow maps in very good agreement with independent data. Furthermore, the elevations and dates of snow retreat estimated by the dynamical system are found to be in much better agreement with estimates obtained from optical satellites and forecaster expertise when compared with the SAR-based Copernicus products for southern facing slopes.

We propose a new approach for wet snow extent mapping in Synthetic Aperture Radar (SAR) images by using a convolutional neural network (CNN) designed to learn with respect to snowpack outputs from the state-of-the-art snow model Crocus. The CNN was trained to classify the wet snow conditions based on features extracted from the SAR images, using both the VV,VH channel and the ratio between these channels and those of a reference image in summer. One of the key points of this work is the comprehensive comparison we have made between the performance of the CNN method and other advanced statistical methods. We found that the CNN was able to achieve good accuracy in wet snow classification, and giving a complementary vision of the solutions obtained by other machine learning algorithms such as the Random Forest classifier. The results of this study demonstrate the potential of using CNNs and SAR images for wet snow classification and highlight the importance of using physical information model for training machine learning models in snow state identification, a domain where collecting ground truth is intricate due to the complexity of the snowpack moisture measurement systems.

Snow holds a significant importance as a fundamental environmental factor in multiple domains. Obtaining accurate ground truth data for snow mapping in mountainous areas presents a significant challenge. To address this issue, this paper presents a deep semantic learning framework for the segmentation of Sentinel-1 images for wet snow detection in mountainous areas. Firstly, we propose to create a deep leaning database based on snow products derived from Sentinel-1 and Sentinel-2 data. Afterward, we introduce a deep convolutional neural network called ReXcepUnet, which combines the U-Net architecture and the powerful Xception backbone. Finally, the proposed framework has been successfully applied to monitor wet snow in the Mont Blanc massif, yielding high accuracy results. The ReXcepUnet model demonstrates a good performance in wet snow detection, particularly in high-relief regions like the Mont Blanc massif.

In this study, we develop a novel method to automatically detect areas of snow avalanche debris using a color space segmentation technique applied to synthetic aperture radar (SAR) image time series through January 2018 in the Swiss Alps. Debris avalanche zones are detected assuming that these areas are characterized by a significant and localized increase in SAR signal relative to the surrounding environment. We undertake a sensitivity study by calculating debris products by varying the D-M reference images (a stable reference image taken several weeks before the event). We examine the results according to the direction of the orbit, the characteristics of the terrain (slope, altitude, orientation), and also by evaluating the relevance of the detection with the help of an independent SPOT database by Hafner and Buhler[1]including 18 737 avalanche events. Small avalanches are not detected by SAR images, and depending on the orientation of the terrain some avalanches are not detected by either the ascending or the descending orbit. The detection results vary with the reference image; best detection results are obtained with some selected individual dates with almost 70% of verified avalanche events using the ascending orbit.

This paper proposes to use ground-based high-resolution 3-D radar imaging, in order to study the interaction between electromagnetic waves and snow packs, and to provide a physical interpretation for the reflectivity of Sentinel 1 images over snow covered regions. Preliminary results show that, according to usually assumed behaviors, fresh snow has an extremely low reflectivity at C band, and wet snow does not let waves go through. Between these two extreme configurations, this study reveals that snow packs may have complex and significant scattering patterns, mainly due to the presence of transformed snow and of rough interfaces between the layers, again due to transformation phenomena.

This article discusses the issue of snow avalanches monitoring at regional scale in the French Alps using SAR observations from Sentinel-1. A color space segmentation technique applied to SAR image time series has recently been used to detect areas of avalanche debris and successfully evaluated in the Swiss Alps. The objective of this paper is to generalize the detection of avalanche debris to all the French massifs and to develop indicators derived from our satellite product Sentinel-1 that can provide information about the avalanche activity at the spatial scale of the massifs. Evaluations are also planned using model outputs, in-situ measurements and very high resolution optical satellite observations.

