Runoff from deep mountain snowpacks is the primary source of much-needed water for arid to semiarid regions in the western United States as well as in many other parts of the world. Each year, water managers in these regions must balance their water budgets, which account for water gained, lost, and stored in the watersheds they oversee, affecting everything from water supply to agriculture to tourism to wildfire containment.
To do so, water managers primarily rely on established statistical models that predict the volume and timing of mountain runoff. However, the information available to feed these models comes mainly from a sparse network of snow-monitoring weather stations, as well as from snow cover maps derived from optical satellite imagery that provide information on snow extent but not on the amount of water stored in the snowpack.
Managers of some basins, typically those home to watersheds that serve major population centers and agricultural producers, can also fund efforts to collect airborne high-resolution remotely sensed snow depth and snow mass estimations (e.g., from the Airborne Snow Observatories). These data significantly improve runoff models and streamflow forecasting for local water management and dam operations. However, the significant cost of these airborne surveys prevents many jurisdictions from accessing these types of data.
Detailed satellite snow volume and mass observations could give more water managers access to more complete information.
Data collected by satellites are more cost-effective and more frequent relative to airborne surveys. Therefore, detailed satellite snow volume and mass observations could give more water managers access to more complete information. For over 3 decades, researchers have developed snow remote sensing methods, working toward a satellite mission capable of sensing snow volume and mass—typically measured by snow depth and snow water equivalent, or SWE—at high spatial and temporal resolutions. Progress has been made, but amid ongoing warming-driven snowpack losses [Hale et al., 2023], there is still no funded global snow-focused satellite mission.
One way forward may involve the use of interferometric synthetic aperture radar (InSAR) to map changes in snowpacks. InSAR is commonly used in the geosciences to explore fault activity and volcanism through measurements of ground surface deformation. But the technique has been difficult to apply to snow because repeat intervals and radar wavelengths of current InSAR satellite platforms were not designed with snow retrievals in mind.
However, recent results from NASA’s 2017–2023 SnowEx campaign and the capabilities of the NASA–Indian Space Research Organisation SAR (NISAR) satellite mission—launched in late July 2025—spotlight InSAR’s potential as a novel, spaceborne snow remote sensing approach with high spatial resolution and near-global coverage. If this method is fully realized, high-resolution snow volume and mass measurements may be freely available for critical snow-dominant basins around the planet, with the potential to drastically improve water management sustainability practices. Such a resource could also enable scientific investigation within remote and inaccessible basins.
The NASA–Indian Space Research Organisation SAR (NISAR) satellite mission recently launched from India, as shown in the image at left. At right, the deployed satellite is shown above the western coast of the United States in this artist’s illustration. Credit: left,
ISRO; right,
NASA/JPL-Caltech
Measuring Snow with Radar
Numerous ground-based and airborne studies over the past 50 years have established that snow depth and snow mass can be calculated from the travel times of radar waves in snowpack. Radar signals span the microwave and radio wave portions of the electromagnetic spectrum and have much longer wavelengths than those used in optical imaging. Radar signals with wavelengths greater than 1 centimeter transmit through dry snowpacks, which contain no melted water, whereas wavelengths longer than 20 centimeters can penetrate both dry and wet snowpacks [e.g., Bradford et al., 2009]. However, spatial resolution and bandwidth limitations prevent direct measurements of signal travel times from space using conventional radar systems.
Synthetic aperture radar methods have found many applications for Earth observation, especially because radar signals pass through cloud cover and because they can be used at night.
On the other hand, SAR methods, which leverage the phase and amplitude of the returned radar signal, have found many applications for Earth observation, especially because radar signals pass through cloud cover and because they can be used at night. SAR uses Doppler effect principles to combine multiple overlapping radar observations from a wide-swath radar antenna to simulate a larger antenna aperture, enabling imaging at very high spatial resolution (<10 meters) and recording the amplitude and phase of backscattered radar signals. SAR methods using backscattered amplitudes or phases have been studied and developed for snow applications for more than 25 years [e.g., Shi and Dozier, 1997; Guneriussen et al., 2001].
