Northern Hemisphere Snow Research Articles

Evidence of human influence on Northern Hemisphere snow loss

Documenting the rate, magnitude and causes of snow loss is essential to benchmark the pace of climate change and to manage the differential water security risks of snowpack declines1–4. So far, however, observational uncertainties in snow mass5,6 have made the detection and attribution of human-forced snow losses elusive, undermining societal preparedness. Here we show that human-caused warming has caused declines in Northern Hemisphere-scale March snowpack over the 1981–2020 period. Using an ensemble of snowpack reconstructions, we identify robust snow trends in 82 out of 169 major Northern Hemisphere river basins, 31 of which we can confidently attribute to human influence. Most crucially, we show a generalizable and highly nonlinear temperature sensitivity of snowpack, in which snow becomes marginally more sensitive to one degree Celsius of warming as climatological winter temperatures exceed minus eight degrees Celsius. Such nonlinearity explains the lack of widespread snow loss so far and augurs much sharper declines and water security risks in the most populous basins. Together, our results emphasize that human-forced snow losses and their water consequences are attributable—even absent their clear detection in individual snow products—and will accelerate and homogenize with near-term warming, posing risks to water resources in the absence of substantial climate mitigation.

Synergistic impacts of wintertime regional snow anomalies in the Northern Hemisphere on the summer rainfall pattern in China

Changes in winter snow cover in the Northern Hemisphere (NH) could have a profound impact on mid-latitude weather. Previous studies have focused on the role of regional, e.g. Eurasian or Tibetan, snow cover in summer precipitation anomaly, without considering the synergistic impacts of hemispheric wintertime snow. In this study, we find that the dominant pattern of the NH winter overall snow cover anomaly with a synergistic impact, has a stronger cross-seasonal association with the China’s summer rainfall pattern than regional snow cover anomaly. We summarize three synergistic impact paths of regional snow cover. One is extratropical path, that is the westerlies are affected by less snow in Europe through the snow-soil moisture-atmospheric feedback, and the influence is strengthened by less snow in Mongolia through enhanced temperature anomalies. The second is subtropical path, that is the meridional thermal difference anomaly caused by more snow anomaly on the Tibetan Plateau is strengthened by less Mongolian snow and then impacts the behavior of the upper-tropospheric westerly jet. Third, concurrently, more North American snow enhances the above two synergistic influence paths via the Circumglobal Teleconnection pattern. These three paths can be simultaneously reflected in the associated circulations of the first mode of NH snow cover. Their synergistic impacts eventually influence the meridional East Asia-Pacific pattern circulation anomalies in summer, leading to increased precipitation in the Yangtze River Basin. The cross-seasonal influences of synergistic effects of multiple regional snow anomalies can be identified by CMIP6 multi-model ensembles, particularly the impact of European snow cover.

Northern Hemisphere Snow Drought in Earth System Model Simulations and ERA5‐Land Data in 1980–2014

AbstractLow snow levels over the past few decades and predictions of a low‐to‐no snow future have spurred research into snow droughts, which pose a threat to water security and management. Systematic data‐model comparisons of snow drought have been lacking, hindering our understanding of the drivers of snow drought in the past. To address this gap, we analyzed snow drought events using standardized snow water equivalent index derived from monthly results of four numerical experiments using the E3SM Land Model (ELM) and ERA5‐Land data during the period of 1980–2014. Additionally, we compared snow drought duration calculated from models with those from the ERA5‐Land data during selected El Niño‐Southern Oscillation (ENSO) years. The numerical experiments were conducted with ELM driven by two prescribed atmospheric forcings, and with the coupled land‐atmosphere configuration of E3SM with and without plant hydraulics scheme feedback. Analysis reveals that 20%–30% of snow droughts occur due to factors other than above‐normal temperature and low snowfall, such as low soil moisture, warm soil temperature, and low relative humidity, etc., especially in high latitudes (50° North). Furthermore, our study highlights the exacerbating effect of ENSO events on snow drought conditions in various regions, despite some discrepancies between model and ERA5‐Land results. We also identified limitations of the coupled land‐atmosphere models in our current configuration in capturing the spatial patterns of snow droughts. This study underscores the challenge of predicting and mitigating snow drought and the need for a comprehensive understanding of the factors contributing to snow drought.

