Evaluation and blending of ATMS and AMSR2 snow water equivalent retrievals over the conterminous United States
Introduction
Snowpack plays an important role in modulating global climate and hydrologic cycle (Dong, 2018; Lettenmaier et al., 2015; Sturm, 2015). Accurate estimates of snowpack properties are of critical importance to a variety of hydrologic and climate-related applications (Chang et al., 2005; Dozier et al., 2016). Many gridded products have been created to provide long-term snow depth (SD) or snow water equivalent (SWE) estimates. Such products include land surface reanalysis (Dee et al., 2011; Gelaro et al., 2017; Rodell et al., 2004; Xia et al., 2012), snow model simulations (Brun et al., 2013), regional climate model simulations (Wrzesien et al., 2018), and ground-based interpolation data (Brown and Brasnett, 2010; Broxton et al., 2016a). Among these, the model simulations and reanalysis are subject to large uncertainties stemming from those in model structures, parameters, as well as forcing data (Mortimer et al., 2020; Mudryk et al., 2015). Meanwhile, the interpolation data are constrained by the density and locations of stations.
In recent decades, satellite retrievals are seeing increasing applications in snowpack monitoring and prediction, especially in regions with poor ground measurements (Frei et al., 2012; Nolin, 2010). In particular, passive microwave (PMW) SD/SWE retrievals have the advantage of being directly relevant to water balance calculation, available for both day and night-time conditions, and not subject to interference by clouds or atmospheric gases as are snow cover retrieved by optical sensors (Clifford, 2010; Lee et al., 2015). Currently, operational spaceborne PMW sensors that can retrieve SD/SWE include the Defense Meteorological Satellite Program (DMSP) Special Sensor Microwave Imager/Sounder (SSMIS; Bommarito, 1993; Kunkee et al., 2008), the Global Change Observation Mission (GCOM) Advanced Microwave Scanning Radiometer (AMSR) series (Imaoka et al., 2002, Imaoka et al., 2010), and the Joint Polar Satellite System (JPSS) Advanced Technology Microwave Sounder (ATMS; Weng et al., 2012). Typically, the radiance observations from the PMW sensors are used to retrieve SD by exploiting empirical brightness temperature (Tb) – SD relationships (Kelly, 2009), and SD is then converted to SWE through empirical estimates of snow density (Sturm et al., 2010). The ATMS SWE retrieval algorithm is somewhat unique that it assimilates radiance observations from sounding channels into the Community Radiative Transfer Model (CRTM; Han et al., 2006) using the Microwave Integrated Retrieval System (MiRS; Boukabara et al., 2011). SWE is then retrieved by comparing the MiRS retrieved emissivity spectra with those from a precomputed catalog that relates surface emissivity to SWE to find the closest match. The catalog is generated from a dense medium radiative transfer snow emissivity model (Weng et al., 2001).
In spite of the many promising aspects of PWM SD/SWE retrievals, these products are limited in spatial resolution and are known to suffer from large errors (Dawson et al., 2018; De Lannoy et al., 2010; Frei et al., 2012). The errors may stem from sensor signal saturation, vegetation and terrain interference, snow wetness, and simplifying assumptions underpinning the retrieval algorithms (Dong et al., 2005; Liu et al., 2015). For example, Vuyovich et al. (2014) suggested that forest cover and deep snow have significant impact on AMSR – Earth Observing System (AMSR-E; Imaoka et al., 2002) and Special Sensor Microwave Imager (SSM/I; Hollinger, 1989) SWE estimates. Dai et al. (2017) showed that the mountainous topography and the coarse resolution of PMW sensor underlie the large disagreement between AMSR-E SD and in situ observations. Cho et al. (2020) illustrated that slope and surface heterogeneity impact the SWE difference between the SSMI/S (i.e., SSM/I and SSMIS) and gamma SWE. Tuttle et al. (2018) found that up to half of the error in AMSR-E SWE is potentially due to subpixel scale variability. While these studies advance our understanding of the error sources of PMW data, most of them fall short in proposing or establishing effective mechanisms for mitigating the errors. Furthermore, considering the number of sensors and retrieval products that are currently available, there is a clear, and heretofore unfulfilled demand for identifying and leveraging the complementary strengths of different PMW retrievals and in situ products, and thereby facilitating the application adoption of the retrievals.
