1. Introduction
Over the decades, the Arctic has been warming faster than anywhere else on the globe (IPCC 2021). This warming, known as Arctic amplification, is responsible for the accelerated melting of Arctic sea ice and the associated ice–albedo feedback (Manabe and Stouffer 1980; Serreze et al. 2009; Taylor et al. 2022), which further contributes to warming. Concurrent with sea ice loss, exchanges of heat and moisture from the ocean to the atmosphere have increased, leading to a warmer, moister Arctic environment (Boisvert et al. 2015; Boisvert and Stroeve 2015; Taylor et al. 2018; Boisvert et al. 2022). These large, rapid changes are likely affecting all aspects of the hydrologic cycle (Vihma et al. 2016). For example, warming temperatures are associated with a moistening of the atmosphere because warmer air can hold more water vapor (Webster 1994). Warming temperatures are also associated with enhanced moisture transport into the Arctic (Barnes and Polvani 2015) due to larger moisture gradients between lower and higher latitudes (Vihma 2014; Harvey et al. 2014). Several climate modeling studies predict an increase in precipitation in the Arctic in the future (Holland et al. 2007; Lique et al. 2016; Bintanja and Andry 2017; Bintanja 2018; Bintanja et al. 2020; McCrystall et al. 2021; Webster et al. 2021; Parker et al. 2022), with rainfall becoming more prevalent, particularly in the summer (Bintanja and Andry 2017). Future increases in rainfall have been attributed to increasing temperatures and shifts in precipitation phase (Bintanja and Andry 2017), increases in local evaporation due to the loss of sea ice (Bintanja and Selten 2014), and greater moisture transport into the Arctic (Min et al. 2008). Although precipitation is expected to increase in the future, current trends of Arctic precipitation remain inconclusive (Walsh et al. 2020).
Changes in Arctic precipitation, specifically the phase, will have significant impacts on the snow and sea ice pack. Since snow is an efficient insulator, changes to the snowpack due to rainfall events will have implications not only for the surface albedo, but also for the insulating properties of the snow and melt/growth of the ice (Sturm et al. 2002; Stammerjohn and Maksym 2017; Webster et al. 2018). However, the effects of rainfall on the snowpack and sea ice are not well understood since rain can both enhance and mitigate sea ice growth, depending on the timing and amount of rainfall.
Given the importance of precipitation to the Arctic hydrologic cycle, surface energy budget, ocean circulation (Hibler and Zhang 1995; Cullather et al. 2000; Lique et al. 2016), its impact on the melt, growth, and sustainability of the sea ice pack (Serreze and Hurst 2000; Marcovecchio et al. 2022), and the prediction that rainfall will become more prevalent in the future, it is important to understand how the phase of precipitation is represented in current reanalysis products. With the paucity of observational data, gridded and continuous precipitation datasets from global models are often used to force regional models and conduct analysis of Arctic processes and trends. However, since global models do not assimilate observations of precipitation, simulating precipitation is dependent on the representation of large-scale atmospheric circulations and plagued by inaccuracies and assumptions in the parameterization of subgrid-scale cumulus and convection, cloud microphysics, boundary layer processes (Dai 2006). Several of these processes remain poorly understood and are not easily validated in the Arctic, where observations are sparse, and precipitation is difficult to measure (Walsh et al. 1998; Serreze et al. 2005). Thus, precipitation in the Arctic is one of the variables with the largest uncertainty in global climate models (Trenberth 2011).
Previous analyses have demonstrated that the magnitude of precipitation over the Arctic Ocean varies widely between reanalysis products (Lindsay et al. 2014; Boisvert et al. 2018; Barrett et al. 2020; Song et al. 2020) and precipitation frequency is overestimated (Boisvert et al. 2018). Spatially, precipitation patterns tend to agree, and exhibit similar interannual variability (Boisvert et al. 2018; Barrett et al. 2020). Reanalyses can capture the frequency of larger precipitation events, but issues remain in generating the correct magnitude (Lindsay et al. 2014; Boisvert et al. 2018, 2021). The discrepancies are likely due to how each reanalysis treats the fall speeds and condensation of liquid and ice particles in clouds. While there are still many unknowns in Arctic cloud properties and microphysics that control the amount and phase of precipitation, assessing the partitioning of precipitation phase in reanalyses is vital given the implications for the sea ice pack and energy budget analysis.
This work examines the occurrence and timing of rainfall in the Arctic in 1980–2016 from three widely used reanalyses: the European Center for Medium-Range Weather Forecasts (ECMWF) ERA-Interim, ECMWF ERA5, and NASA’s Modern-Era Retrospective Analysis for Research and Application version 2 (MERRA-2). We explore how the rainfall and rainy season have changed in the Arctic and the connection between these changes and temperature trends. We then explain the differences between the products based on their cloud microphysics schemes. The findings provide important insight for using these datasets in Arctic analyses, particularly those related to precipitation and Arctic amplification.