What Determines the Leading Mode of Wintertime Northern Hemisphere Snow-Cover Interannual Variability?

Abstract The Northern Hemisphere snow-cover (NHSC) interannual variability can serve as a crucial source of climate predictability. However, the spatiotemporal characteristics of the NHSC interannual variability and associated formation mechanism lack a thorough investigation using the latest reliable datasets. In this observationally based study, we identify the leading mode of the wintertime NHSC interannual variability and reveal the major determinants and mechanisms for forming the mode. Results show that the leading mode exhibits a hemispheric dipole pattern characterized by snow-cover anomalies in the central-eastern United States and Europe, contrary to those in the Tibetan Plateau (TP). Importantly, the leading mode of TP-only snow-cover variability is essentially a regional manifestation of the dipole NHSC anomaly pattern. The wintertime hemispheric atmospheric circulation anomaly mode, that is, the Arctic Oscillation (AO), primarily determines the dipole NHSC anomaly pattern via altering snowfall and snowmelt. During a positive phase of AO, an anomalous atmospheric anticyclone over the midlatitude North Atlantic favors increased warm-air advection over the central-eastern United States and Europe, reducing the snowfall and the snow cover there. Meanwhile, the anomalously warm atmospheric thermal condition in Europe accompanied by enhanced westerly moisture flux convergence tends to heat the underlying land surface via reducing upward sensible heat flux and increasing downward longwave radiation, thereby causing increased snowmelt and decreased snow cover in northeastern Europe. Associated with the aforementioned atmospheric circulation anomalies, an anomalous atmospheric cyclone prevailing over the TP tends to cause increased southeasterly moisture flux convergence, favoring snowfall and snow increase in the TP.

NH-SWE: Northern Hemisphere Snow Water Equivalent dataset based on in situ snow depth time series

Abstract. Ground-based datasets of observed snow water equivalent (SWE) are scarce, while gridded SWE estimates from remote-sensing and climate reanalysis are unable to resolve the high spatial variability of snow on the ground. Long-term ground observations of snow depth, in combination with models that can accurately convert snow depth to SWE, can fill this observational gap. Here, we provide a new SWE dataset (NH-SWE) that encompasses 11 071 stations in the Northern Hemisphere (NH) and is available at https://doi.org/10.5281/zenodo.7515603 (Fontrodona-Bach et al., 2023). This new dataset provides daily time series of SWE, varying in length between 1 and 73 years, spanning the period 1950–2022, and covering a wide range of snow climates including many mountainous regions. At each station, observed snow depth was converted to SWE using an established snow-depth-to-SWE conversion model, with excellent model performance using regionalised parameters based on climate variables. The accuracy of the model after parameter regionalisation is comparable to that of the calibrated model. The key advantages and strengths of the regionalised model presented here are its transferability across climates and the high performance in modelling daily SWE dynamics in terms of peak SWE, total snowmelt and duration of the melt season, as assessed here against a comparison model. This dataset is particularly useful for studies that require accurate time series of SWE dynamics, timing of snowmelt onset, and snowmelt totals and duration. It can, for example, be used for climate change impact analyses, water resources assessment and management, validation of remote sensing of snow, hydrological modelling, and snow data assimilation into climate models.