We develop and evaluate a new methodology of wet snow detection over the French mountains using a colour space segmentation technique applied to SAR image time series during the period ranging from August 2016 to August 2020. Wet snow products are evaluated against snow products from optical measurements and snow simulations from a state-of-the-art snowpack model. The variability of wet snow, in space and time, is examined to infer knowledge about the seasonal and inter-annual variability of wet snow at the scale of the French mountain massifs.

This paper presents experimental results of the 3-D characterization of alpine snowpacks, obtained using a low complexity portable MIMO radar system that operates at C-band. Different types of snow at different altitudes and seasons are studied. The acquired datasets are processed using the Back Projection Algorithm and the resulting tomograms are compared to Météo France ground measurements (Snow Micro Pen transects, density and stratigraphy profiles and liquid water content). The obtained tomograms show that the system mostly detects melt forms and faceted crystals. These measurements provide 3-D electromagnetic ground truth that can be used to confirm the results obtained by Sentinel-1.

In this paper, the surface emissivity is retrieved over the Arctic sea ice/open seas using observations from the advanced microwave sounding unit window channels during the year 2009. The emissivity computation is performed using two contrasted surface assumptions: specular and Lambertian assumptions. The obtained sea ice surface emissivities are studied in this paper with a focus on the effect of the surface assumption. Some factors of variability of the obtained emissivities are analyzed: variability in space, in time, with the zenith angle, and with respect to the frequency. We show that the near-nadir surface emissivity and emissivity difference (obtained using two contrasted surface assumptions) could be used as an excellent proxy to detect ice/no ice regions. We also show that near-nadir sea ice emissivity at some selected frequencies and the combination of both high and low window frequencies could also be very useful to better characterize sea ice surface physical properties and provide additional information for existing sea ice classifications, as they bring relevant information about first year and multiyear sea ice properties and their seasonal evolution.

Over the last ten years Earth observing systems have grown considerably and thus allow the use of Global Navigation Satellite System (GNSS) signals for remote sensing. Several studies in the U.S. have proven the ability of existing GNSS ground networks to measure environmental parameters quantifying essential surface conditions for the understanding of the water cycle. This is possible by taking advantage of reflected signals around geodetic stations. Improved characterization of environmental disturbances in GNSS signals is also essential to increase the accuracy of future GNSS measurement for applications in Earth Sciences. GNSS reflectometry provides these surface parameters at an intermediate scale that can bridge the lack of data between in situ and satellite observations. Here we adapt and develop for the first time the GNSS reflectometry method using the existing sites of the French national GNSS permanent network (RENAG) for snow and soil moisture applications.

Passive microwave (PMW) satellite-based precipitation over land algorithms rely on physical models to define the most appropriate channel combinations to use in the retrieval, yet typically require considerable empirical adaptation of the model for use with the satellite measurements. Although low-frequency channels are better suited to measure the emission due to liquid associated with rain, most techniques to date rely on high-frequency, scattering-based schemes since the low-frequency methods are limited to the highly variable land surface background, whose radiometric contribution is substantial and can vary more than the contribution of the rain signal. Thus, emission techniques are generally useless over the majority of the Earth's surface. As a first step toward advancing to globally useful physical retrieval schemes, an intercomparison project was organized to determine the accuracy and variability of several emissivity retrieval schemes. A three-year period (July 2004-June 2007) over different targets with varying surface characteristics was developed. The PMW radiometer data used includes the Special Sensor Microwave Imagers, SSMI Sounder, Advanced Microwave Scanning Radiometer (AMSR-E), Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI), Advanced Microwave Sounding Units, and Microwave Humidity Sounder, along with land surface model emissivity estimates. Results from three specific targets in North America were examined. While there are notable discrepancies among the estimates, similar seasonal trends and associated variability were noted. Because of differences in the treatment surface temperature in the various techniques, it was found that comparing the product of temperature and emissivity yielded more insight than when comparing the emissivity alone. This product is the major contribution to the overall signal measured by PMW sensors and, if it can be properly retrieved, will improve the utility of emission techniques for over land precipitation retrievals. As a more rigorous means of comparison, these emissivity time series were analyzed jointly with precipitation data sets, to examine the emissivity response immediately following rain events. The results demonstrate that while the emissivity structure can be fairly well characterized for certain surface types, there are other more complex surfaces where the underlying variability is more than can be captured with the PMW channels. The implications for Global Precipitation Measurement-era algorithms suggest that physical retrievals are feasible over vegetated land during the warm seasons.