2. Data
Global model configurations, parameterizations, and methods and observations for data assimilation can result in diverging representations of precipitation phase, magnitude, and timing (Stephens et al. 2010; Lindsay et al. 2014). Therefore, we chose to compare three different widely used reanalysis products and satellite-derived observation: ERA-Interim (Dee et al. 2011), ERA5 (Hersbach et al. 2020), and MERRA-2 (Gelaro et al. 2017), as well as CloudSat. We use these datasets to assess the range of spatial patterns and trends and explore the sources of their differences.
a. ECMWF ERA-Interim
ERA-Interim (Dee et al. 2011) is a global reanalysis with an approximate 0.7° (78 km) grid spacing and covers the 1979–present period. The reanalysis uses a 4D-variational assimilation method with a 12-h analysis cycle (Dee et al. 2011). Four different types of sea ice concentration datasets are used for the 1980–2016 period; these include the NCEP two-dimensional variational data assimilation until June 2001, weekly Optimum Interpolation Sea Surface Temperature (OISST) version 2 (Reynolds et al. 2002) (July–December 2001), NCEP real-time global daily SST (2002–April 2009), and the daily Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) product (Donlon et al. 2012). Model cloud microphysics, described in Tiedtke (1993), are based in part on the prognostic cloud condensate scheme of Sundqvist et al. (1989). The model has prognostic variables for cloud condensate and subgrid cloud fraction. At temperatures between 250 and 273 K, cloud condensate phase is determined by a temperature-dependent function (Tiedtke 1993). Precipitation is diagnosed by an autoconversion parameterization that uses the same temperature-dependent function to define the phase of the precipitation, where only snow occurs at temperatures below 250 K, the fraction of rain increases with increasing temperatures between 250 and 273 K, and only rain occurs for temperatures warmer than 273 K (see Table 5 for more details on the relevant physics scheme).
b. ECMWF ERA5
ERA5 is ECMWF’s newer reanalysis product, which offers higher spatial and temporal resolutions. The ERA-Interim and ERA5 reanalyses are produced using the ECMWF Integrated Forecasting System (IFS) model version that was available at the initial time of the production of each reanalysis. Given that ERA5 is a more modern reanalysis, it is produced with a newer version of the ECMWF IFS model. The model version and all associated parameterizations used for each reanalysis remains constant over time. Key improvements for precipitation processes include the increased resolution and changes in the microphysics scheme (see Table 5). ERA5 provides hourly data on a 31-km grid. Like ERA-Interim, it uses a 4D variational assimilation scheme (Hersbach et al. 2020). Until August 2007, ERA5 utilized the Hadley Centre Sea Ice and Sea Surface Temperature dataset version 2 (HadISST2; Titchner and Rayner 2014) and OSI-SAF reprocessing for daily sea ice concentration (Eastwood et al. 2014). Since September 2007, ERA5 uses the OSI-SAF daily sea ice and the OSTIA product (Donlon et al. 2012). ERA5 has an improved global balance of precipitation and evaporation compared to ERA-Interim (Hersbach et al. 2020). The cloud microphysics scheme is significantly upgraded from the version in ERA-Interim, with the same prognostic subgrid cloud fraction, but separate prognostic variables for cloud liquid, cloud ice, rain, and snow (Forbes and Tompkins 2011; Forbes et al. 2011; Table 5). In ERA5, the phase partition for cloud condensate and precipitation is determined by the result of parameterized microphysical processes and is no longer determined by a fixed diagnostic temperature-dependent function.
c. NASA MERRA-2
MERRA-2 (Gelaro et al. 2017) is the most recent version of the MERRA reanalysis system, with output fields available from 1980 to the present. MERRA-2 employs a constraint such that the globally averaged net precipitation minus evaporation (P − E) equals the change in total atmospheric water (Takacs et al. 2016). This was designed to inhibit artificial jumps in precipitation associated with changes in the observing system, but it is thought to have a negligible impact on high latitudes. As compared with MERRA, MERRA-2 employs an updated 3D-variational assimilation system with an incremental analysis update (IAU) technique and has a grid spacing of 0.5° (69 km) (Molod et al. 2015; Bosilovich et al. 2017). Prior to output, computations are performed on a cubed-sphere grid to better resolve polar regions. Sea ice concentrations use the daily SST product of Reynolds et al. (2007) until April 2006 and then OSTIA thereafter. For MERRA-2, the cloud parameterization of Bacmeister et al. (2006) includes large-scale condensation governed by a probability distribution function as described in Molod (2012). For cloud condensate, partitioning between liquid and ice is determined diagnostically by linear ramp function of temperature between 273 and 233 K (see Table 5 for more details).