Forcing and impact of the Northern Hemisphere continental snow cover in 1979–2014

Abstract. The main drivers of the continental Northern Hemisphere snow cover are investigated in the 1979–2014 period. Four observational datasets are used as are two large multi-model ensembles of atmosphere-only simulations with prescribed sea surface temperature (SST) and sea ice concentration (SIC). A first ensemble uses observed interannually varying SST and SIC conditions for 1979–2014, while a second ensemble is identical except for SIC with a repeated climatological cycle used. SST and external forcing typically explain 10 % to 25 % of the snow cover variance in model simulations, with a dominant forcing from the tropical and North Pacific SST during this period. In terms of the climate influence of the snow cover anomalies, both observations and models show no robust links between the November and April snow cover variability and the atmospheric circulation 1 month later. On the other hand, the first mode of Eurasian snow cover variability in January, with more extended snow over western Eurasia, is found to precede an atmospheric circulation pattern by 1 month, similar to a negative Arctic oscillation (AO). A decomposition of the variability in the model simulations shows that this relationship is mainly due to internal climate variability. Detailed outputs from one of the models indicate that the western Eurasia snow cover anomalies are preceded by a negative AO phase accompanied by a Ural blocking pattern and a stratospheric polar vortex weakening. The link between the AO and the snow cover variability is strongly related to the concomitant role of the stratospheric polar vortex, with the Eurasian snow cover acting as a positive feedback for the AO variability in winter. No robust influence of the SIC variability is found, as the sea ice loss in these simulations only drives an insignificant fraction of the snow cover anomalies, with few agreements among models.

Projected changes to Northern Hemisphere snow conditions over the period 1950–2100, given two emission scenarios

In this study Northern Hemisphere (NH) snow conditions were simulated and analyzed for the period 1950–2100 under two emission scenarios RCP45 and RCP85. Daily snow conditions were simulated using the Snow17 model run on the Google Earth Engine (GEE) parallel cloud computing platform with each of the 21 GCMs in the NASA NEX GDDP dataset. The model was evaluated based on Snow Water Equivalent (SWE) and Snow Cover Area (SCA) data from the GLDAS-2 dataset and SCA remote sensing data from the MODIS/Terra instrument. The results showed that both NH SWE and the number of Snow-Covered Days (SCD) have and will continue to decrease at lower latitudes, whereas in many high latitude regions SWE is expected to increase, even as SCD is decreasing. The average annual NH SCD is expected to decrease by 12.5% and 23.1% between the periods 1950–1975 and 2075–2100, under RCP45 and RCP85, respectively. Between the same periods however, the NH mean winter SWE is expected to increase slightly in the beginning of the period under both RCPs. By 2100, SWE will have reduced to the 1950–1975 level under RCP45. Under RCP85, SWE will be 10% lower than the 1950–1975 level. These changes will pose a great challenge for water resource management across a wide spectrum, including impacting current way of living.

A long-term daily gridded snow depth dataset for the Northern Hemisphere from 1980 to 2019 based on machine learning

ABSTRACT A high-quality snow depth product is very import for cryospheric science and its related disciplines. Current long time-series snow depth products covering the Northern Hemisphere can be divided into two categories: remote sensing snow depth products and reanalysis snow depth products. However, existing gridded snow depth products have some shortcomings. Remote sensing-derived snow depth products are temporally and spatially discontinuous and tend to underestimate snow depth, while reanalysis snow depth products have coarse spatial resolutions and great uncertainties. To overcome these problems, in our previous work we proposed a novel data fusion framework based on Random Forest Regression of snow products from Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E), Advanced Microwave Scanning Radiometer-2 (AMSR2), Global Snow Monitoring for Climate Research (GlobSnow), the Northern Hemisphere Snow Depth (NHSD), ERA-Interim, and Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), incorporating geolocation (latitude and longitude), and topographic data (elevation), which were used as input independent variables. More than 30,000 ground observation sites were used as the dependent variable to train and validate the model in different time periods. This fusion framework resulted in a long time series of continuous daily snow depth product over the Northern Hemisphere with a spatial resolution of 0.25°. Here, we compared the fused snow depth and the original gridded snow depth products with 13,272 observation sites, showing an improved precision of our product. The evaluation indices of the fused (best original) dataset yielded a coefficient of determination R2 of 0.81 (0.23), Root Mean Squared Error (RMSE) of 7.69 (15.86) cm, and Mean Absolute Error (MAE) of 2.74 (6.14) cm. Most of the bias (88.31%) between the fused snow depth and in situ observations was in the range of −5 cm to 5 cm. The accuracy assessment of independent snow observation sites – Sodankylä (SOD), Old Aspen (OAS), Old Black Spruce (OBS), and Old Jack Pine (OJP) – showed that the fused snow depth dataset had high precision for snow depths of less than 100 cm with a relatively homogeneous surrounding environment. The results of random point selection and independent in situ site validation show that the accuracy of the fused snow depth product is not significantly improved in deep snow areas and areas with complex terrain. In the altitude range of 100 m to 2000 m, the fused snow depth had a higher precision, with R2 varying from 0.73 to 0.86. The fused snow depth had a decreasing trend based on the spatiotemporal analysis and Mann-Kendall trend test method. This fused snow depth product provides the basis for understanding the temporal and spatial characteristics of snow cover and their relation to climate change, hydrological and water cycle, water resource management, ecological environment, snow disaster and hazard prevention.