Ground-based multifrequency (L-band to W-band, 1.41-90 GHz) and multiangular (20°-50°) bipolarized (V and H) microwave radiometer observations, acquired over a dense wheat field, are analyzed in order to assess the sensitivity of brightness temperatures (Tb) to land surface properties: surface soil moisture (mv) and vegetation water content (VWC). For each frequency, a combination of microwaveTbobserved at either two contrasting incidence angles or two polarizations is used to retrievemvand VWC, through regressed empirical logarithmic equations. The retrieval performance of the regression is used as an indicator of the sensitivity of the microwave signal to eithermvor VWC. In general, L-band measurements are shown to be sensitive to bothmvand VWC, with lowest root mean square errors (0.04 m3·m-3and 0.52 kg ·m-2, respectively) obtained at H polarization, 20° and 50° incidence angles. In spite of the dense vegetation, it is shown thatmvinfluences the microwave observations from L-band to K-band (23.8 GHz). The highest sensitivity to soil moisture is observed at L-band in all configurations, while observations at higher frequencies, from C-band (5.05 GHz) to K-band, are only moderately influenced bymvat low incidence angles (e.g., 20°). These frequencies are also shown to be very sensitive to VWC in all the configurations tested. The highest frequencies (Q- and W-bands) are shown to be moderately sensitive to VWC only. These results are used to analyze the response of W-band emissivities derived from the Advanced Microwave Sounding Unit instruments over northern France.

This paper describes several sensitivity studies carried out with the French global 4-D-Var system to check its ability to assimilate surface-sensitive observations over land from the Special Sensor Microwave Imager (SSM/I). As well as a sound knowledge of land-surface parameters, the assimilation of SSM/I observations requires effective rain-detection and bias-correction algorithms. Three sensitivity components are hence analyzed with a special emphasis on the land-surface emissivity at SSM/I frequencies estimated from satellite observations. Several rain algorithms were tested to reject cloudy/rainy observations over land, and the bias-correction scheme was adapted to improve its performance over land and sea surfaces. Once these problems have been outlined, a global 4-D-Var assimilation experiment which assimilates SSM/I observations over land surfaces was run and compared with a control experiment. The impact on forecast scores has been found to be globally positive. Nevertheless, the very high sensitivity of SSM/I to each of the three components presented in this study is characterized by opposite effects that, once clustered together, lead to some residual biases over land due to their combined effects.

The 12 papers in this special issue focus on the remote sensing and modeling of surface properties.

This work is in direct line with the Concordiasi international project. It aims to better constrain atmospheric analyses by improving the assimilation of low-level Advanced Microwave Sounding Unit (AMSU)-A and AMSU-B microwave observations over Antarctica. So far, a very small amount of available AMSU observations is effectively assimilated over Antarctica. To assimilate more observations, different issues have to be dealt with. In this work, the surface emissivity issue over Antarctica is examined. In a first step, a thorough review of the use of a specular assumption to calculate emissivity from AMSU-A measurements has been undertaken. The effect of five different assumptions about the surface on retrieved AMSU emissivities has then been evaluated using a one-year database: specular, Lambertian, and three intermediate assumptions. Simulations of brightness temperatures at AMSU sounding frequencies have been produced using a radiative transfer model. The emissivities obtained using the five assumptions have been found very useful in improving these simulations. The most successful schemes are found to be the Lambertian scheme during the winter season and a specular or an intermediate scheme (50% specular, 50% Lambertian) during Antarctica's short summer.

To determine land surface emissivity from satellite microwave measurements, the surface is usually assumed to be specular. Questions about the validity of this approximation to estimate emissivity from nadir viewing radiometers were raised. This work aims to examine the validity of the specular assumption by evaluating errors induced when deriving emissivities from near-nadir measurements over snow-free areas. Brightness temperature simulations near nadir above both a specular and a Lambertian surface are compared. Errors on the retrieved emissivity introduced by the specular assumption are also quantified. The results show that the impact of the specular assumption when the surface is Lambertian is limited: less than 1% error in most atmospheric situations over natural snow-free surfaces.