d. CloudSat
NASA’s CloudSat satellite was launched in April 2006 (Stephens et al. 2008) and remains operational. However, in 2011, CloudSat switched to a daytime-only mode due to a battery power failure, requiring sunlight for it to operate. The Cloud Profiling Radar (CPR) onboard is used to produce cloud and precipitation products. The CPR produces reflectivity profiles in the lowest 30 km of the atmosphere (Tanelli et al. 2008). Precipitation is estimated from these profiles, together with cloud microphysical assumptions and temperatures from ECMWF operational analyses (Kulie and Bennartz 2009; Wood et al. 2013b). The retrieval of snowfall and rainfall profiles from CloudSat data are complex; for more information, we refer readers to L’Ecuyer and Stephens (2002), Haynes et al. (2009), Wood et al. (2013a, 2014), and the CloudSat website (https://www.cloudsat.cira.colostate.edu).
To briefly summarize the retrieval steps, radar reflectivity profiles and physical models are used to derive snowfall and rainfall. These models estimate the vertical distribution of cloud water vertically, evaporation of rain below the cloud base, the droplet size distribution, and the precipitation phase. Some of the models are forced by ECMWF variables, including vertical temperature, pressure, specific humidity, 2-m air temperature, 10-m winds, and SST. The precipitation rate is derived from the path integrated attenuation, which is retrieved from the radar reflectivity profiles. Precipitation rates are classified as stratiform, convective, or shallow based on lookup tables of cloud depths and freezing level heights from a forward model of multiple scattering and the height of the radar attenuation (Hogan and Battaglia 2008). The forward model assumes dry snow properties, which may lead to overestimations of snowfall in situations where melting snow grains are present.
The characteristics of the radar attenuation are unique to convective and stratiform profiles. In particular, the attenuation characteristics of stratiform profiles typically have an inflection point near the freezing level, which corresponds to the melting layer. This inflection point is assumed to be the top of the liquid precipitation in the cloud column, unless it is 500 m above the ECMWF freezing point; in that case, it is classified as convective. Since the ECMWF temperatures are used in this determination, there may be a discrepancy between the location of the melting layer detected in the radar reflectivity profiles versus that simulated by ECMWF. In this situation, snowfall rates may be overestimated if melting snow particles occur above the ECMWF temperature threshold. For further information on caveats and uncertainties in the retrievals, we additionally refer readers to the Algorithm Theoretical Basis Documents of the CloudSat products.
The monthly 1° × 1° gridded N_Rain and N_Precip products used in the analysis give the total number of CloudSat profiles in bins collocated with 2C-RAIN-PROFILE rain_rate and 2C-SNOW-PROFILE snowfall_rate_sfc, which are greater than zero; N_Rain is the number of rainfall profiles found, and N_precip is the total number of both snow and rain profiles found. The CloudSat data products are used to assess the precipitation frequency estimated by the reanalyses. CloudSat snowfall products have previously been used in other studies over the Arctic Ocean (Cabaj et al. 2020; Edel et al. 2020; von Lerber et al. 2022), although validation in these data-sparse areas is difficult.
3. Methods
In this study, we use daily averages of precipitation and 2-m air temperatures for ERA-Interim, ERA5, and MERRA-2 for 1980–2016 and focus on the occurrence (days) of rainfall and above freezing temperatures. The study domain covers the Arctic Ocean areas north of 60°N latitude. The “entire Arctic” encompasses all ocean regions in Fig. 1. We further split up the Arctic Ocean into the central Arctic, peripheral seas, and North Atlantic in our analysis.
Region map. The Arctic Ocean at latitudes greater than 60°N. Light green is the central Arctic, yellow is the peripheral seas, and orange is the North Atlantic as used in this study.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Previous work has demonstrated that reanalyses tend to agree on the occurrence of major precipitation events, but produce widely different and potentially biased magnitudes (e.g., Boisvert et al. 2018). It is important to note that reanalyses have their caveats: the magnitude of MERRA-2 precipitation over the Arctic is much higher compared to other reanalyses, and ERA-Interim tends to produce fewer days with snowfall [see Boisvert et al. (2018) for more detail], while the overall magnitude is like other reanalyses. Comparisons of precipitation in the Southern Ocean (Boisvert et al. 2020) and Arctic Ocean (Barrett et al. 2020; Cabaj et al. 2020) using ERA5 showed that its magnitude and occurrence of precipitation fall largely in the middle of the spread between other reanalyses. All reanalyses used here produce modest precipitation amounts near-daily, which is likely unrealistic. Therefore, this residual “drizzle” is removed from the rainfall products using a 1 mm day−1 threshold to isolate more geophysically realistic precipitation for the analysis. Therefore, in this analysis only days where the rainfall is greater than 1 mm day−1 are classified as rainfall days.