One way to address the shortcomings of stand-alone PMW SD/SWE data is through the joint use of PMW radiometry and ground observations as information sources in the retrieval algorithms (Pulliainen, 2006). For example, the Global Snow Monitoring for Climate Research (GlobSnow) SWE product (Pulliainen et al., 2020; Takala et al., 2011) assimilates different sources of PMW Tb (from 18.7 and 36.5 GHz channels) and in situ SD into a semi-empirical snow emission model and provides 25-km daily SWE estimates from 1979 to 2018 over the Northern Hemisphere excluding alpine areas. This product, however, still exhibits large errors inherited from structural limitations of the Tb–SD relation and the snow emission models (Hancock et al., 2013; Larue et al., 2017; Mudryk et al., 2015). An alternative way is through the assimilation of the PMW observations into snow models (Dong et al., 2007; Dziubanski and Franz, 2016) or land surface models (Che et al., 2014; De Lannoy et al., 2012; Kwon et al., 2017; Liu et al., 2013). For example, Kumar et al. (2019) assimilated different sources of PMW SD into the Noah model (Ek et al., 2003) to improve SD estimates from 1979 to 2015 for the conterminous United States (CONUS). While these model-based products were demonstrated to be generally superior to stand-alone PMW SWE (Cho et al., 2020; Dawson et al., 2018; Mortimer et al., 2020), their creation entails high computational costs and their accuracy remains subject to questions in data scarce areas (Broxton et al., 2016b; Clark et al., 2011; Rutter et al., 2009).
The objectives of this study are twofold. The first is to assess the complementary skills of the two different PMW SWE retrievals, namely that from GCOM AMSR2 and the one based on JPSS ATMS, over the CONUS. The second is to develop a lightweight, computationally efficient blending algorithm that optimally combining the two satellite products and in situ observations. The blending scheme has the advantage of being simple and independent of any snow model, and its product can be assimilated to the latter to further improve the prediction of snow and other hydrologic variables (Kumar et al., 2015; Liu et al., 2015). We choose to focus on SWE rather than SD, as the former can be directly used in hydrologic analysis and predictions. The AMSR2 and ATMS SWE retrievals are selected for the following reasons. First, these products are based on two relatively new instruments that contrast sharply in scanning and channel configurations. AMSR2, like its predecessor AMSR-E, is a conical scanner measuring orthogonally polarized radiation (vertical and horizontal) at specific window frequencies (Imaoka et al., 2010), whereas ATMS is a cross-track scanner measuring radiation at all its channels at varying scan angles (Weng et al., 2012). The second reason is that the retrieval algorithms for the two products are quite different. While the AMSR2 retrieval algorithm has undergone extensive assessments (Lee et al., 2015; Wang et al., 2019; Zhang et al., 2017), neither the ATMS SWE retrieval nor the associated algorithm in MiRS has received much attention. This study is intended to fill the latter knowledge gap by gauging the relative strengths of ATMS retrieval against the more established AMSR2 counterpart.
The primary research questions of this study are as follows. First, how do the AMSR2 and ATMS SWE retrievals differ in their accuracy among different geographic regions, and what are the complementary strengths (if any) of the two products? Second, how the differences in skill can be attributed to differences in scanning patterns of each instrument and in the retrieval algorithms? Third, can the introduction of a priori bias correction improve upon the optimal interpolation-based blending scheme, which is typically employed in the field of snow analysis (Brasnett, 1999; Brown et al., 2003; Liu et al., 2015)?
The rest of this paper is structured as follows. Section 2 describes the study area, the two PWM SW retrievals, and the in situ observations for analysis and validation. Sections 3 offers an overview of the blending algorithm, and Section 4 presents and interprets key observations emerging from the evaluation. Section 5 summarizes the major findings and concludes the study.
