Snow Estimates Research Articles

A Spatially‐Distributed Machine Learning Approach for Fractional Snow Covered Area Estimation

AbstractSnowpack in mountainous areas often provides water storage for summer and fall, especially in the Western United States. In situ observations of snow properties in mountainous terrain are limited by cost and effort, impacting both temporal and spatial sampling, while remote sensing estimates provide more complete spacetime coverage. Spatial estimates of fractional snow covered area (fSCA) at 30m are available every 16 days from the series of multispectral scanning instruments on Landsat platforms. Daily estimates at 463m spatial resolution are also available from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument on the Terra satellite. Fusing Landsat and MODIS fSCA images creates high resolution daily spatial estimates of fSCA that are needed for various uses: to support scientists and managers interested in energy and water budgets for water resources and to understand the movement of animals in a changing climate. Here, we propose a new machine learning approach conditioned on MODIS fSCA, as well as a set of physiographic features, and fit to Landsat fSCA over a portion of the Sierra Nevada USA. The predictions are daily 30m fSCA. The approach relies on two stages of spatially‐varying models. The first classifies fSCA into three categories and the second yields estimates within (0, 100) percent fSCA. Separate models are applied and fitted within sub‐regions of the study domain. Compared with a recently‐published machine learning model (Rittger, Krock, et al., 2021), this approach uses spatially local (rather than global) random forests, and improves the classification error of fSCA by 16%, and fractionally‐covered pixel estimates by 18%.

Arctic Freeboard and Snow Depth From Near‐Coincident CryoSat‐2 and ICESat‐2 (CRYO2ICE) Observations: A First Examination of Winter Sea Ice During 2020–2022

AbstractIn the summer of 2020, ESA changed the orbit of CryoSat‐2 to align periodically with NASA's ICESat‐2 mission, a campaign known as CRYO2ICE, which allows for near‐coincident CryoSat‐2 and ICESat‐2 observations in space and time over the Arctic until summer 2022, where the CRYO2ICE Antarctic campaign was initiated. This study investigates the Arctic CRYO2ICE radar and laser freeboards acquired by CryoSat‐2 and ICESat‐2, respectively, during the winter seasons of 2020–2022, and derives snow depths from their differences along the orbits. Along‐track snow depth observations can provide high‐resolution snow depth distributions which are vital for air‐ice‐ocean heat and momentum transfer, understanding light transmission, and snow‐ice‐interactions. Generally, ICESat‐2 is backscattered at a surface above the elevation of the CryoSat‐2 signal. CRYO2ICE snow depths are thinner than the daily model‐ or passive‐microwave‐based snow depth composites used for comparison, with differences being most pronounced in the Atlantic and Pacific Arctic. Satellite‐derived and model‐based snow estimates show similar seasonal accumulation over first‐year ice, but CRYO2ICE has limited seasonal accumulation over multi‐year ice which is linked to a slow increase in ICESat‐2, and to some extent CryoSat‐2, freeboards. We present a first estimation of spaceborne along‐track snow depth estimates with average uncertainty of 10–11 ± 2–3 cm for 7‐km segments, with random and systematic contributions of 7 and 4 cm. These observations show the potential for along‐track dual‐frequency observations of snow depth from the future Copernicus mission CRISTAL; but they also highlight uncertainties in radar penetration and the correlation length scales of snow topography that still require further research.

Assessment of snow simulation using Noah-MP land surface model forced by various precipitation sources in the Central Tianshan Mountains, Central Asia