Data assimilated for producing reanalysis products have varied over the time record, meaning that observed trends over time need to be carefully considered. This work addresses these caveats by analyzing a range of reanalysis products and discussing the pertinent model differences and associated changes that may influence precipitation trends.
4. Results
a. Rainfall days in the Arctic Ocean
Figure 2 shows rainfall days in the entire Arctic between 1980 and 2016. Reanalyses agree that the least frequent rainfall occurs in the central Arctic (8 days yr−1), rainfall frequency increases in the peripheral seas (18 days yr−1) and is most frequent in the North Atlantic (57 days yr−1) (Fig. 2a, Table 1). Consistent with previous findings (e.g., Boisvert et al. 2018), ERA-Interim overall has the most frequent rainfall days and MERRA-2 has the least (Fig. 2a), with an average of 37 and 27 days for the entire Arctic, respectively. The patterns of interannual variability in rainfall frequency mirrors those of average frequency, where the least variability occurs in the central Arctic (3 days yr−1) and increases at lower latitudes, with the greatest in the North Atlantic (37 days yr−1) in ERA5 and MERRA-2 (Fig. 2b, Table 1). While the spatial variability of rainfall frequency in ERA-Interim is somewhat similar to other reanalyses, the magnitude is larger, particularly over the central Arctic. The increased variability is caused by more frequent rainfall compared to the other reanalyses.
Rainfall days from ERA-Interim, ERA5, and MERRA-2 reanalyses between 1980 and 2016. (a) Average number of rainfall days per year, (b) interannual variability in the number of rainfall days per year, and (c) trend in the number of rainfall days per year.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Averages, standard deviations, and trends for 1980–2016 rainfall days in ERA-Interim, ERA5, and MERRA-2. First day means the first day of the year when rainfall occurs, and last day means the last day of the year when rainfall occurs. Trend units are days per decade. Trends in bold are statistically significant at the 90th percentile based off a Student’s t test.
1) Comparison with CloudSat observations
To assess whether the reanalyses are producing too much or too little rainfall, the rainfall counts for each reanalysis product are compared with those from CloudSat in JJA for 2007–10. The percentage of rainfall counts to total precipitation counts are shown in Fig. 3. Similar spatial patterns are apparent between CloudSat, ERA5, and MERRA-2, with the least amount of rainfall occurring in the central Arctic and more in the peripheral seas and the North Atlantic. During the sea ice minimum of 2007, for CloudSat, ERA5, and MERRA-2 there were more occurrences of rainfall in the Chukchi and Beaufort Seas where there was a drastic reduction in sea ice coverage compared to the other years. For all years, ERA-Interim produces the most rainfall, even compared to the CloudSat observations. The differences between these rainfall percentages between CloudSat and ERA-Interim, ERA5, and MERRA-2 are shown in Fig. 4. ERA-Interim produces up to 30% more rainfall in JJA compared to CloudSat over the majority of the Arctic Ocean. CloudSat shows slightly less rainfall in the Kara and Barents Seas, and slightly more rainfall in the central Arctic compared to ERA5. This is similar to MERRA-2, except that MERRA-2 produced much less rainfall compared to CloudSat over the central Arctic by 30%. Therefore, it appears that ERA-Interim consistently produced too much rainfall relative to satellite observations and other reanalyses, whereas ERA5 rainfall days are more in line with the observations.
Rainfall count percentages (number of rainfall counts/total precipitation counts) for JJA between 2007 and 2010 for (a) CloudSat, (b) ERA-Interim, (c) ERA5, and (d) MERRA-2.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Rainfall count percentages (number of rainfall counts/total precipitation counts) differences for JJA between 2007 and 2010: (a) ERA-Interim and CloudSat differences, (b) ERA5 and CloudSat differences, and (c) MERRA-2 and CloudSat differences.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
2) Rainfall trends
Although the frequency of rainfall varies, the overall average trend for the entire Arctic in 1980–2016 for the aggregated reanalyses is a general increase (+2 days decade−1). This trend is statistically significant for MERRA-2 and ERA5 (Fig. 2c, Table 1). However, there are notable differences between geographic regions among the reanalyses. The largest trends in rainfall frequency averaged across all reanalyses occurs in the North Atlantic (+5 days decade−1; statistically significant). When considering the reanalyses separately, they all show an increase in the rainfall frequency in the Kara and Barents Seas, and off the southeast coast of Greenland (more than 10 days decade−1). ERA5 and MERRA-2 produce smaller increases in rainfall frequency in the East Siberian and Chukchi Seas. ERA-Interim has an increasing rainfall frequency in most of the central Arctic (+2 days decade−1, statistically significant), where ERA5 and MERRA-2 have no trend. Statistically significant positive trends in ERA5 and MERRA-2 occur in the North Atlantic (Table 1).
