In recent decades, the Earth Observation (EO) wave interaction modelling domain has witnessed a proliferation of both forward and inverse models. These models are developed by the scientific community to understand the relationship between electromagnetic waves and natural surfaces, and to support methodologies for extracting bio-geophysical variables from remotely sensed data. However, the current landscape exposes certain limitations such as the absence of systematic implementation, validation on limited datasets, and a scarce integration with emerging Artificial Intelligence (AI)-based inversion techniques. This manuscript introduces the Wave Interaction Models Exploitation Framework (WIMEX), developed to address these challenges in the frame of an ESA-funded project. Leveraging EO data available today, and exploiting Graphical Processing Unit and parallel computing, the framework proposes a systematic approach to create, validate, and disseminate forward and inverse models. WIMEX aims to offer a flexible development environment supporting the evolving needs of the scientific community.

Interferometric SAR (InSAR) is a promising tool for monitoring seasonal snow and for retrieving of Snow Water Equivalent (SWE) as the interferometric phase can be related to changes in SWE (ΔSWE). The boreal forest is a challenging landscape for the InSAR retrieval of SWE since it contributes to the signal by adding an undesired component originating from the vegetation. Although the technique has been validated extensively, most of these works are limited to discrete points. For comparison, we used snowpack simulations from the SnowModel, a high-resolution spatially distributed snow evolution model. This enables a better understanding of the limitations of L-band InSAR for SWE retrieval since it allows to evaluate its performance under different conditions. We analyzed the impact on coherence of snow melt between acquisitions and analyzed pairs with wet snow presence. The interferometric phase was inverted and compared to the simulated ΔSWEfrom the SnowModel distributions for three interferometric pairs. The results indicate a good spatial match between SnowModel and InSAR estimations. However, an increased difference was observed over densely forested areas when the air temperature was close to zero in at least one of the interferometric pairs. We hypothesize that the increase in permittivity of the forest for close to zero temperatures also increases the contribution from the canopy, consequently inducing errors in the retrieval. Both ALOS2 and SnowModel ΔSWEestimates were compared with in-situ data including a snow scale, snow depth from an Automatic Weather Station (AWS), a snow pit, and manual courses.

An investigation of boreal forest attenuation of a radar signal in winter is presented, applying a multifrequency (1–10 GHz) ground-based synthetic aperture radar (GB-SAR). As stable targets, corner reflectors (CRs) with known radar cross section (RCS) were used under the forest canopy. This enabled to relate changes in observed wideband backscattering from the reflectors to attenuation of the radar signal in forest vegetation, eliminating the influence of the background, such as snow and soil. We found that ambient temperature affected the observed attenuation of the radar signal in the entire 1–10-GHz frequency range. For temperatures$T < 0 ~^{\circ }\text{C}$, attenuation was found to decrease by up to 4.3 dB at the lowest observed temperatures of −36 °C, with peak attenuation occurring at$T \approx 0 ~^{\circ }\text{C}$. The overall apparent two-way attenuation increased by up to 18 dB from L- to X-band. The presence of snow on the canopy was found to increase attenuation by 1–4 dB, the effect increasing with frequency while having only negligible effects on vegetation backscatter.

Monitoring freeze/thaw variations of the Earth surfaces is of paramount importance for the study of biogeochemical processes and climate change. At present, the use of passive sensors is well established, but, very recently, some studies demonstrated the potentialities of observations exploiting signals of opportunity. We propose here an advanced study to demonstrate the capability of spaceborne Global Navigation Satellite System Reflectometry (GNSS-R) to provide accurate and systematic information about the Earth-surface freeze/thaw state. Reflectivity values derived from TechDemoSat-1 (TDS-1) data are elaborated and compared against the Soil Moisture and Ocean Salinity (SMOS) freeze/thaw product, while state-of-the-art land cover data are used to select GNSS-R data within an estimated footprint. In spite of the limited data availability due to sparse spatial coverage and calibration issues of TDS-1 observations, the proposed analysis demonstrates the possibility of monitoring the freeze/thaw state by analyzing the calibrated reflectivity, including also the possibility of detecting the transition state between frozen and thawed conditions across seasonal variations. This feature makes the design of next-generation GNSS-R satellite missions a unique opportunity to achieve high-resolution freeze/thaw monitoring with small and low-cost platforms.