Implementing spatially and temporally varying snow densities into the GlobSnow snow water equivalent retrieval

Abstract. Snow water equivalent (SWE) is a valuable characteristic of snow cover, and it can be estimated using passive spaceborne radiometer measurements. The radiometer-based GlobSnow SWE retrieval methodology, which assimilates weather station snow depth observations into the retrieval, has improved the reliability and accuracy of SWE retrieval when compared to stand-alone radiometer SWE retrievals. To further improve the GlobSnow SWE retrieval methodology, we investigate implementing spatially and temporally varying snow densities into the retrieval procedure. Thus far, the GlobSnow SWE retrieval has used a constant snow density throughout the retrieval despite differing locations, snow depth, or time of winter. This constant snow density is a known source of inaccuracy in the retrieval. Four different versions of spatially and temporally varying snow densities are tested over a 10-year period (2000–2009). These versions use two different spatial interpolation techniques: ordinary Kriging interpolation and inverse distance weighted regression (IDWR). All versions were found to improve the SWE retrieval compared to the baseline GlobSnow v3.0 product, although differences between versions are small. Overall, the best results were obtained by implementing IDWR-interpolated densities into the algorithm, which reduced RMSE (root mean square error) and MAE (mean absolute error) by about 4 mm (8 % improvement) and 5 mm (16 % improvement) when compared to the baseline GlobSnow product, respectively. Furthermore, implementing varying snow densities into the SWE retrieval improves the magnitude and seasonal evolution of the Northern Hemisphere snow mass estimate compared to the baseline product and a product post-processed with varying snow densities.

The hazard components of representative key risks. The physical climate perspective

The framework of Representative Key Risks (RKRs) has been adopted by the Intergovernmental Panel on Climate Change Working Group II (WGII) to categorize, assess and communicate a wide range of regional and sectoral key risks from climate change. These are risks expected to become severe due to the potentially detrimental convergence of changing climate conditions with the exposure and vulnerability of human and natural systems. Other papers in this special issue treat each of eight RKRs holistically by assessing their current status and future evolution as a result of this convergence. However, in these papers, such assessment cannot always be organized according to a systematic gradation of climatic changes. Often the big-picture evolution of risk has to be extrapolated from either qualitative effects of “low”, “medium” and “high” warming, or limited/focused analysis of the consequences of particular mitigation choices (e.g., benefits of limiting warming to 1.5 or 2C), together with consideration of the socio-economic context and possible adaptation choices.In this study we offer a representation – as systematic as possible given current literature and assessments – of the future evolution of the hazard components of RKRs.We identify the relevant hazards for each RKR, based upon the WGII authors’ assessment, and we report on their current state and expected future changes in magnitude, intensity and/or frequency, linking these changes to Global Warming Levels (GWLs) to the extent possible. We draw on the assessment of changes in climatic impact-drivers relevant to RKRs described in the 6th Assessment Report by Working Group I supplemented when needed by more recent literature.For some of these quantities - like regional trends in oceanic and atmospheric temperature and precipitation, some heat and precipitation extremes, permafrost thaw and Northern Hemisphere snow cover - a strong and quantitative relationship with increasing GWLs has been identified. For others - like frequency and intensity of tropical cyclones and extra-tropical storms, and fire weather - that link can only be described qualitatively. For some processes - like the behavior of ice sheets, or changes in circulation dynamics - large uncertainties about the effects of different GWLs remain, and for a few others - like ocean pH and air pollution - the composition of the scenario of anthropogenic emissions is most relevant, rather than the warming reached. In almost all cases, however, the basic message remains that every small increment in CO2 concentration in the atmosphere and associated warming will bring changes in climate phenomena that will contribute to increasing risk of impacts on human and natural systems, in the absence of compensating changes in these systems’ exposure and vulnerability, and in the absence of effective adaptation. Our picture of the evolution of RKR-relevant climatic impact-drivers complements and enriches the treatment of RKRs in the other papers in at least two ways: by filling in their often only cursory or limited representation of the physical climate aspects driving impacts, and by providing a fuller representation of their future potential evolution, an important component – if never the only one – of the future evolution of risk severity.