When looking at seasonal rainfall, Figs. 5a–c reveal that the largest increasing trend in the number of rainfall days across all reanalyses is occurring in autumn (September–November), specifically in the North Atlantic and peripheral seas (+2–5 days decade−1). The spatial pattern of trends in summer (June–August) differs between reanalyses, with MERRA-2 having increases in the North Atlantic and peripheral seas, ERA5 in the Kara and Barents Seas, and ERA-Interim in the central Arctic along with decreases in the lower latitudes of the North Atlantic. The positive trends in spring (March–May) in all reanalyses occur in the North Atlantic and off the coast of southeast Greenland, but not over areas of the central Arctic and the consolidated sea ice pack.
Seasonal trends in rainfall for 1980–2016 for ERA-Interim, ERA5, and MERRA-2. (a) Trends in the number of rainfall days in March, April, and May (MAM), (b) Trends in the number of rainfall days in June, July, and August (JJA). (c) Trends in the number of rainfall days in September, October, and November (SON).
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Figures 2c and 5a–c show that the number of rainfall days are increasing in the entire Arctic over 1980–2016, but is rainfall becoming a more predominant phase of precipitation? Looking at the ratio of rainfall to total precipitation days (Fig. 6a) and the change in this ratio over time (Fig. 6b), we can determine if the occurrence of rainfall is increasing compared to snowfall. Over 1980–2016, all reanalyses show the smallest percentage of rainfall (snowfall dominates) just north of Greenland in the central Arctic, and the percentage increases as latitudes decrease. In the central Arctic, ERA-Interim produces rainfall ∼29.5%, ERA5 ∼10.8%, and MERRA-2 ∼5.2% of the time, demonstrating that snowfall overall dominates precipitation (Table 2). Over the period, all reanalyses show large and statistically significant increases to this ratio in the North Atlantic (Table 2). MERRA-2 depicts large increases in the peripheral seas (statistically significant), and ERA-Interim has large increases in the central Arctic. Although the spatial trends are not uniform (Fig. 6b), all reanalyses show a statistically significant increase in this ratio of 1.90%–3.16% decade−1 in the North Atlantic (Table 2). These increases in the ratio of rainfall to total precipitation days reinforce the conclusion that regardless of whether the frequency of precipitation is changing, the liquid phase of precipitation is becoming more predominant.
Ratio of rainfall days to total precipitation days for 1980–2016 for ERA-Interim, ERA5, and MERRA-2. (a) Average rainfall ratio. (b) Trend in the rainfall ratio.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Ratio of rainfall days to total precipitation days for ERA-Interim, ERA5, and MERRA-2 for the entire Arctic and regions. Standard deviations are in parentheses. Trends are in percent per decade. Trends that are statistically significant at the 90th percentile are in bold.
b. Rainfall days and air temperature
To better understand the changes in rainfall occurrence in the Arctic, we have analyzed trends in air temperatures as a possible driver of precipitation changes. Figure 7 suggests that warming temperatures indeed may be a driver for a rainier Arctic. The number of days that the 2-m air temperatures are above freezing is statistically significant and increasing across the Arctic in all reanalyses (∼+6 days decade−1; Fig. 7a, Table 3). The largest increases are found in the North Atlantic (+8 days decade−1, statistically significant), southeast of Greenland, in the Chukchi and Beaufort Seas, and the Bering Strait. The spatial patterns are similar to the trends in the rainfall occurrences and ratios shown in Figs. 2c and 6b. Trends in the central Arctic differ between the reanalyses, with MERRA-2 showing an increase in warm days (5 days decade−1, statistically significant), but with the smallest increases in the eastern portion of the central Arctic, while ERA-Interim and ERA5 show no significant trend.
Rainfall and temperature trends and relationship from ERA-Interim, ERA5, and MERRA-2 reanalyses for 1980–2016. (a) Trends in the number of days when 2-m temperatures are above freezing, (b) correlation between the number of rainfall days and the number of days where the 2-m temperature is above freezing, and (c) scatterplot of the trend in the number of rainfall days and the trend in the number of days when the 2-m temperatures are above freezing. Correlation coefficients (r) in bold are statistically significant at the 90th percentile using a P value.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Averages, standard deviations, and trends for 1980–2016 2-m temperature days in ERA-Interim, ERA5, and MERRA-2. First T > 0°C denotes the first day of the year when the temperature is above freezing. Last T > 0°C denotes the last day of the year when the temperature is above freezing. Trend units are days per decade. The 2-m air temperature is represented by T. Trends in bold are statistically significant at the 90th percentile based off a Student’s t test.