Accurate mountainous snow estimation is paramount for hydrological processes and water resources estimation in arid regions. Using Land surface models (LSMs) is a practical numerical approach for snow estimation in a complex orography region such as the Tianshan Mountains. However, the bias of precipitation forcing is a major source of uncertainty for snow simulation over scarce-data regions. This study evaluated the performance of 4 km snow simulations using the Noah-MP LSM driven by eight precipitation sources. They are retrieved from non-model products and climate model simulations during the 2018-2019 cold season in the Central Tianshan Mountains (CTS) region. Multi-source validation datasets (in-situ observation, field snow pits, and remote sensing products) have been used for evaluation and uncertainty estimation. The results showed that the difference in precipitation amount significantly affected the snowpack simulation performance. Compared with non-model products (Global Land Data Assimilation System (GLDAS), Global Precipitation Measurement (GPM), and Gauge-adjusted Global Satellite Mapping of Precipitation (GSMaP)), the cold season precipitation from climate model simulations exhibited a better performance overall in the high-elevation regions (elevation > 1000 m) evaluated by in-situ observations. In addition, the 4 km convection-permitting modeling (CPM) precipitation (WRF-Morrison and WRF-WSM6) showed higher accuracy (RMSE: 23.48 mm/season and 18.94 mm/season, respectively) than gray-zone resolution simulations and ERA5-land driven runs in the high-elevation regions. The snow depth simulation driven by WRF-Morrison precipitation had the second smallest RMSE (4.67 cm/day) and lowest bias (-0.74 cm/day) value in the high-elevation regions compared with in-situ observation. Meanwhile, CPM precipitation-driven runs exhibited the smallest RMSE value based on the assessment of snow field pits. In addition, the CPM-driven simulation achieved the closest match to the elevation-based distribution of snow cover days in regions with elevation over 1000 m and duration over 60 days compared to the MODIS product. The findings of this study highlight that CPM with proper parameterization configurations has added values in producing realistic topographic precipitation for snow modeling using LSMs over data-scarcity mountainous areas.

Testing a novel sensor design to jointly measure cosmic-ray neutrons, muons and gamma rays for non-invasive soil moisture estimation

Abstract. Cosmic-ray neutron sensing (CRNS) has emerged as a reliable method for soil moisture and snow estimation. However, the applicability of this method beyond research has been limited due to, among others, the use of relatively large and expensive sensors. This paper presents the tests conducted on a new scintillator-based sensor especially designed to jointly measure neutron counts, muons and total gamma rays. The neutron signal is first compared against two conventional gas-tube-based CRNS sensors at two locations. The estimated soil moisture is further assessed at four agricultural sites, based on gravimetric soil moisture collected within the sensor footprint. Muon fluxes are compared to the incoming neutron variability measured at a neutron monitoring station and total gammas counts are compared to the signal detected by a gamma ray spectrometer. The results show that the neutron dynamic detected by the new scintillator-based CRNS sensor is well in agreement with conventional CRNS sensors. The derived soil moisture also agreed well with the gravimetric soil moisture measurements. The muons and the total gamma rays simultaneously detected by the sensor show promising features to account for the incoming variability and for discriminating irrigation and precipitation events, respectively. Further experiments and analyses should be conducted, however, to better understand the accuracy and the added value of these additional data for soil moisture estimation. Overall, the new scintillator design shows to be a valid and compact alternative to conventional CRNS sensors for non-invasive soil moisture monitoring and to open the path to a wide range of applications.

Assimilation of airborne gamma observations provides utility for snow estimation in forested environments

Abstract. An airborne gamma-ray remote-sensing technique provides a strong potential to estimate a reliable snow water equivalent (SWE) in forested environments where typical remote-sensing techniques have large uncertainties. This study explores the utility of assimilating the temporally (up to four measurements during a winter period) and spatially sparse airborne gamma SWE observations into a land surface model (LSM) to improve SWE estimates in forested areas in the northeastern US. Here, we demonstrate that the airborne gamma SWE observations add value to the SWE estimates from the Noah LSM with multiple parameterization options (Noah-MP) via assimilation despite the limited number of measurements. Improvements are witnessed during the snow accumulation period, while reduced skills are seen during the snowmelt period. The efficacy of the gamma data is greater for areas with lower vegetation cover fraction and topographic heterogeneity ranges, and it is still effective at reducing the SWE estimation errors for areas with higher topographic heterogeneity. The gamma SWE data assimilation (DA) also shows a potential to extend the impact of flight-line-based measurements to adjacent areas without observations by employing a localization approach. The localized DA reduces the modeled SWE estimation errors for adjacent grid cells up to 32 km distance from the flight lines. The enhanced performance of the gamma SWE DA is evident when the results are compared to those from assimilating the existing satellite-based SWE retrievals from the Advanced Microwave Scanning Radiometer 2 (AMSR2) for the same locations and time periods. Although there is still room for improvement, particularly for the melting period, this study shows that the gamma SWE DA is a promising method to improve the SWE estimates in forested areas.

