Land surface models provide self-consistent estimates of the water stored in various components of the land system, i.e. in the soil, vegetation and snow, and of water fluxes. Assimilation of satellite-based data helps to update these estimates to some extent, but it has some limitations when human processes, such as irrigation, are missing in the modeling system. Furthermore, the simulated water distribution heavily depends on the choice of meteorological input and other model choices. In this study, we aim to quantify the relative contribution of (i) Sentinel-1 data assimilation for soil moisture and snow updating, (ii) meteorological input, and (iii) modeling irrigation over the Po river basin in Italy. The Po river network channels the discharge of snow melt water from the Alps and Apennines, combined with surface and deep subsurface runoff from the hillslopes and valley. During the summer, the river network supplies irrigation water to the large agricultural area in the Po river valley. The Po basin is thus a unique testbed to study various water budget components in an environment with pronounced seasonal water storage dynamics and human water management. More specifically, we assimilate 1-km Sentinel-1 data into the Noah-MP land surface model coupled to an irrigation module and the hydrological modeling and analysis platform (HyMAP) as runoff routing module. The Noah-MP simulations are forced with meteorological data from either the fifth generation ECMWF atmospheric reanalysis (ERA5) or the Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) for the years 2015-2023. Sentinel-1 snow depth retrievals are assimilated over the mountains in the winter, whereas Sentinel-1 backscatter are mainly assimilated in the valley during the spring, summer and fall. The state updates are applied to snow depth, snow water equivalent, and soil moisture. These updates subsequently trigger updates in estimates of irrigation, leaf area index, discharge and other variables, resulting in a self-consistent re-analysis of the entire water budget of the Po basin. The impact of the Sentinel-1 data assimilation relative to that of the activation of irrigation modeling is quantified using independent in situ and remotely sensed measurements of soil moisture, leaf area index, snow depth, evaporation, irrigation and discharge.

Sled-based side-looking C-band radar profiles were collected around Fairbanks, Alaska, in March 2023 during the NASA SnowEx campaign to improve the conceptual understanding of C-band radar wave interactions with snow in a boreal forest environment. Seven transects with different vegetation and ground conditions were studied. Significant volume scattering from snow was observed in this shallow snowpack, indicating sensitivity at lower snow depths (SDs) which are common in high-latitude snowpacks. Manual removal of the snowpack decreased the backscatter by more than 2 dB in all polarizations, with a larger decrease in the cross-polarization, supporting the potential use of Sentinel-1 to retrieve SD.

Synthetic aperture radar will be at the forefront of future advancements in global remote sensing of snow depth and snow water equivalent. Recently, snow depth retrievals using an empirical volume scattering approach with C-band Sentinel-1 (S1) data have been demonstrated over the European Alps and Northern Hemisphere, with the most accurate results obtained in regions with dry, deep (>1.5 m) snowpacks and little vegetation influence. However, these S1-based snow depth retrievals have previously been compared only to point-based measurements or modeled snow depth products. In this study we develop an open-source version of the S1 snow depth retrieval technique and compare the results to spatially-distributed lidar snow depth measurements. The highly accurate and fine resolution lidar datasets were collected during the NASA Snow SnowEx 2020 and 2021 field campaigns at six study sites across the western United States. These regions represent different snow environments and characteristics than the datasets used for comparison in previous investigations. We compare the S1 and lidar snow depths at a range of spatial resolutions and interpret the results within the context of snowpack, vegetation, and terrain characteristics. At 90 m resolution, comparisons between lidar and S1 snow depth retrievals show low to moderate correlations (R = 0.38) and high RMSE (0.98 m) averaged across the study sites, with improved performance at 500 m resolution (R = 0.59, RMSE = 0.69 m). The distribution of S1 and lidar snow depths are more similar in regions of deeper snow, lower forest coverage, higher incidence angles, dry snow, and at coarser spatial resolutions. Our results highlight limitations of the current S1 snow depth algorithm and present opportunities to improve the technique for future snow depth retrievals across varied snow environments.

Snow depth retrievals from spaceborne C-band synthetic aperture radar (SAR) backscatter have the potential to fill an important gap in the remote monitoring of seasonal snow. Sentinel-1 (S1) SAR data have been used previously in an empirical algorithm to generate snow depth products with near-global coverage, subweekly temporal resolution and spatial resolutions on the order of hundreds of meters to 1 km. However, there has been no published independent validation of this algorithm. In this work we develop the first open-source software package that implements this Sentinel-1 snow depth retrieval algorithm as described in the original papers and evaluate the snow depth retrievals against nine high-resolution lidar snow depth acquisitions collected during the winters of 2019-2020 and 2020-2021 at six study sites across the western United States as part of the NASA SnowEx mission. Across all sites, we find agreement between the Sentinel-1 snow depth retrievals and the lidar snow depth measurements to be considerably lower than requirements placed for remotely sensed observations of snow depth, with a mean root mean square error (RMSE) of 0.92 m and a mean Pearson correlation coefficient r of 0.46. Algorithm performance improves slightly in deeper snowpacks and at higher elevations. We further investigate the underlying Sentinel-1 data for a snow signal through an exploratory analysis of the cross- to co-backscatter ratio (sigma VH/sigma VV; i.e., cross ratio) relative to lidar snow depths. We find the cross ratio increases through the time series for snow depths over similar to 1.5 m but that the cross ratio decreases for snow depths less than similar to 1.5 m. We attribute poor algorithm performance to (a) the variable amount of apparent snow depth signal in the S1 cross ratio and (b) an algorithm structure that does not adequately convert S1 backscatter signal to snow depth. Our findings provide an open-source framework for future investigations, along with insight into the applicability of C-band SAR for snow depth retrievals and directions for future C-band snow depth retrieval algorithm development. C-band SAR has the potential to address gaps in radar monitoring of deep snowpacks; however, more research into retrieval algorithms is necessary to better understand the physical mechanisms and uncertainties of C-band volume-scattering-based retrievals.