InSAR detects the change in phase of radar signals between two SAR data acquisitions. Any snow accumulation between data acquisitions causes a phase change in backscattered signals because radar waves move slower in snowpack than in air. This change in radar phase represents a change in the signals’ travel times and can be used to estimate changes in SWE directly; together with an estimated snow density, it can also be used to estimate changes in snow depth (Figure 1) [Guneriussen et al., 2001].
Fig. 1. This illustration shows the interaction of a synthetic aperture radar (SAR) signal with a snow-free (left) and subsequently snow-covered (right) environment. The snow-covered illustration is representative of snowpacks up to a few meters deep. Accumulated snow causes the signal to refract and slow slightly, causing a delay in the time it takes the signal to return to the satellite, which can be used to estimate changes in snow water equivalent (SWE). For visual clarity, the respective paths of backscattered and forward-scattered signals are not shown.
Until recently, InSAR for snowpack detection saw little evaluation and development, primarily because in situ SWE observations, which are needed to validate the method, were not collected coincident with InSAR time series. Other factors included imprecise satellite orbital information that is problematic for processing InSAR data, the shortage of satellites sensing at longer wavelengths and their respective acquisition strategies, and the fact that SAR data were largely proprietary (these data have become accessible since the launch of Sentinel-1 in 2014).
Long periods of time between InSAR data acquisitions (e.g., several weeks to months) further complicate application of the method, because longer time intervals between observations result in less accurate or often unresolvable phase information. In addition, when large snow accumulations cause more than 360° of phase change in the backscattered signal, there is ambiguity in the resulting SWE and snow depth estimations.
Previous work has therefore shown that frequent and regular observations are required to measure sequential changes in phase and accurately detect changes in snowpack SWE (e.g., from accumulation, ablation, or redistribution) [Deeb et al., 2011]. To then estimate the total SWE of a snowpack, changes in SWE between sequential pairs of InSAR acquisitions must be added together (Figure 2), an approach recently demonstrated using InSAR data collected by Sentinel-1 every 6 days [Oveisgharan et al., 2024].
Fig. 2. SWE accumulation was measured during water year 2024 at the Grizzly Peak SNOTEL (snow telemetry) station in Colorado (left). SWE has been subsampled to 12-day intervals to illustrate how an SWE accumulation curve from NISAR might look. Background colors represent the studied feasibility of the L-band InSAR method throughout the snow season. The highest feasibility is expected for December through mid-April, when the snowpack is likely dry. Lower feasibility is expected during warmer months, when liquid water within the wetter snowpack absorbs the radar signal energy. As measured using InSAR, snow accumulation or ablation events cause phase changes (i.e., changes in the signal path length or travel time) in the detected signals. The plot at right provides an idealized and simplified example of what those phase changes (φsnow) might look like based on the SWE accumulation and ablation shown at left.
SnowEx-UAVSAR Puts InSAR to the Test
NASA’s SnowEx campaign served as a testing ground for many of the leading snow remote sensing methodologies, including interferometric SAR (InSAR).
NASA’s SnowEx campaign served as a testing ground for many of the leading snow remote sensing methodologies, including InSAR. SnowEx partnered with the NASA Jet Propulsion Laboratory Uninhabited Aerial Vehicle SAR (UAVSAR) program to collect airborne InSAR imagery over SnowEx field sites during 2017, 2020, and 2021 (Figure 3). (The UAVSAR was originally intended to fly on an autonomous aircraft, hence its name, but is instead flown in a piloted aircraft.)