Benchmarking algorithm changes to the Snow CCI+ snow water equivalent product

The European Space Agency (ESA) Snow Climate Change Initiative (CCI+) provides long-term, global time series of daily snow cover fraction and snow water equivalent (SWE). The Snow CCI+ SWE Version 1 (CCIv1) product is built on the GlobSnow algorithm, which combines passive microwave (PMW) data with in situ snow depth (SD) measurements to estimate SWE. While CCIv1 remains algorithmically similar to the most recent GlobSnow product (GlobSnow Version 3), Snow CCI+ SWE Version 2 (CCIv2) incorporates two notable differences. CCIv2 uses updated PMW data from the NASA MEaSUREs Calibrated Passive Microwave Daily EASE-Grid 2.0 Earth Science Data Record and is generated in EASE-Grid 2.0 with 12.5 km grid spacing. It also adjusts SWE retrievals in post-processing by incorporating spatially and temporally varying snow density information. Due to the phased product development framework CCI+ employs, proposed changes between CCIv1 and CCIv2 were implemented in a series of step-wise developmental datasets. Using these developmental datasets, we analyze how changes to input PMW and SD data and the snow density parameterization affect the resulting SWE product. Using in situ snow courses as reference data, we demonstrate that the correlation and RMSE of the CCIv2 developmental product improved 18% (0.10) and 12% (5 mm), respectively, relative to CCIv1. The timing of peak snow mass is shifted two weeks later and a temporal discontinuity in the monthly northern hemisphere snow mass time series associated with the shift from the Special Sensor Microwave/Imager (SSM/I) to the Special Sensor Microwave Imager/Sounder (SSMIS) in 2009 is also removed.

Evaluation of Northern Hemisphere snow water equivalent in CMIP6 models during 1982–2014

Abstract. Seasonal snow cover of the Northern Hemisphere (NH) is a major factor in the global climate system, which makes snow cover an important variable in climate models. Previously, substantial uncertainties have been reported in NH snow water equivalent (SWE) estimates. A recent bias-correction method significantly reduces the uncertainty of NH SWE estimation, which enables a more reliable analysis of the climate models' ability to describe the snow cover. We have intercompared NH SWE estimates between CMIP6 (Coupled Model Intercomparison Project Phase 6) models and observation-based SWE reference data north of 40∘ N for the period 1982–2014 and analyzed with a regression approach whether model biases in temperature (T) and precipitation (P) could explain the model biases in SWE. We analyzed separately SWE in winter and SWE change rate in spring. For SWE reference data, we used bias-corrected SnowCCI data for non-mountainous regions and the mean of Brown, MERRA-2 and Crocus v7 data for the mountainous regions. The SnowCCI SWE data are based on satellite passive microwave radiometer data and in situ snow depth data. The analysis shows that CMIP6 models tend to overestimate SWE; however, large variability exists between models. In winter, P is the dominant factor causing SWE discrepancies especially in the northern and coastal regions. T contributes to SWE biases mainly in regions, where T is close to 0∘ C in winter. In spring, the importance of T in explaining the snowmelt rate discrepancies increases. This is to be expected, because the increase in T is the main factor that causes snow to melt as spring progresses. Furthermore, it is obvious from the results that biases in T or P cannot explain all model biases either in SWE in winter or in the snowmelt rate in spring. Other factors, such as deficiencies in model parameterizations and possibly biases in the observational datasets, also contribute to SWE discrepancies. In particular, linear regression suggests that when the biases in T and P are eliminated, the models generally overestimate the snowmelt rate in spring.