The increasing number of above-freezing days in the Arctic might be tied to increasing rainfall frequency. Figure 7b shows the spatial correlations between the number of days above freezing and the number of rainfall days. MERRA-2 has the highest spatial correlations overall, specifically in the peripheral seas and the central Arctic. All reanalyses show high correlations in the North Atlantic. Figure 8 presents scatterplots of the number of days where the temperature is above freezing and the number of rainfall occurrences, for the entire Arctic and specific regions, and demonstrates that these variables, although not one-to-one, are highly correlated. While there are many more days above freezing than rainfall occurrences, the more often above-freezing days occur, the higher the likelihood of rainfall (Fig. 8a). Regionally, the highest correlations between temperature and rainfall occur in the North Atlantic, followed by the peripheral seas, and the lowest correlations in the central Arctic (Figs. 8c,d). ERA-Interim (MERRA-2) has the lowest (highest) correlations overall.
The number of days above freezing and the number of rainfall days scatterplots of 2-m temperature and rainfall for 1980–2016 for ERA-Interim, ERA5, and MERRA-2 for the (a) entire Arctic and for the regions of the (b) North Atlantic, (c) peripheral seas, and (d) central Arctic (see Fig. 1). Correlations (r) in bold are statistically significant at the 90th percentile using a P value.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
The air temperature and rainfall frequency correlations prompt the following question: Has the trend in the number of above-freezing days driven the trends in the rainfall? Figure 7c shows that indeed the trends in the rainfall days and the days above freezing are correlated, but increasing days above freezing can only explain 17%–69% of the variance in the rainfall frequency trend. ERA5 has the highest correlation and MERRA-2 has the lowest correlation. The trend in the number of above-freezing days per year is larger than the trends in rainfall days, demonstrating that just because there are more warm days in the Arctic, does not mean there will be more rain days. Other factors and weather variability are clearly at play here.
c. Changes to the rainy season
To our knowledge, recent changes in the rainy season length and timing have not been previously assessed and could have a significant impact on the albedo, and melt and growth of the snow and sea ice pack in the Arctic. However, future mid-twentieth-century changes to the length of the rainfall season have been assessed by Landrum and Holland (2020), who found an increase of 20–60 days in the rainfall season over the Arctic based on large ensembles from five Earth system models from phase 5 of the Coupled Model Intercomparison Project (CMIP5). Dou et al. (2022) have also found an increase in rainfall and a lengthening of the rainy season in future climate projections from CMIP5 Earth system models. Figure 9a shows the 1980–2016 average date of the first year’s rainfall and the differences between the reanalyses. Here, ERA-Interim has the earliest average first rainfall day (day 96, 6 April), whereas the average first day in ERA5 and MERRA-2 occurs roughly 40 days later (day 137, 17 May) for the entire Arctic (Table 1). Across the reanalyses, average rainfall occurs earliest in the year at lower latitudes in the North Atlantic (day 61, 2 March), and occurs latest in the central Arctic at high latitudes (day 178, 27 June). The average rainfall first day occurs on day 150 (30 May) in the peripheral seas (Table 1). The interannual variability of the average first rainfall day is shown in Fig. 9b. The North Atlantic has areas of both the lowest and highest variability, with the lowest occurring in the lower latitudes of the North Atlantic where it rains most of the time, and the highest occurring at the sea ice edge, which varies from year to year (Fig. 10). The central Arctic and peripheral seas have similar variability (42 and 35 days, respectively). ERA-Interim’s variability behaves most differently, with large areas of the Chukchi, Beaufort, and Kara Seas along with areas off the northeast of Greenland showing large variability in rain onset.
First rainfall and 2-m temperatures from 1980 to 2016 for ERA-Interim, ERA5, and MERRA-2. (a) Average day of rainfall onset (first day with rainfall greater than 1 mm day−1). (b) Variability in the rainfall onset date. (c) Trend in the day of rainfall onset. (d) Trends in the day of the year when the first day above freezing occurs.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Sea ice concentration trends over 1980–2016: (a) annual, (b) MAM, (c) JJA, and (d) SON.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
While not statistically significant, all reanalyses show an average general trend toward rainfall occurring earlier in the year (−2 days decade−1) over 1980–2016 for the entire Arctic (Fig. 9c, Table 1). The largest trends for ERA5 and MERRA-2 are in the North Atlantic along the marginal ice zone, whereas ERA-Interim has largest trends slightly farther to the north in the central Arctic (−7 days decade−1). Most other areas of the Arctic Ocean show a near neutral but slightly negative trend, with no statistical significance. (Fig. 9c, Table 1).