Assessment of L-band InSAR snow estimation techniques over a shallow, heterogeneous prairie snowpack

Snow water equivalent (SWE) is a critical input for weather, climate, and water resource management models at local to global scales. Despite its importance, global SWE measurements that are accurate, consistent, and at sufficiently high spatiotemporal resolutions are not currently available. L-band interferometric synthetic aperture radar (InSAR) techniques have been used to measure SWE at local to regional scales, and two upcoming L-band SAR satellite missions have renewed interest in these techniques to provide regular SWE measurements at the global scale. However, previous research demonstrating the capabilities of L-band InSAR-SWE measurement has been limited to mountain or tundra snowpack regimes. Here we examine the feasibility of applying the same techniques over a prairie snowpack, which are typically characterized by shallow snow depths (mean snow depth of 0.22 m in this study), exposed agricultural vegetation, and high spatial variability over short distances. Our study area in central Montana, USA (47.060, -109.951) was a validation site for NASA SnowEx 2021, as part of the UAVSAR snow timeseries. Airborne L-band SAR imagery was acquired by the UAVSAR platform while concurrent snow measurements were collected using uncrewed aerial vehicle (UAV)-based LiDAR, UAV-based photogrammetry, and ground-based manual techniques. This validation dataset enables an investigation of the effects of sub-pixel snow cover heterogeneity and exposed agricultural vegetation stubble on SAR data and the resulting SWE estimations. Results based on repeated application of the Kolmogorov–Smirnov test show that UAVSAR VV phase change is sensitive to differences in snow cover but relatively unaffected by differences in agricultural stubble height. However, we did not find similarly definitive results when we used the same phase change data to estimate SWE. Although broad spatial patterns were similar in both LiDAR-derived and InSAR-derived SWE estimates, considerable differences in the two estimates were apparent in areas with large sub-pixel snow depth variability. Our results indicate that additional work is necessary to derive accurate SWE estimates in prairie environments. Regular measurements from L-band SAR satellites will provide an excellent opportunity to refine InSAR-based snow estimation techniques over shallow, heterogeneous snowpacks.

Snow coverage estimation using camera data for automated driving applications

In the U.S., over 38,000 fatalities occur every year due to automotive accidents where 24% of these accidents are attributable to inclement weather. Automated driving systems have shown to decrease up to 21% of potential collisions, however, these systems do not operate in inclement weather. The camera’s reliance on clear lane line detections cease the functionality of the safety systems when occlusions occur due to precipitation. For these systems to become operational during conditions such as snow coverage, therefore leading to a greater impact on safety, new research and development is needed to focus on inclement weather scenarios. This study addresses this need by first collecting a new dataset consisting of raw camera images along arterial roads in Kalamazoo, MI and additionally collecting snow precipitation data from the National Center for Environmental Information. With this data, snow coverage estimation models were developed to automatically determine categories of snow coverage. The models were developed by investigating various machine learning algorithm types, image predictors, and the presence of snow precipitation data. The final model resulted in 95.63% accuracy for categorizing the instance as either none, standard, or heavy snow coverage. These categories are important for future development of purpose-build algorithms that identify drivable regions in various levels of snow coverage for future automated driving systems. The results demonstrate that snow estimation is a near-term achievable task and that the presence weather data improves accuracy. With the addition of snow-coverage estimation, automated driving systems can be developed to react to these different conditions respectively and further reduce the nearly 6,000 annual fatalities caused driving in adverse weather.

Combination of Snow Process Model Priors and Site Representativeness Evaluation to Improve the Global Snow Depth Retrieval Based on Passive Microwaves

The spatiotemporal distribution of snow depth (SD) has a significant impact on the energy and water balances of the Earth’s system. However, passive microwave remote sensing widely used for SD estimation has large uncertainties due to the variations in snow physical properties. In this study, we demonstrate a new method to minimize these uncertainties and to increase the accuracy of SD estimation. Our method is based on the synergy between the passive microwave AMSR-2 brightness temperature (TB) and a physical snow process model (SNTHERM) to estimate snow grain size, snow density and first-guess SD as priors. On one hand, we used TB from three frequencies and removed non-representative ground measurements at the stations to improve deep snow estimation. Then, we applied a machine learning (ML) algorithm based on both the AMSR-2 TB and the SNTHERM simulations to retrieve the global SD. The results showed that the root-mean-squared error (RMSE) of the retrieved SD was 12.4 cm at the meteorological stations. Independent validations showed that our method significantly reduced the SD and snow water equivalent (SWE) underestimation in the mountains compared to the current satellite products.