Variability and Changes of Unfrozen Soils Below Snowpack

AbstractUsing four reanalysis data sets and ground‐based observations, this paper uncovers that on average, 30% of the time, Northern Hemisphere snow cover experiences unfrozen bottom soil. It is demonstrated that the probability of occurrence of unfrozen soils is correlated with the snow types and is maximum over the ephemeral followed by the maritime and prairie snow. The results based on reanalysis data unveil that the seasonal evolution of the unfrozen soil areas is not synchronous with the snow cover extent and exhibits sub‐annual bi‐modality with two annual maxima in April and October. Interannual trend analyses indicate that shrinkage of spring snow in the past few decades has been accompanied by an increase in the proportion of unfrozen bottom soils, more significantly over polar climate regimes dominated by the tundra and taiga snow. The findings imply that the snowpack basal melting could have increased due to global warming.

Detection of climate transitions and discontinuities by Hurst rescaling

AbstractThe method of Outcalt et al., based on work developed originally by Hurst, is re‐examined to evaluate its efficacy in delineating changes in trends and identifying regime shifts in climatic‐related time series. This technique is based on the concept of the normalized rescaled running sum where temporal changes in the Hurst exponent can be used to identify climatic trends from one regime to another as each regime has a characteristic distribution that differs from the statistical characteristics of the complete time series. An examination of the temporal change in the amplitude of the normalized rescaled running sum can be used as a method to identify these regime changes, which may be either real (i.e., a true climatic shift) or induced (i.e., through a change in measurement bias, station location, or other nonclimatic influence). Examples shown here focus on examining time series of the Pacific Decadal Oscillation, Arctic thaw depth, the Northern Hemisphere snow cover extent, treeflow data from Lees Ferry (AZ), North Atlantic tropical cyclone frequency, and central England air temperatures.

Historical evolution and future trend of Northern Hemisphere snow cover in CMIP5 and CMIP6 models

Historical snow cover over the Northern Hemisphere was examined in the satellite-based NOAA-CDR data for the period of 1970–2019. Observed annual snow cover fraction (SNF) has reduced over most areas by up to 2%/decade, while annual snow cover area (SCA) has reduced by 2 × 105 km2/decade. However, SCA in the October–December season has increased by about 5 × 105 km2/decade. CMIP5 and CMIP6 historical experiments were validated against the NOAA-CDR data. Snow cover was generally well simulated in both CMIPs, with CMIP6 models performing better. The biases in SCA reduction were larger and smaller during summer and winter, respectively. The observed increase of October–November–December SCA in the 2000s was not reproduced. Climate projections of future snow cover were evaluated in CMIP6. SNF is projected to decrease in the next 80 years, under all four scenarios evaluated (SSP126, SSP245, SSP370 and SSP585). The higher the greenhouse emissions in the shared socio-economic pathways, the faster the reduction. Under the SSP585 scenario, the rate of SCA reduction is projected to exceed −1.2 × 106 km2/decade. By 2081–2100, annual (January–March) SCA is projected to decrease by more than 30% (20%). Under the SSP126 scenario, annual (January–March) SCA is projected to only reduce by about 10% (5%) relative 1995–2014 values. The reduction of Greenhouse gas emissions is critical to controlling the loss of snow cover; future snow cover only stabilizes under the SSP126 scenario, but continue to decrease under the other three scenarios.