Fig. 3. Data collection sites were located across the U.S. West. Each labeled site saw at least one pair of Uninhabited Aerial Vehicle SAR (UAVSAR) flights (white boxes). Locations of sites with ground-based radar measurements and SNOTEL/CDEC (California Data Exchange Center) stations, which provided complementary ground-based data, are indicated by red markers and pink dots, respectively. Credit: 2020–2021 NASA SnowEx Experimental Plan
The UAVSAR aircraft flies at about 12-kilometer altitude, carrying a SAR instrument that emits signals over an approximately 15-kilometer swath width, with a spatial resolution of about 5 meters and a wavelength of about 24 centimeters, which is within the L-band radar wavelength range. L-band radar waves are long enough to penetrate deep snowpacks (with minimal scattering in the snowpack) and some forest canopies, with the trade-off that the longer wavelength reduces sensitivity for mapping small snow accumulations or small wind redistribution events.
In February 2017, NASA SnowEx conducted airborne and ground campaigns, including UAVSAR flights, at sites in Grand Mesa and in Senator Beck Basin in western Colorado. The UAVSAR instrument was flown over each site on five dates from February to March. Direct evaluation of the repeat-pass L-band InSAR approach was not possible because the field campaign strategy was designed for evaluating other remote sensing methods. Still, the phase-change measurements were valuable for predicting snow depths with a machine learning algorithm, because the measured changes in SWE had a very similar spatial pattern to the total measured snow depth [Alabi et al., 2025].
On the basis of these early results, UAVSAR flew at weekly to biweekly intervals from January through March of 2020 and 2021 over 13 field sites in the mountains of the western United States and one site in Montana’s prairies. Accompanying ground campaigns emphasized repeat observations at specific locations to better evaluate InSAR measurements of SWE and snow depth changes. At each site, researchers collected a unique set of ground observations. At some, for example, they emphasized snow pit and snow depth collections, whereas at others the focus was on ground-based radar collections. To provide a more spatially expansive dataset for InSAR evaluation, airborne lidar snow depths were also collected at select sites.
These studies also demonstrated the utility of InSAR for mapping snowpacks over a variety of landscapes.
Four UAVSAR studies were conducted in mountain ranges with continental climates (characterized by hot summers and cold winters), where snowpacks are relatively shallow. At Grand Mesa, Colorado, InSAR snow depth and SWE change measurements were evaluated against spatially distributed airborne lidar snowpack measurements. Marshall et al. [2021] showed that InSAR snow measurements can be remarkably accurate in flat terrain and dry snow conditions.
Studies over 3-month periods in the mountains of northern Colorado further support the accuracy of InSAR-based findings, particularly during the accumulation season when snowpacks are dry [Bonnell et al., 2024a, 2024b]. These studies also demonstrated the utility of InSAR for mapping snowpacks over a variety of landscapes, including densely vegetated wetland meadows, severely burned forest stands, steep topography, and coniferous forests with low to moderate forest coverage.
A study in the Valles Caldera of New Mexico used InSAR to map snow accumulation and ablation early in the snowmelt season and found that the ablation patterns resembled snow losses observed in coincident optical imagery [Tarricone et al., 2023]. Until this study, measuring SWE with InSAR during this part of the snow season was considered infeasible because it was thought that wet snow would absorb and attenuate the radar signal too much.
Another two studies evaluated the InSAR method for snowpacks in the mountains of Idaho and in a Montana prairie. Idaho’s mountain snowpacks are classified as intermountain, which means they are generally deeper than continental snowpacks but shallower than maritime snowpacks (e.g., in California’s Sierra Nevada). Compared with continental mountain ranges, the intermountain climate regime also tends to be warmer, so midwinter snowmelt events are more common, though the snowpack remains colder and drier than maritime snow for much of the winter. The UAVSAR study in Idaho showed that L-band InSAR estimates generally agreed with manual SWE measurements and modeled SWE estimates at higher elevations. However, at lower elevations, InSAR SWE measurements had larger uncertainties where wet snow was identified [Hoppinen et al., 2024].
Prairie snowpacks, including those in Montana, can be intermittent, with winds scouring away snow in some areas and redistributing it into deep snowdrifts elsewhere. Palomaki and Sproles [2023] found that InSAR snow measurements had increased uncertainty where the ground was only partly covered by snow.