To better understand the interactions between C-band radar waves and snow, a tower-based experiment was set up in the Idaho Rocky Mountains for the period of 2021–2023. The experiment objective was to improve understanding of the sensitivity of Sentinel-1 C-band backscatter radar signals to snow. The data were collected in the time domain to measure the backscatter profile from the various snowpack and ground surface layers. The data show that scattering is present throughout the snow volume, although it is limited for low snow densities. Contrasting layer interfaces, ice features and metamorphic snow can have considerable impact on the backscatter signal. During snowmelt periods, wet snow absorbs the signal, and the soil backscatter becomes negligible. A comparison of the vertically integrated tower radar data with Sentinel-1 data shows that both systems have similar temporal behavior, and both feature an increase in backscatter during the dry-snow period in 2021–2022, even during weeks of nearly constant snow depth, likely due to morphological changes in the snowpack. The results demonstrate that C-band radar is sensitive to the dominant seasonal patterns in snow accumulation but that changes in microstructure, stratigraphy, melt–freeze cycles and snow wetness may complicate satellite-based snow depth retrievals.

Seasonal snow plays a crucial role in society and understanding trends in snow depth and mass is essential for making informed decisions about water resources and adaptation to climate change. However, quantifying snow depth in remote, mountainous areas with complex topography remains a significant challenge. The increasing availability of high-resolution synthetic aperture radar (SAR) observations from active microwave satellites has prompted opportunistic use of the data to retrieve snow depth remotely across large spatial and frequent temporal scales and at a high spatial resolution. Nevertheless, these novel SAR-based snow depth retrieval methods face their own set of limitations, including challenges for shallow snowpacks, high vegetation cover, and wet snow conditions. In response, here we introduce a machine learning approach to enhance SAR-based snow depth estimation over the European Alps. By integrating Sentinel-1 SAR imagery, optical snow cover observations, and topographic, forest cover and snow class information, our machine learning retrieval method more accurately estimates snow depth at independent in-situ measurement sites than current methods. Further, our method provides estimates at 100 m horizontal resolution and is capable of better capturing local-scale topography-driven snow depth variability. Through detailed feature importance analysis, we identify optimal conditions for SAR data utilization, thereby providing insight into future use of C-band SAR for snow depth retrieval.

High-resolution water budget estimates benefit from modeling of human water management and satellite data assimilation (DA) in river basins with a large human footprint. Utilizing the Noah-MP land surface model, in combination with an irrigation module, Sentinel-1 backscatter and snow depth observations, we produce a set of 0.7-km$^2$ digital water budget replicas of the Po river basin (Italy) for 2015-2023. The results demonstrate that irrigation modeling consistently improves the seasonal soil moisture variation and summer streamflow at all gauges in the valley after withdrawal of irrigation water from the streamflow (12\% error reduction relative to observed low summer streamflow), even if the basin-wide irrigation amount is underestimated. Sentinel-1 backscatter DA for soil moisture updating strongly interacts with irrigation modeling: when both are activated, the soil moisture updates are limited, and the simulated irrigation amounts are reduced. Backscatter DA systematically reduces soil moisture in the spring, which improves downstream spring streamflow. Assimilating Sentinel-1 snow depth retrievals over the surrounding Alps and Apennines further improves spring streamflow in a complementary way (2\% error reduction relative to observed high spring streamflow). Despite the seasonal improvements, irrigation modeling and Sentinel-1 backscatter DA cannot significantly improve short-term or interannual variations in soil moisture, irrigation results in a systematically prolonged high vegetation productivity, and snow depth DA only impacts the deep snowpacks. These findings help to advance the design and production of digital water budget replicas for river basins.

AbstractHigh‐resolution water budget estimates benefit from modeling of human water management and satellite data assimilation (DA) in river basins with a large human footprint. Utilizing the Noah‐MP land surface model with dynamic vegetation growth and river routing, in combination with an irrigation module, Sentinel‐1 backscatter and snow depth retrievals, we produce a set of 0.7‐km2water budget estimates of the Po river basin (Italy) for 2015–2023. The results demonstrate that irrigation modeling improves the seasonal soil moisture variation and summer streamflow at all gauges in the valley after withdrawal of irrigation water from the streamflow in postprocessing (12% error reduction relative to observed low summer streamflow), even if the basin‐wide irrigation amount is underestimated. Sentinel‐1 backscatter DA for soil moisture updating strongly interacts with irrigation modeling: when both are activated, the soil moisture updates are limited, and the simulated irrigation amounts are reduced. Backscatter DA systematically reduces soil moisture in the spring, which improves downstream spring streamflow. Assimilating Sentinel‐1 snow depth retrievals over the surrounding Alps and Apennines further improves spring streamflow in a complementary way (2% error reduction relative to observed high spring streamflow). Despite the seasonal improvements, irrigation modeling and Sentinel‐1 backscatter DA cannot significantly improve short‐term or interannual variations in soil moisture, irrigation modeling causes a systematically prolonged high vegetation productivity, and snow depth DA only impacts the deep snowpacks. This study helps advancing the design of digital water budget replicas for river basins.