The microwave radiometers on altimeter missions are specified to provide the "wet" troposphere path delay with an uncertainty of 1 cm or lower, at the location of the altimeter footprint. The constraints on the calibration and stability of these instruments are therefore particularly stringent. The paper addresses the questions of long-term stability and absolute calibration of the National Aeronautics and Space Administration Topography Experiment (TOPEX) and European Space Agency European Remote Sensing 2 (ERS-2) radiometers over the entire range of brightness temperatures. Selecting the coldest measurements over ocean from the two radiometers, the drift of the TOPEX radiometer 18-GHz channel is confirmed to be about 0.2 K/year over the seven first years of the mission, and the one of the ERS-2 radiometer 23.8-GHz channel to be -0.2 K/year. The good stability of the other channels is confirmed (drift less than 0.04 K/year). The use of continental targets for analyzing the long-term drift is evaluated: the natural interannual variability prevents one from directly monitoring the drift of each channel, but the relative variation between two channels of the same instrument is found reliable. Over cold areas (Antarctic and Greenland plateau), results are consistent with the "cold ocean" analysis. Intercomparison of radiometer absolute calibrations is performed over the same continental area, leading to an anomalously high difference between channels 36.5 and 37 GHz of the ERS-2 and TOPEX radiometers, respectively, over "hot" targets (Sahara desert and Amazon forest). To quantify and analyze this difference, other radiometer measurements are analyzed over the Amazon forest, from the Special Sensor Microwave Imager (SSM/I) and the Advanced Microwave Sounding Unit (AMSU). Biases are confirmed for both TOPEX and ERS-2 radiometers by comparing brightness temperatures and derived surface emissivities: the TOPEX radiometer channels exhibit a negative bias with respect to SSM/I and AMSU-A, whereas the ERS-2 radiometer 36.5-GHz channel is positively biased, by several kelvin in brightness temperature in both cases. The method presented here could be used for controlling the in-flight calibration of any radiometer, and correct for remaining calibration errors after launch.

In this work, two microwave emissivity parameterizations are proposed to estimate the land emissivity at Advanced Microwave Sounding Units (AMSU) frequencies and scanning conditions in order to help processing AMSU measurements over land surfaces. Both parameterizations are derived from previously calculated land emissivities directly from satellite observations and take into account different surface types from bare soil to areas with high vegetation density. The first parameterization uses best-fit functions derived from February 2000 observations whereas the second parameterization is based on the first one with the addition of a mean nadir emissivity map at 23.8 GHz to allow a more precise surface description. The emissivity parameterizations have been evaluated by comparing emissivity and Tb simulations to target emissivity and Tb observations.

Atmospheric parameter retrievals over land from Advanced Microwave Sounding Unit (AMSU) measurements, such as atmospheric temperature and moisture profiles, could be possible using a reliable estimate of the land emissivity. The land surface emissivities have been calculated using six months of data, for 30 beam positions (observation zenith angles from -58/spl deg/ to +58/spl deg/) and the 23.8-, 31.4-, 50.3-, 89-, and 150-GHz channels. The emissivity calculation covers a large area including Africa, Eurasia, and Eastern South America. The day-to-day variability of the emissivity is less than 2% in these channels. The angular and spectral dependence of the emissivity is studied. The obtained AMSU emissivities are in good agreement with the previously derived SSMI ones. The scan asymmetry problem has been evidenced for AMSU-A channels. And possible extrapolation of the emissivity from window channels to sounding ones has been successfully tested.

Land surface emissivities at AMSU frequencies are estimated using collocated brightness temperatures, the International Satellite Cloud Climatology Project (ISCCP) data and the European Centre for Medium-Range Weather Forecasts (ECMWF) temperature-humidity profiles. The impact of some sources of errors is estimated. Angular dependence of the AMSU emissivities is examined. Preliminary AMSU emissivity maps are presented as well as comparison with SSMI derived emissivities and a microwave land emissivity model.