From SnowEx to NISAR
The NASA SnowEx campaign has enabled significant advances in developing a remotely sensed InSAR approach for measuring snowpacks. However, more work is needed to determine the approach’s suitability across environments, and it is not expected to work everywhere in all snow conditions. The presence of liquid water within snowpack is the biggest inhibiting factor, so it is uncertain how well L-band InSAR can handle wet maritime snowpacks, regions that accumulate snow near its melting point, and the spring snowmelt period. Although the method appears to work with high accuracy in some forests, it also remains to be seen whether it can be adapted for high-density forests.
Through these NASA SnowEx InSAR studies, the method appears successful for estimating SWE in areas covered by dry snowpacks that persist throughout the winter. Thus, it has applications in many critical snow-dominated basins. If widely applied, it could dramatically expand our understanding of seasonal snow dynamics around the world and aid prediction of melt season streamflow.
The NISAR satellite mission has attributes that could help achieve the goal of applying InSAR for snow water resources globally.
The NISAR satellite mission has attributes that could help achieve the goal of applying InSAR for snow water resources globally. First, like UAVSAR, NISAR will use an L-band radar signal, potentially allowing for accurate observations of phase changes over some forested areas and from deep snowpacks. Second, NISAR will have an exact revisit period of 12 days. This period is longer than the 7-day revisit period often tested during the SnowEx campaign but should be short enough to produce high-quality SWE measurements across many snow climates. Third, the Alaska Satellite Facility, which will distribute NISAR data, will provide InSAR datasets at 80-meter resolution within 2 days of acquisition, timely enough for water management decisions.
Unfortunately, the method’s potential was not demonstrated until after the NISAR science plan was developed, so the mission’s science objectives do not include seasonal snow measurements and a standard snow product will not be released. Also, although the 2020–2021 SnowEx-UAVSAR studies served as a partial proof of concept for satellite InSAR snow monitoring, the higher imaging altitude of NISAR could raise additional complications that will need to be studied and addressed. For example, NISAR will have lower-resolution imaging capabilities than the airborne UAVSAR platform, and the higher imaging altitude will introduce additional atmospheric and ionospheric artifacts in the satellite observations, some of which the NISAR team will attempt to estimate and remove.
Despite these obstacles, the results of SnowEx and the availability of NISAR data (plus the upcoming launches of other L-band SAR satellites such as ROSE-L (Radar Observing System for Europe in L-band) and the development of SWE mapping methods using higher radar frequencies) show that modern radar techniques are lighting the path to the future of global snowpack monitoring. To progress on this path, cross-disciplinary collaborations involving snow researchers, radar experts, data scientists, and, importantly, local water managers must continue evaluating and harnessing InSAR’s potential to detect changing snowpacks and inform water management decisions that affect people and habitats around the world.
Acknowledgments
We thank the participants, coordinators, and site leaders of the NASA SnowEx campaign and the NASA UAVSAR team, particularly Yunling Lou and Yang Zheng. Much of this research culminated from collaborations in the NASA L-band InSAR Snow Working Group (2021 to present) and the open-science tools developed during the NASA SnowEx Hackweeks (2021–2023). In particular, we acknowledge the efforts of Zach Hoppinen, Ross Palomaki, Shadi Oveisgharan, Ibrahim Alabi, Dan McGrath, Ryan Webb, Kelly Elder, Eric Sproles, Rick Forster, and Anne Nolin. We also acknowledge InSAR tower-based and satellite-borne studies that were produced in tandem with the SnowEx campaigns by Jorge Ruiz and Juha Lemmetyinen. Finally, we thank John Hammond and John Fulton for their constructive feedback. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. government.
References
Alabi, I. O., et al. (2025), Advancing terrestrial snow depth monitoring with machine learning and L-band InSAR data: A case study using NASA’s SnowEx 2017 data, Front. Remote Sens., 5, 1481848, https://doi.org/10.3389/frsen.2024.1481848.