Within the framework of European Space Agency (ESA) activities, several campaigns were carried out in the last decade with the purpose of exploiting the capabilities of multifrequency synthetic aperture radar (SAR) data to retrieve snow information. This article presents the results obtained from the ESA SnowSAR airborne campaigns, carried out between 2011 and 2013 on boreal forest, tundra and alpine environments, selected as representative of different snow regimes. The aim of this study was to assess the capability of X- and Ku-bands SAR in retrieving the snow parameters, namely snow depth (SD) and snow water equivalent (SWE). The retrieval was based on machine learning (ML) techniques and, in particular, of artificial neural networks (ANNs). ANNs have been selected among other ML approaches since they are capable to offer a good compromise between retrieval accuracy and computational cost. Two approaches were evaluated, the first based on the experimental data (data driven) and the second based on data simulated by the dense medium radiative transfer (DMRT). The data driven algorithm was trained on half of the SnowSAR dataset and validated on the remaining half. The validation resulted in a correlation coefficient$R \simeq 0.77$between estimated and target SD, a root-mean-square error (RMSE)$\simeq 13$cm, and bias = 0.03 cm. ANN algorithms specific for each test site were also implemented, obtaining more accurate results, and the robustness of the data driven approach was evaluated over time and space. The algorithm trained with DMRT simulations and tested on the experimental dataset was able to estimate the target parameter (SWE in this case) with$R =0.74$, RMSE = 34.8 mm, and bias = 1.8 mm. The model driven approach had the twofold advantage of reducing the amount ofin situdata required for training the algorithm and of extending the algorithm exportability to other test sites.

A tower-based global navigation satellite system reflectometry (GNSS-R) experiment is set up in an Arctic wetland environment for investigating the possibility of monitoring wetland inundation and freeze/thaw (FT) dynamics which are additionally impacted by snow on the ground. Effects of inundation, snow cover, and soil FT state on observed GNSS-R signal-to-noise ratios (SNRs) are analyzed for horizontal (H) and vertical (V) polarizations. A simple classification approach is suggested to detect the inundated, frozen, or thawed soil state. A simple forward reflectivity model is formulated to evaluate the influence of snow cover, overlying frozen, or thawed soil, on the reflected GNSS signals. Reflectivity time series are simulated in H- and V-polarizations usingin situobservations of the Arctic wetland site. The simulations are used to verify the tower-based observations, which show a significant impact of wet snow on reflectivity during melting conditions in spring. The observed SNR is strongly correlated with the Sentinel-1 backscatter coefficient. Generally, soil states detected by GNSS-R are in high agreement with ground truth soil states, especially for inundated and frozen soils. Wet snow conditions, however, complicate the correct timing estimation of soil thawing by inducing reflectivities of a similar order as thawing soil. It is recommended that GNSS-R land application models and retrieval algorithms consider snow cover effects to reduce false classification, especially in FT detection. Overall, the outcome of this study is relevant to the upcoming ESA HydroGNSS mission.

Monitoring freeze/thaw variations of the Earth surfaces is of great value for the study of biogeochemical processes and climate changes. Over the last decade, the use of passive sensors has been established and, more recently, some researches demonstrated the potential of exploiting observations based on signals of opportunity. We propose an advanced study to assess the capability of spaceborne Global Navigation Satellite System Reflectometry (GNSS-R) to give systematic information about the soil freeze/thaw state of high-latitude regions. To this aim, reflectivity values derived from TechDemoSat-1 (TDS-1) data are elaborated using state-of-art land cover data.

This paper presents a new approach for monitoring soil frost in the northern boreal forest region using co-polarized C-band synthetic aperture radar (SAR) data. Due to the high sensitivity of the C-band signal to vegetation, estimating the soil freeze/thaw (F/T) state directly from the measured backscatter is not feasible over dense vegetation, such as boreal forests. The presented method is based on applying a simple zeroth-order model to estimate the contribution of the ground and the forest canopy on the observed total backscatter. The retrieved ground and canopy backscatter values were compared with in situ information on soil F/T state. By using a linear least sum of square errors classification algorithm, the retrieved ground and canopy backscatter values representing frozen and thawed ground were successfully separated. The method was tested for various soil types and incidence angles. For soil types with higher water holding capacities and lower infiltration rates such as fine Haplic Podzol and Umbric Gleysol, the estimation accuracy of the F/T state was over 97%, whereas for drier, well-drained soil types such as Haplic Arenosol and Coarse Haplic Podzol it was over 94%. Estimation accuracy slightly increased with higher incidence angle. The method is not feasible in rocky terrain due to very low water content, or in wet snow conditions due to lack of penetration of the C-band SAR signal through wet snow. With low ancillary data and computational requirements, the proposed method is applicable for continuous near real-time monitoring of soil F/T state.

Reliable information on snow cover across the Northern Hemisphere and Arctic and sub-Arctic regions is needed for climate monitoring, for understanding the Arctic climate system, and for the evaluation of the role of snow cover and its feedback in climate models. In addition to being of significant interest for climatological investigations, reliable information on snow cover is of high value for the purpose of hydrological forecasting and numerical weather prediction. Terrestrial snow covers up to 50 million km2of the Northern Hemisphere in winter and is characterized by high spatial and temporal variability. Making satellite observations the only means for providing timely and complete observations of the global snow cover.