A western United States snow reanalysis dataset over the Landsat era from water years 1985 to 2021

Water stored in mountain snowpacks (i.e., snow water equivalent, SWE) represents an important but poorly characterized component of the terrestrial water cycle. The Western United States snow reanalysis (WUS–SR) dataset is novel in its combination of spatial resolution (~500 m), spatial extent (31°–49° N; 102°–125° W), and temporal continuity (daily over 1985–2021). WUS–SR is generated using a Bayesian framework with model-based snow estimates updated through the assimilation of cloud-free Landsat fractional snow-covered area observations. Over the WUS, the peak SWE verification with independent in situ measurements show correlation coefficient, mean difference (MD), and root mean squared difference (RMSD) of 0.77, −0.15 m, and 0.28 m, respectively. The effects of forest cover and Landsat image availability on peak SWE are assessed. WUS–SR peak SWE is well correlated (ranging from 0.75 to 0.91) against independent lidar-derived SWE taken near April 1st, with MD <0.15 m and RMSD <0.38 m. The dataset is useful for characterizing WUS mountain snow storage, and ultimately for improving snow-derived water resources management.

Development of a novel approach for snow wetness estimation using hybrid polarimetric RISAT-1 SAR datasets in North-Western Himalayan region

Snow wetness estimates are critical inputs in the understanding of snow hydrological processes. With early warming in winters in the Himalayas, snowpack shows early signs of melting in February with increased snow wetness. Typically, in February, the snowpack comprises a mix of dry and wet snow layers. In the literature, methods for quantitative analysis of snow wetness are mostly based on fully polarimetric synthetic aperture radar (SAR) data. Methods based on the hybrid polarimetric SAR (PolSAR) data for snow monitoring are virtually inexistent. This study proposes a novel methodology for the estimation of snow wetness utilizing the C-band hybrid polarimetric RISAT-1 SAR dataset. Using radar remote sensing to analyze the behavior of such a snowpack requires information on the surface and volume scattering characteristics. The modeled generalized surface and volume scattering parameters (α), and (γ), based on the X-Bragg’s reflection coefficients and Fresnel transmission coefficients, are used for the inversion of surface and volume snow permittivities, respectively. The investigations are carried out for February 2014 for a study area in the Manali region in Himachal Pradesh, India. The retrieved snow estimates showed a coefficient of determination of 0.86 and a root mean square error of 0.667 with respect to in-situ measurements. Further, it was observed that the snow wetness estimates derived from the proposed method using RISAT-1 dataset outperformed the estimates based on fully polarimetric RADARSAT-2 dataset using the conventional Shi and Dozier method.

Snow water equivalent change mapping from slope-correlated synthetic aperture radar interferometry (InSAR) phase variations

Abstract. Area-based measurements of snow water equivalent (SWE) are important for understanding earth system processes such as glacier mass balance, winter hydrological storage in drainage basins, and ground thermal regimes. Remote sensing techniques are ideally suited for wide-scale area-based mapping with the most commonly used technique to measure SWE being passive microwave, which is limited to coarse spatial resolutions of 25 km or greater and to areas without significant topographic variation. Passive microwave also has a negative bias for large SWE. Another method is repeat-pass synthetic aperture radar interferometry (InSAR) that allows measurement of SWE change at much higher spatial resolution. However, it has not been widely adopted because (1) the phase unwrapping problem has not been robustly addressed, especially for interferograms with poor coherence, and (2) SWE change maps scaled directly from repeat-pass interferograms are not an absolute measurement but contain unknown offsets for each contiguous coherent area. We develop and test a novel method for repeat-pass InSAR-based dry-snow SWE estimation that exploits the sensitivity of the dry-snow refraction-induced InSAR phase to topographic variations. The method robustly estimates absolute SWE change at spatial resolutions of &lt; 1 km without the need for phase unwrapping. We derive a quantitative signal model for this new SWE change estimator and identify the relevant sources of bias. The method is demonstrated using both simulated SWE distributions and a 9-year RADARSAT-2 (C-band, 5.405 GHz) spotlight-mode dataset near Inuvik, Northwest Territories (NWT), Canada. SWE results are compared to in situ snow survey measurements and estimates from ERA5 reanalysis. Our method performs well in high-relief areas, thus providing complementary coverage to passive-microwave-based SWE estimation. Further, our method has the advantage of requiring only a single wavelength band and thus can utilize existing spaceborne synthetic aperture radar systems.