Section snippets
Study area and data
Fig. 1 shows the elevation and spatial distribution of Snow Telemetry (SNOTEL) and Cooperative Observer Program (COOP) stations in the 18 hydrologic units (HUs) over the CONUS. The research domain extends from 25° to 53° north and 125° to 67° west with a 0.125° grid spacing, which is same as the North American Land Data Assimilation System (NLDAS; Mitchell et al., 2004). The elevation data were aggregated from the 30 arc sec Global Multi-resolution Terrain Elevation Data (GMTED2010; Danielson
Blending algorithm framework
The algorithm framework for blending satellite retrievals with in situ observations is shown in Fig. 3. The first step is to mask the PMW SWE with the IMS snow cover maps. Namely, the PMW SWE retrievals are retained only over snow covered grids in order to remove possible false alarms; in addition, any grids that are seen to be snow covered in IMS but not detected by the PMW retrievals are filled with 5 mm of SWE. The outcome from the masking then undergoes the following processing steps: i)
Results and discussion
In this section we first present the outcomes from the comparison of ATMS and AMSR2 SWE retrievals, then we describe the results of cross-validation experiments with a focus on the differential skills of products generated through each scheme. In the end we further compare the best product as determined from the cross-validation experiments with SNODAS analysis.
Summary and conclusions
This paper compares two different PMW SWE products, namely ATMS and AMSR2, with SNOTEL and COOP in situ observations and the SNODAS analysis over the CONUS. A blending algorithm is then designed to optimally combine the in situ and satellite SWE to obtain a reliable gridded product at a relatively high spatial resolution (0.125° × 0.125°).
The comparison results indicate that the accuracy of ATMS and AMSR2 SWE, despite derived using different instruments and retrieval algorithms, have much in
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the National Oceanic and Atmospheric Administration (grant #NA18OAR4590410) and this support is graciously acknowledged here. We would also like to thank Greg Fall and Brian Cosgrove at the Office of Water Prediction (OWP), and Hui Shao at the Joint Center for Satellite Data Assimilation (JCSDA) for their helpful suggestions. We appreciate the constructive comments made by the three anonymous reviewers. The daily 0.125° blended in situ-satellite SWE data are available
References (95)
- et al.
Assessment of spring snow cover duration variability over northern Canada from satellite datasets
Remote Sens. Environ.
(2007) - et al.
Assimilating passive microwave remote sensing data into a land surface model to improve the estimation of snow depth
Remote Sens. Environ.
(2014) Remote sensing, hydrological modeling and in situ observations in snow cover research: a review
J. Hydrol.
(2018)- et al.
Factors affecting remotely sensed snow water equivalent uncertainty
Remote Sens. Environ.
(2005) - et al.
Assimilation of AMSR-E snow water equivalent data in a spatially-lumped snow model
J. Hydrol.
(2016) - et al.
Quantifying the uncertainty in passive microwave snow water equivalent observations
Remote Sens. Environ.
(2005) - et al.
A review of global satellite-derived snow products
Adv. Space Res.
(2012) - et al.
Evaluating global snow water equivalent products for testing land surface models
Remote Sens. Environ.
(2013) - et al.
Validation of GlobSnow-2 snow water equivalent over eastern Canada
Remote Sens. Environ.
(2017) - et al.
Assimilating satellite-based snow depth and snow cover products for improving snow predictions in Alaska
Adv. Water Resour.
(2013)
The airborne snow observatory: fusion of scanning lidar, imaging spectrometer, and physically-based modeling for mapping snow water equivalent and snow albedo
Remote Sens. Environ.
Mapping of snow water equivalent and snow depth in boreal and sub-arctic zones by assimilating space-borne microwave radiometer data and ground-based observations
Remote Sens. Environ.
Estimating northern hemisphere snow water equivalent for climate research through assimilation of space-borne radiometer data and ground-based measurements
Remote Sens. Environ.
AMSR2 snow depth downscaling algorithm based on a multifactor approach over the Tibetan plateau, China
Remote Sens. Environ.
Spatial Distribution and Evolution of a Seasonal Snowpack in Complex Terrain: An Evaluation of the SNODAS Modeling Product
DMSP special sensor microwave imager sounder (SSMIS)
MiRS: an all-weather 1DVAR satellite data assimilation and retrieval system
IEEE T. Geosci. Remote
A physical approach for a simultaneous retrieval of sounding, surface, hydrometeor, and cryospheric parameters from SNPP/ATMS
J. Geophys. Res. Atmos.
A global analysis of snow depth for numerical weather prediction
J. Appl. Meteorol.
Canadian Meteorological Centre (CMC) Daily Snow Depth Analysis Data
Gridded North American monthly snow depth and snow water equivalent for GCM evaluation
Atmosphere-Ocean
Linking snowfall and snow accumulation to generate spatial maps of SWE and snow depth
Earth Space Sci.