All reanalyses have very similar averages of the first day above freezing, with their cumulative averages for the entire Arctic (day 125, 4 May), central Arctic (day 164, 13 June), peripheral seas (day 156, 5 June), and North Atlantic (day 66, 7 March) (Table 1). The first day of rainfall always occurs after the first day above freezing for MERRA-2 and ERA5, but not for ERA-Interim (Table 1). Reanalyses have similar trends in the first occurrence of 2-m air temperatures above freezing, suggesting that warmer temperatures are occurring earlier in the year (−3 days decade−1) (Fig. 9d). MERRA-2 tends to have the largest change in the first day of above freezing over the majority of the Arctic (−5 days decade−1, statistically significant), with ERA-Interim and ERA5 having these trends confined mostly in the peripheral seas (−2 and −3 days decade−1, respectively, statistically significant) and the North Atlantic for ERA5. Other statistically significant trends are occurring in MERRA-2 in the central Arctic, peripheral seas, and North Atlantic, with warmer days occurring earlier in the year (Table 1).
When comparing the first day above freezing and the first rainfall day, there are high correlations in the North Atlantic and the peripheral seas near the coasts, with much smaller and slightly negative correlations present in the central Arctic (Fig. 11a). ERA-Interim has many occurrences when the first rainfall day occurs much earlier than the first day above freezing (Fig. 11b). MERRA-2 shows some instances of earlier rainfall, while ERA5 has rainfall occurring nearly always after the first day above freezing. When split into the individual regions (Fig. 11c), it is apparent that ERA-Interim produces rainfall before warm temperatures in all regions, especially in the central Arctic, where rainfall is occurring around day 50 (19 February) but the first day above freezing is consistent around day 160 (9 June). These scatterplots also highlight a delay or lag in ERA5 and MERRA-2 between the first day above freezing and first day of rainfall (points below the one-to-one line) (Figs. 11b,c), and not for ERA-Interim. This is driven by the fact that just because average daily temperatures reach the freezing point does not guarantee that is going to precipitate that day, just that if it were to precipitate, then atmospheric temperatures could be warm enough for that precipitation to be rainfall. This lag for ERA5 and MERRA-2 tends to be around 20 days. For MERRA-2, and to a greater extent for ERA-Interim, in the central Arctic, the first rainfall day sometimes occurs before the first warm day of the year (points above the one-to-one line) and could be occurring when temperatures are hovering just below freezing. This occurs to a much lesser extent in ERA5. These instances may occur when the temperature was above freezing for a brief period of the day when rainfall occurred, but the average daily temperature could have been below freezing due to the diurnal cycle that is more extreme in the spring and fall months.
Relationship between first day 2-m temperatures above freezing and first day of rainfall for ERA-Interim, ERA5, and MERRA-2 for 1980–2016. (a) Correlations between temperatures and rainfall days, (b) scatterplot of first day above freezing and first rainfall day for the entire Arctic, and (c) scatterplots of first day above freezing and first rainfall day for the North Atlantic, peripheral seas, and the central Arctic. Correlations (r) in bold are statistically significant at the 90th percentile using a P value.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Overall, there is a trend in the Arctic toward rainfall occurring later in the year during 1980–2016 (Fig. 12a). On average, ERA-Interim always has the latest last day of rainfall on day 291, 18 October, and MERRA-2 always has the earliest last day of rainfall on day 258, 15 September, in the entire Arctic. Each reanalysis has the latest day of rainfall occurring in the North Atlantic (day 320, 16 November), followed by the peripheral seas (day 257, 14 September) and last in the central Arctic (day 226, 14 August) (Table 1). The interannual variability with regard to the last day of rainfall differs between the reanalyses (Fig. 12b). For example, ERA-Interim has large variability (>30 days) that covers most of the central Arctic, whereas areas of large variability are only present in the marginal ice zone east of Greenland and the Barents Sea for ERA5 and MERRA-2 (Figs. 12b and 10). Little to no variability in the last day of rainfall is present in lower latitudes of the North Atlantic, where rainfall is most prevalent throughout the year.
Last rainfall and 2-m temperatures from 1980 to 2016 for ERA-Interim, ERA5, and MERRA-2. (a) Average day of last day of rainfall (with rainfall greater than 1 mm day−1). (b) Variability in the last day of rainfall. (c) Trend in the last day of rainfall. (d) Trends in the day of the year when the last day above freezing occurs.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Autumn rainfall is occurring later in the season, specifically in the North Atlantic (+4 days decade−1) and the peripheral seas (+3 day decade−1) (Fig. 12c, Table 1); however, the locations of these trends are not consistent between the reanalyses. In the central Arctic for example, ERA-Interim shows a large, statistically significant trend of 7 days decade−1 delay in the end of the rainy season, while the other reanalyses demonstrate a nonsignificant trend of 2–3 days decade−1 (Table 1). The delay in the end of rainfall is likely driven by an increase in the occurrence of temperatures above freezing extending later in the year (Fig. 12d). Positive trends are seen in the North Atlantic near the marginal ice zone, and in areas of the peripheral seas. On average, the warming extended roughly 4 days decade−1, with MERRA-2 and ERA-Interim having statistically significant trends for the entire Arctic (Table 3). Statistically significant trends also occur in the peripheral seas and North Atlantic in MERRA-2 and ERA5. The timing of the last day of warm temperatures in reanalyses is similar for the entire Arctic, with ERA-Interim (MERRA-2) always having the latest (earliest) warm day, except for the North Atlantic where ERA5 has the earliest (Table 1).