Polarimetric Radar Quantitative Precipitation Estimation

Radar quantitative precipitation estimation (QPE) is one of the primary tasks of weather radars. The QPE quality was substantially improved after polarimetric upgrade of the radars. This study provides an overview of existing polarimetric methodologies for rain and snow estimation and their operational implementation. The variability of drop size distributions (DSDs) is a primary factor affecting the quality of rainfall estimation and its impact on the performance of various radar rainfall relations at S, C, and X microwave frequency bands is one of the focuses of this review. The radar rainfall estimation algorithms based on the use of specific attenuation A and specific differential phase KDP are the most efficient. Their brief description is presented and possible ways for their further optimization are discussed. Polarimetric techniques for the vertical profile of reflectivity (VPR) correction at longer distances from the radar are also summarized. Radar quantification of snow is particularly challenging and it is demonstrated that polarimetric methods for snow measurements show good promise. Finally, the article presents a summary of the latest operational radar QPE products available in the US by integration of the information from the WSR-88D radars via the Multi-Radar Multi-Sensor (MRMS) platform.

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.

Development of a “nature run” for observing system simulation experiments (OSSEs) for snow mission development

Abstract Snow is a fundamental component of global and regional water budgets, particularly in mountainous areas and regions downstream that rely on snowmelt for water resources. Land surface models (LSMs) are commonly used to develop spatially distributed estimates of snow water equivalent (SWE) and runoff. However, LSMs are limited by uncertainties in model physics and parameters, among other factors. In this study, we describe the use of model calibration tools to improve snow simulations within the Noah-MP LSM as the first step in an Observing System Simulation Experiment (OSSE). Noah-MP is calibrated against the University of Arizona (UA) SWE product over a Western Colorado domain. With spatially varying calibrated parameters, we run calibrated and default Noah-MP simulations for water years 2010-2020. By evaluating both simulations against the UA dataset, we show that calibration decreases domain averaged temporal RMSE and bias for snow depth from 0.15 to 0.13 m and from -0.036 to -0.0023 m, respectively, and improves the timing of snow ablation. Increased snow simulation performance also improves estimates of model-simulated runoff in four of six study basins, though only one has statistically significant improvement. Spatially distributed Noah-MP snow parameters perform better than default uniform values. We demonstrate that calibrating variables related to snow albedo calculations and rain-snow partitioning, among other processes, is a necessary step for creating a nature run that reasonably approximates true snow conditions for the OSSEs. Additionally, the inclusion of a snowfall scaling term can address biases in precipitation from meteorological forcing datasets, further improving the utility of LSMs for generating reliable spatiotemporal estimates of snow.

Seasonal snow cover classification based on SAR imagery and topographic data

ABSTRACT Melting snow is a sensitive indicator of regional climate change and snowmelt runoff. Optical remote sensing offers an effective tool to delineate the extent of snow cover, but its application is limited by sky conditions and illumination. In contrast, synthetic aperture radar (SAR) has the advantages of high penetration and can obtain the bulk properties of snowpack, such as snow wetness, in addition to the snow-covered area (SCA). The widely used algorithm based on the multitemporal backscattering coefficient can delineate wet snow and further estimate dry snow cover by empirical rules. However, employing a fixed threshold on wet snow extraction and only considering altitude in dry snow estimation lead to the misclassification of snow cover. In this study, an algorithm was developed for dry and wet snow discrimination based on C-band dual-polarization SAR data. First, the adaptive threshold obtained by the Otsu thresholding method (OTSU) was implemented on the ratio image to improve wet snow extraction, and then topographic information, including elevation, slope and aspect, was added to further infer dry snow cover. Snow field observations were carried out in March 2021 and February 2017 to evaluate the effectiveness of the presented method in typical snow regions with different snowpack characteristics. Compared with using a single equation with a fixed threshold (57.9% and 56.5%, respectively), the overall accuracy of the novel approach in prairie and alpine snow areas was 84.2% and 73.9%, respectively. The results show that considering the adaptive threshold and topography offers superior performance in identifying dry and wet snow.