Assessing Prior Emergent Constraints on Surface Albedo Feedback in CMIP6

AbstractAn emergent constraint (EC) is a popular model evaluation technique, which offers the potential to reduce intermodel variability in projections of climate change. Two examples have previously been laid out for future surface albedo feedbacks (SAF) stemming from loss of Northern Hemisphere (NH) snow cover (SAFsnow) and sea ice (SAFice). These processes also have a modern-day analog that occurs each year as snow and sea ice retreat from their seasonal maxima, which is strongly correlated with future SAF across an ensemble of climate models. The newly released CMIP6 ensemble offers the chance to test prior constraints through out-of-sample verification, an important examination of EC robustness. Here, we show that the SAFsnow EC is equally strong in CMIP6 as it was in past generations, while the SAFice EC is also shown to exist in CMIP6, but with different, slightly weaker characteristics. We find that the CMIP6 mean NH SAF exhibits a global feedback of 0.25 ± 0.05 W m−2 K−1, or ~61% of the total global albedo feedback, largely in line with prior generations despite its increased climate sensitivity. The NH SAF can be broken down into similar contributions from snow and sea ice over the twenty-first century in CMIP6. Crucially, intermodel variability in seasonal SAFsnow and SAFice is largely unchanged from CMIP5 because of poor outlier simulations of snow cover, surface albedo, and sea ice thickness. These outliers act to mask the noted improvement from many models when it comes to SAFice, and to a lesser extent SAFsnow.

Snow Depth Fusion Based on Machine Learning Methods for the Northern Hemisphere

In this study, a machine learning algorithm was introduced to fuse gridded snow depth datasets. The input variables of the machine learning method included geolocation (latitude and longitude), topographic data (elevation), gridded snow depth datasets and in situ observations. A total of 29,565 in situ observations were used to train and optimize the machine learning algorithm. A total of five gridded snow depth datasets—Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) snow depth, Global Snow Monitoring for Climate Research (GlobSnow) snow depth, Long time series of daily snow depth over the Northern Hemisphere (NHSD) snow depth, ERA-Interim snow depth and Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) snow depth—were used as input variables. The first three snow depth datasets are retrieved from passive microwave brightness temperature or assimilation with in situ observations, while the last two are snow depth datasets obtained from meteorological reanalysis data with a land surface model and data assimilation system. Then, three machine learning methods, i.e., Artificial Neural Networks (ANN), Support Vector Regression (SVR), and Random Forest Regression (RFR), were used to produce a fused snow depth dataset from 2002 to 2004. The RFR model performed best and was thus used to produce a new snow depth product from the fusion of the five snow depth datasets and auxiliary data over the Northern Hemisphere from 2002 to 2011. The fused snow-depth product was verified at five well-known snow observation sites. The R2 of Sodankylä, Old Aspen, and Reynolds Mountains East were 0.88, 0.69, and 0.63, respectively. At the Swamp Angel Study Plot and Weissfluhjoch observation sites, which have an average snow depth exceeding 200 cm, the fused snow depth did not perform well. The spatial patterns of the average snow depth were analyzed seasonally, and the average snow depths of autumn, winter, and spring were 5.7, 25.8, and 21.5 cm, respectively. In the future, random forest regression will be used to produce a long time series of a fused snow depth dataset over the Northern Hemisphere or other specific regions.

Historical Northern Hemisphere snow cover trends and projected changes in the CMIP6 multi-model ensemble

Abstract. This paper presents an analysis of observed and simulated historical snow cover extent and snow mass, along with future snow cover projections from models participating in the World Climate Research Programme Coupled Model Intercomparison Project Phase 6 (CMIP6). Where appropriate, the CMIP6 output is compared to CMIP5 results in order to assess progress (or absence thereof) between successive model generations. An ensemble of six observation-based products is used to produce a new time series of historical Northern Hemisphere snow extent anomalies and trends; a subset of four of these products is used for snow mass. Trends in snow extent over 1981–2018 are negative in all months and exceed -50×103 km2 yr−1 during November, December, March, and May. Snow mass trends are approximately −5 Gt yr−1 or more for all months from December to May. Overall, the CMIP6 multi-model ensemble better represents the snow extent climatology over the 1981–2014 period for all months, correcting a low bias in CMIP5. Simulated snow extent and snow mass trends over the 1981–2014 period are stronger in CMIP6 than in CMIP5, although large inter-model spread remains in the simulated trends for both variables. There is a single linear relationship between projected spring snow extent and global surface air temperature (GSAT) changes, which is valid across all CMIP6 Shared Socioeconomic Pathways. This finding suggests that Northern Hemisphere spring snow extent will decrease by about 8 % relative to the 1995–2014 level per degree Celsius of GSAT increase. The sensitivity of snow to temperature forcing largely explains the absence of any climate change pathway dependency, similar to other fast-response components of the cryosphere such as sea ice and near-surface permafrost extent.