Why do global reanalyses and land data assimilation products underestimate snow water equivalent?
J. Hydrometeorol.
Simulation of northern Eurasian local snow depth, mass, and density using a detailed snowpack model and meteorological reanalyses
J. Hydrometeorol.
NOHRSC Operations and the Simulation of Snow Cover Properties for the Coterminous US, in: Proceedings of the 69th Annual Western Snow Conference
Analysis of ground-measured and passive-microwave-derived snow depth variations in midwinter across the northern Great Plains
J. Hydrometeorol.
Assessing objective techniques for gauge-based analyses of global daily precipitation
J. Geophys. Res. Atmos.
Validation of NOAA-interactive multisensor snow and ice mapping system (IMS) by comparison with ground-based measurements over continental United States
Remote Sens.
The value of long-term (40 years) airborne gamma radiation SWE record for evaluating three observation-based gridded SWE data sets by seasonal snow and land cover classifications
Water Resour. Res.
Representing spatial variability of snow water equivalent in hydrologic and land-surface models: a review
Water Resour. Res.
Global estimates of snow water equivalent from passive microwave instruments: history, challenges and future developments
Int. J. Remote Sens.
Evaluation of SNODAS snow depth and snow water equivalent estimates for the Colorado Rocky Mountains, USA
Hydrol. Process.
Evaluation of snow cover and snow depth on the Qinghai–Tibetan plateau derived from passive microwave remote sensing
Cryosphere
Global Multi-Resolution Terrain Elevation Data 2010 (GMTED2010)
A new snow density parameterization for land data initialization
J. Hydrometeorol.
Evaluation of remotely sensed snow water equivalent and snow cover extent over the contiguous United States
J. Hydrometeorol.
Satellite-scale snow water equivalent assimilation into a high-resolution land surface model
J. Hydrometeorol.
Multiscale assimilation of advanced microwave scanning radiometer–EOS snow water equivalent and moderate resolution imaging Spectroradiometer snow cover fraction observations in northern Colorado
Water Resour. Res.
The ERA-interim reanalysis: configuration and performance of the data assimilation system
Q. J. Roy. Meteor. Soc.
Scanning multichannel microwave radiometer snow water equivalent assimilation
J. Geophys. Res. Atmos.
Estimating the spatial distribution of snow water equivalent in the world's mountains
WIREs Water
Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale eta model
J. Geophys. Res. Atmos.
A blended global snow product using visible, passive microwave and scatterometer satellite data
Int. J. Remote Sens.
Estimation of prediction error by using K-fold cross-validation
Stat. Comput.
Stepwise sensitivity analysis from qualitative to quantitative: application to the terrestrial hydrological modeling of a conjunctive surface-subsurface process (CSSP) land surface model
J. Adv. Model. Earth Sys.
Objective Analysis of Meteorological Fields
The modern-era retrospective analysis for research and applications, version 2 (MERRA-2)
J. Clim.
Cited by (12)
Mapping of snow water equivalent by a deep-learning model assimilating snow observations
2023, Journal of HydrologyAssimilation of blended in situ-satellite snow water equivalent into the National Water Model for improving hydrologic simulation in two US river basins
2022, Science of the Total EnvironmentCitation Excerpt :The objective of this paper is to investigate the potential of assimilating a recently developed in situ-satellite blended SWE product for improving snow and streamflow simulations of the National Water Model (NWM). This product, developed by Gan et al. (2021), blends two PWM SWE products that undergo bias adjustment using in situ observations. These products are the SWE retrievals from Suomi National Polar-orbiting Partnership (S-NPP) Advanced Technology Microwave Sounder (ATMS; Weng et al., 2012) and Global Change Observation Mission – Water (GCOM-W1) Advanced Microwave Scanning Radiometer 2 (AMSR2; Kelly, 2009).
LSTM-Based Data Integration to Improve Snow Water Equivalent Prediction and Diagnose Error Sources
2024, Journal of HydrometeorologySpatial Downscaling of Snow Water Equivalent Using Machine Learning Methods Over the Zayandehroud River Basin, Iran
2023, PFG - Journal of Photogrammetry, Remote Sensing and Geoinformation ScienceModeling and estimation of snow depth spatial correlation structure from observations over North America
2023, Frontiers in Earth Science