Snow Depth Retrieval With an Autonomous UAV-Mounted Software-Defined Radar

We present results from a field campaign to measure seasonal snow depth at Cameron Pass, Colorado, using a synthetic ultrawideband software-defined radar (SDRadar) implemented in commercially available Universal Software Radio Peripheral (USRP) software-defined radio hardware and flown on a small hexacopter unmanned aerial vehicle (UAV). We coherently synthesize an ultrawideband signal from stepped frequency 50-MHz subpulses across 600–2100-MHz frequency bands using a novel nonuniform nonlinear synthetic wideband waveform reconstruction technique that minimizes sweep time and completely eliminates problematic grating lobes and other processing artifacts traditionally seen in stepped waveform synthesis. We image seasonal snow across two transects: a 400-m open Meadow Transect and a 380-m forested transect. We present a surface detection algorithm that fuses data from LiDAR, global navigation satellite system (GNSS)/global positioning system (GPS), and features in the radargram itself to obtain high precision estimates of both snow and ground surface reflections, and thus total snow depth, represented as two-way travel time. The measurements are validated against independent ground-based ground-penetrating radar measurements with correlations coefficients as high as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\rho = 0.9$ </tex-math></inline-formula> demonstrated. Finally, we compare backscattered radar data collected by the UAV-SDRadar while hovering proximal to a known snow pit with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in situ</i> measured snow dielectric profiles and demonstrate imaging of snow stratigraphy.

Quantifying the observational requirements of a space-borne LiDAR snow mission

This study quantifies the level of observational accuracy required from a spaceborne light detection and ranging (LiDAR) snow depth retrieval mission for enabling beneficial impacts for snow estimation. The study is conducted over a region in Western Colorado using a suite of observing system simulation experiments (OSSEs). The Joint UK Land Environment Simulator, version 5.0 (JULES v5.0) is employed to simulate a suite of idealized LiDAR observations, considering a range of LiDAR snow depth retrieval errors, different hypothetical sensor swath widths, and the impact of cloud cover on observability. These simulated observations are then assimilated into the Noah land surface model with multi-parameterization options, version 3.6 (Noah-MP v3.6) model. This data assimilation setup is used to systematically evaluate the potential utility of LiDAR observations for improving modeled snow water equivalent (SWE) estimates and water budget variables such as runoff. Results from the OSSE runs show that, in general, assimilation of synthetic LiDAR observations provide beneficial impacts when the LiDAR snow depth retrieval error standard deviation (σerror) is below 60 cm. Based on comparisons between the realistic (i.e., swath-limited and cloud-attenuated) case and the idealized (i.e., infinite swath width in the absence of cloud cover) case, this study concludes that observations with a conservative error standard deviation threshold of 40 cm (i.e., upper limit of the snow depth retrieval error that adds value to the SWE estimates via assimilation) are needed for improving modeled snow estimates. More than a 33% reduction in SWE root mean square errors and more than a 15% increase in correlation coefficients are achieved when σerror ≤ 40 cm using a 170-km sensor swath width in the presence of cloud attenuation effects. Further, the integrated hydrologic response, as represented by total (surface and subsurface) runoff estimates during the snow ablation season, are also enhanced when assimilating synthetic LiDAR snow depth retrievals with errors below this level.