We describe the methodology for deriving yearly pixel-wise snow melt-off day maps from optical data-based FSC (Fractional Snow Cover) without conducting any interpolation for cloud-obscured pixels or otherwise missing data. The Copernicus CryoLand Pan-European FSC time series for 2001-2016 re-gridded to 0.1 ° serves as input for the production of 16 years of melt-off day maps for Europe. These maps are compared with passive microwave radiometer (MWR) melt retrievals. These independent datasets are evaluated against melt-off day derived from in situ snow depth (SD) time series observed at European weather stations. Our results show that the melt-off day derived from optical springtime FSC time series provides the best correlation with the snow melt-off day as indicated by in situ data. The obtained bias is 0.9 days, and RMSE is 13.1 days. For 85 % of the analyzed cases the differences are between ±10 days. Across Europe the MWR-based detection of melt-off day is less accurate, as the applied method performs the best for areas with sustained seasonal snow cover. Based on the time series 1980-2016 for MWR-based melt-off day, separately for boreal forests and tundra, we also found a clear trend towards earlier snow clearance: a decrease of melt-off day by as much as ~5 days per decade in boreal forests was observed.

The Finnish Meteorological Institute, together with Gamma Remote Sensing, Switzerland, has developed a global soil freeze/thaw detection algorithm using passive L-band microwave observations from the European Space Agency's Soil Moisture and Ocean Salinity (SMOS) mission. The current product gives the soil state as “frozen”, “partially frozen”, or “thawed”. Estimates for a given season are derived after each winter period. An operational product with a latency of one day is under development. Global information on soil freezing and thawing has many applications; e.g. in evaluation or as an input prior in carbon and climate models, soil carrying capacity analysis, and hydrological models.

Reliable information on snow cover across the Northern Hemisphere and Arctic and sub-Arctic regions is needed for climate monitoring, for understanding the Arctic climate system, and for the evaluation of the role of snow cover and its feedback in climate models. In addition to being of significant interest for climatological investigations, reliable information on snow cover is of high value for the purpose of hydrological forecasting and numerical weather prediction. Terrestrial snow covers up to 50 million km2of the Northern Hemisphere in winter and is characterized by high spatial and temporal variability. Making satellite observations the only means for providing timely and complete observations of the global snow cover.

There is a significant difference in SWE retrieval performance between the different satellite-based products. The assessment using the Russian and Finnish snow transect data covers an extremely large and varied geographical region and spans a total of ten years (2002–2011). Additionally, the reference data are well suited for assessing coarse resolution data, as they are not point-wise measurements but distributed measurements from the snow transects or snow courses.

Reliable global and regional scale SWE maps can be calculated by the assimilation of space borne derived SWE estimates and ground based SD observations. The spatial resolution of these products is ~25 km per pixel which is good enough for climate research but for hydrology a higher resolution is often optimal. A regional SWE processing system with nominal resolution of ~ 5 km per pixel over Europe is described in this paper. In addition the validation results show that the sensitivity to SWE is on the same level as with the lower resolution products. SWE data are also assimilated with HOPS hydrological model and the results show an improvement in river discharge estimates.

Different researches were addressed to the assessment of the boreal forest environment using active microwave remote sensing. Some of these activities were also devoted to estimate the ground parameters under the forest (i.e. soil moisture, snow mass) and, in order to understand the complex mechanisms which govern the radar backscattering, different electromagnetic models were developed for simulating the boreal scenario. An improvement of these models, for better characterizing the sparse forests, also considered the effect of shadow induced by the trees. Besides the computation of the attenuation caused by shadow on ground backscattering, the first needed parameter was the quantification of the percentage of the pixel affected by shadow. The estimation of this parameter can be obtained from high resolution optical images which are not always available at global scale. An alternative semi- empirical method based on 3D CAD modelling and ancillary information, which allow to quantify the amount of the shaded area is presented in the paper. Different forest profiles, height and densities and different geometry of observation were considered, and semi-empirical relationships between shadow area extension and these parameters were founded. The effect of electromagnetic shadowing was also quantified by performing model simulations. Finally a validation of the method was achieved by using high-resolution data collected from optical sensors in a forested area of Finland.

Satellite-based snow-cover monitoring is performed using optical, synthetic aperture radar (SAR), and passivemicrowave sensors. Effects of forest canopy on the observed signal need to be considered with all of these sensor types. Various models describing the interaction of electromagnetic radiation with forest canopy have been developed, but many of these are overly complex with high computational and ancillary data requirements. However, for retrieval purposes, simple models are preferred. This work aims at increasing the understanding of the effect of forest canopy on remote sensing observations of snow-covered terrain for both microwave and optical regimes and at quantifying the capability of simple zeroth-order models in simulating these effects. To achieve these goals, a spatial analysis of optical, SAR, and passive-microwave remote sensing data in the northern boreal forest region was performed. Model parameters for vegetation transmissivity as well as the properties of the underlying surface were optimized by utilizing lidar-ranging- and Landsat-based simplified proxy parameters describing forest canopy closure and stem volume. The results demonstrated that despite using these relatively simple proxies, a zeroth-order model can accurately estimate the extinction of electromagnetic signals in a forest, particularly for passive microwave and optical data. The SAR model successfully estimated the median of the observations, but larger scatter of the observations was reflected by a higher root mean square error and lower correlation between models and observations. Due to both good estimation accuracy and simplicity, the presented models can be considered to be applicable in existing snow retrieval algorithms.