Bonnell, R., et al. (2024a), L-band InSAR snow water equivalent retrieval uncertainty increases with forest cover fraction, Geophys. Res. Lett., 51(24), e2024GL111708, https://doi.org/10.1029/2024GL111708.
Bonnell, R., et al. (2024b), Evaluating L-band InSAR snow water equivalent retrievals with repeat ground-penetrating radar and terrestrial lidar surveys in northern Colorado, Cryosphere, 18(8), 3,765–3,785, https://doi.org/10.5194/tc-18-3765-2024.
Bradford, J. H., J. T. Harper, and J. Brown (2009), Complex dielectric permittivity measurements from ground-penetrating radar data to estimate snow liquid water content in the pendular regime, Water Resour. Res., 45(8), W08403, https://doi.org/10.1029/2008WR007341.
Deeb, E. J., R. R. Forster, and D. L. Kane (2011), Monitoring snowpack evolution using interferometric synthetic aperture radar on the North Slope of Alaska, USA, Int. J. Remote Sens., 32(14), 3,985–4,003, https://doi.org/10.1080/01431161003801351.
Guneriussen, T., et al. (2001), InSAR for estimation of changes in snow water equivalent of dry snow, IEEE Trans. Geosci. Remote Sens., 39(10), 2,101–2,108, https://doi.org/10.1109/36.957273.
Hale, K. E., et al. (2023), Recent decreases in snow water storage in western North America, Commun. Earth Environ., 4(1), 170, https://doi.org/10.1038/s43247-023-00751-3.
Hoppinen, Z., et al. (2024), Snow water equivalent retrieval over Idaho–Part 2: Using L-band UAVSAR repeat-pass interferometry, Cryosphere, 18, 575–592, https://doi.org/10.5194/tc-18-575-2024.
Marshall, H. P., et al. (2021), L-band InSAR depth retrieval during the NASA SnowEx 2020 campaign: Grand Mesa, Colorado, in 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS, pp. 625–627, https://doi.org/10.1109/IGARSS47720.2021.9553852.
Oveisgharan, S., et al. (2024), Snow water equivalent retrieval over Idaho–Part 1: Using Sentinel-1 repeat-pass interferometry, Cryosphere, 18(2), 559–574, https://doi.org/10.5194/tc-18-559-2024.
Palomaki, R. T., and E. A. Sproles (2023), Assessment of L-band InSAR snow estimation techniques over a shallow, heterogeneous prairie snowpack, Remote Sens. Environ., 296, 113744, https://doi.org/10.1016/j.rse.2023.113744.
Shi, J., and J. Dozier (1997), Mapping seasonal snow with SIR-C/X-SAR in mountainous areas, Remote Sens. Environ., 59(2), 294–307, https://doi.org/10.1016/S0034-4257(96)00146-0.
Tarricone, J., et al. (2023), Estimating snow accumulation and ablation with L-band interferometric synthetic aperture radar (InSAR), Cryosphere, 17(5), 1,997–2,019, https://doi.org/10.5194/tc-17-1997-2023.
Author Information
Randall Bonnell (rbonnell@usgs.gov), U.S. Geological Survey, Denver, Colo.; Jack Tarricone, NASA Goddard Space Flight Center, Greenbelt, Md.; Hans-Peter Marshall, Boise State University, Boise, Idaho; Elias Deeb, U.S. Army Corps of Engineers, Hanover, N.H.; and Carrie Vuyovich, NASA Goddard Space Flight Center, Greenbelt, Md.
Citation: Bonnell, R., J. Tarricone, H.-P. Marshall, E. Deeb, and C. Vuyovich (2025), Satellite radar advances could transform global snow monitoring,
Eos, 106, https://doi.org/10.1029/2025EO250476. Published on 24 December 2025.
Text © 2025. The authors.
CC BY-NC-ND 3.0Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.