Improving Snow Estimates Through Assimilation of MODIS Fractional Snow Cover Data Using Machine Learning Algorithms and the Common Land Model

AbstractIn this study, an innovative MODIS fractional snow cover (SCF) data assimilation (DA) prototype framework that invokes machine learning (ML) techniques and Common land model (CoLM) is proposed to improve the estimation of the snow depth (SD) and the SCF. To validate our new framework, we analyzed two snow seasons from 2013 to 2015 at 46 stations in Northern Xinjiang in China. We developed 12 SCF DA schemes that represent different DA methods (direct insertion (DI) and Ensemble Kalman Filter (EnKF)), observational data (original data and gap‐filled MODIS SCF data), and observation operators (five new snow depletion curves (SDCs) defined using traditional multivariate nonlinear regression and four ML methods). While improving the frequency of the SCF observations in the DI‐based DA scheme only resulted in a marginal improvement in the snow estimates, by adding new SDCs fitted by ML techniques (e.g., deep belief network), and the gap‐filled MODIS SCF data to the EnKF‐based DA scheme, we were able to reduce model structural uncertainties of CoLM and achieve marked improvement in the accuracy of the snow estimates (RMSE = 5.92 cm, mean bias error = −1.94 cm, and average degree of improvement = 32.18% for SD estimates and RMSE = 15. 79%, mean bias error = −1.21%, and average degree of improvement = 47.95% for SCF estimates). Our results demonstrate the feasibility of improving snow estimates by combining the ML techniques with physically based snowpack model in a SCF DA framework.

Physics-based SNOWPACK model improves representation of near-surface Antarctic snow and firn density

Abstract. Estimates of snow and firn density are required for satellite-altimetry-based retrievals of ice sheet mass balance that rely on volume-to-mass conversions. Therefore, biases and errors in presently used density models confound assessments of ice sheet mass balance and by extension ice sheet contribution to sea level rise. Despite this importance, most contemporary firn densification models rely on simplified semi-empirical methods, which are partially reflected by significant modeled density errors when compared to observations. In this study, we present a new drifting-snow compaction scheme that we have implemented into SNOWPACK, a physics-based land surface snow model. We show that our new scheme improves existing versions of SNOWPACK by increasing simulated near-surface (defined as the top 10 m) density to be more in line with observations (near-surface bias reduction from −44.9 to −5.4 kg m−3). Furthermore, we demonstrate high-quality simulation of near-surface Antarctic snow and firn density at 122 observed density profiles across the Antarctic ice sheet, as indicated by reduced model biases throughout most of the near-surface firn column when compared to two semi-empirical firn densification models (SNOWPACK mean bias=-9.7 kg m−3, IMAU-FDM mean bias=-32.5 kg m−3, GSFC-FDM mean bias=15.5 kg m−3). Notably, our analysis is restricted to the near surface where firn density is most variable due to accumulation and compaction variability driven by synoptic weather and seasonal climate variability. Additionally, the GSFC-FDM exhibits lower mean density bias from 7–10 m (SNOWPACK bias=-22.5 kg m−3, GSFC-FDM bias=10.6 kg m−3) and throughout the entire near surface at high-accumulation sites (SNOWPACK bias=-31.4 kg m−3, GSFC-FDM bias=-4.7 kg m−3). However, we found that the performance of SNOWPACK did not degrade when applied to sites that were not included in the calibration of semi-empirical models. This suggests that SNOWPACK may possibly better represent firn properties in locations without extensive observations and under future climate scenarios, when firn properties are expected to diverge from their present state.

Reconstruction of historical surface mass balance, 1984–2017 from GreenTrACS multi-offset ground-penetrating radar

AbstractWe present continuous estimates of snow and firn density, layer depth and accumulation from a multi-channel, multi-offset, ground-penetrating radar traverse. Our method uses the electromagnetic velocity, estimated from waveform travel-times measured at common-midpoints between sources and receivers. Previously, common-midpoint radar experiments on ice sheets have been limited to point observations. We completed radar velocity analysis in the upper ~2 m to estimate the surface and average snow density of the Greenland Ice Sheet. We parameterized the Herron and Langway (1980) firn density and age model using the radar-derived snow density, radar-derived surface mass balance (2015–2017) and reanalysis-derived temperature data. We applied structure-oriented filtering to the radar image along constant age horizons and increased the depth at which horizons could be reliably interpreted. We reconstructed the historical instantaneous surface mass balance, which we averaged into annual and multidecadal products along a 78 km traverse for the period 1984–2017. We found good agreement between our physically constrained parameterization and a firn core collected from the dry snow accumulation zone, and gained insights into the spatial correlation of surface snow density.