Patterns and trends of Northern Hemisphere snow mass from 1980 to 2018

Warming surface temperatures have driven a substantial reduction in the extent and duration of Northern Hemisphere snow cover1-3. These changes in snow cover affect Earth's climate system via the surface energy budget, and influence freshwater resources across a large proportion of the Northern Hemisphere4-6. In contrast to snow extent, reliable quantitative knowledge on seasonal snow mass and its trend is lacking7-9. Here we use the new GlobSnow 3.0 dataset to show that the 1980-2018 annual maximum snow mass in the Northern Hemisphere was, on average, 3,062±35 billion tonnes (gigatonnes). Our quantification is for March (the month that most closely corresponds to peak snow mass), covers non-alpine regions above 40°N and, crucially, includes a bias correction based on in-field snow observations. We compare our GlobSnow 3.0 estimates with three independent estimates of snow mass, each with and without the bias correction. Across the four datasets, the bias correction decreased the range from 2,433-3,380gigatonnes (mean 2,867) to 2,846-3,062 gigatonnes (mean 2,938)-a reduction in uncertainty from 33% to 7.4%. On the basis of our bias-corrected GlobSnow 3.0 estimates, we find different continental trends over the 39-year satellite record. For example, snow mass decreased by 46gigatonnes per decade across North America but had a negligible trend across Eurasia; both continents exhibit high regional variability. Our results enable a better estimation of the role of seasonal snow mass in Earth's energy, water and carbon budgets.

Evaluation of long-term Northern Hemisphere snow water equivalent products

Abstract. Nine gridded Northern Hemisphere snow water equivalent (SWE) products were evaluated as part of the European Space Agency (ESA) Satellite Snow Product Intercomparison and Evaluation Exercise (SnowPEx). Three categories of datasets were assessed: (1) those utilizing some form of reanalysis (the NASA Global Land Data Assimilation System version 2 – GLDAS-2; the European Centre for Medium-Range Weather Forecasts (ECMWF) interim land surface reanalysis – ERA-Interim/Land and ERA5; the NASA Modern-Era Retrospective Analysis for Research and Applications version 1 (MERRA) and version 2 (MERRA-2); the Crocus snow model driven by ERA-Interim meteorology – Crocus); (2) passive microwave remote sensing combined with daily surface snow depth observations (ESA GlobSnow v2.0); and (3) stand-alone passive microwave retrievals (NASA AMSR-E SWE versions 1.0 and 2.0) which do not utilize surface snow observations. Evaluation included validation against independent snow course measurements from Russia, Finland, and Canada and product intercomparison through the calculation of spatial and temporal correlations in SWE anomalies. The stand-alone passive microwave SWE products (AMSR-E v1.0 and v2.0 SWE) exhibit low spatial and temporal correlations to other products and RMSE nearly double the best performing product. Constraining passive microwave retrievals with surface observations (GlobSnow) provides performance comparable to the reanalysis-based products; RMSE over Finland and Russia for all but the AMSR-E products is ∼50 mm or less, with the exception of ERA-Interim/Land over Russia. Using a seven-dataset ensemble that excluded the stand-alone passive microwave products reduced the RMSE by 10 mm (20 %) and increased the correlation from 0.67 to 0.78 compared to any individual product. The overall performance of the best multiproduct combinations is still at the margins of acceptable uncertainty for scientific and operational requirements; only through combined and integrated improvements in remote sensing, modeling, and observations will real progress in SWE product development be achieved.