Spatially, there are positive correlations between the last rainfall day and the last day above freezing for all reanalyses (Fig. 13a, Table 4). Although highly correlated, scatterplots of the entire Arctic’s last day above freezing and last day of rainfall reveal that ERA-Interim’s last day of rainfall occurs much later than the last day above freezing; however, MERRA-2 has many occurrences when the last day of rainfall is well before the last day above freezing, and ERA5’s last day of rainfall typically occurs just slightly before the last day above freezing (Fig. 13b). When broken down into regions, MERRA-2 has the highest correlations in the North Atlantic and peripheral seas, but small correlations in the central Arctic. ERA5 has positive correlations over the entire Arctic, and ERA-Interim has high correlations in parts of the North Atlantic, with small correlations over the rest of the Arctic (Table 4). ERA-Interim consistently has the last rainfall day occurring much later than the last day above freezing and the lowest correlations (Fig. 13c, Table 4). ERA5 has the highest correlations, with both the central Arctic and peripheral seas showing the last rainfall occurring before the last day above freezing, but this same pattern is not seen in the North Atlantic. MERRA-2 is similar to ERA5, but the last rainfall day occurs much earlier compared to the last day above freezing (Fig. 13c, Tables 1 and 3).
Relationship between timing of rainfall and 2-m air temperature above freezing. (a) Correlations of the last day above freezing and the last day of rainfall, (b) scatterplot of the last day above freezing vs the last day of rainfall for the entire Arctic, and (c) scatterplots of the last day above freezing vs the last day of rainfall for the North Atlantic, peripheral seas, and the central Arctic. Correlations (r) in bold are statistically significant at the 99th percentile using a P value.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
Correlation coefficients (R) for 1980–2016 rainfall days and 2-m temperature days in ERA-Interim, ERA5, and MERRA-2. First day means the first day of the year when rainfall occurs, and last day means the last day of the year when rainfall occurs. Similarly, first T > 0°C denotes the first day of the year when the temperature is above freezing, and last T > 0°C denotes the last day of the year when the temperature is above freezing. Correlation coefficients in bold are statistically significant at the 90th percentile using a P value.
Similar to the springtime during the transition of polar night to the sunlit season, in the fall when the sunlit season is transitioning to polar night, there are large diurnal temperature differences. The rainfall (snowfall) could be occurring during the short period of the day when temperatures were above (below) freezing, but the average temperature for the day was below (above) freezing. This could explain the lag between the last rainfall day and the last day above freezing seen MERRA-2 and ERA5 in Fig. 13.
The analysis demonstrates that the timing of rainfall in ERA-Interim is least correlated to the 2-m temperatures (Table 4), and rainfall is often occurring before the first (Fig. 11b) and after the last day above freezing (Fig. 13b). Therefore, it is important to explore on average how many times a year that rainfall occurs when temperatures are below freezing (Fig. 14a). ERA-Interim has more rainfall occurrences when average temperatures are below freezing compared to ERA5 and MERRA-2. Specifically, in the central Arctic, ERA5 and MERRA-2 have zero days when this occurs, but ERA-Interim has ∼7 days yr−1. MERRA-2 and ERA5 have 2–3 days yr−1 in the peripheral seas, where ERA-Interim has 5 days yr−1. In the North Atlantic, ERA-Interim has 12 days yr−1, while MERRA-2 and ERA5 have 6 and 4 days yr−1, respectively. ERA-Interim also has greater interannual variability in the entire Arctic compared to the other reanalyses (Fig. 14b). These findings warrant a more detailed discussion of precipitation phase partitioning in the cloud-microphysics schemes employed by these reanalysis models.
Temperatures below freezing and rainfall occurrence over 1980–2016 for ERA-Interim, ERA5, and MERRA-2. (a) Average frequency of rainfall that occurs when the 2-m temperatures are below freezing, and (b) variability in the frequency of rainfall that occurs when the 2-m temperatures are below freezing.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
d. What are the drivers of differences in precipitation frequency and timing in reanalysis datasets?
The precipitation phase and amount in reanalysis products largely depends on how the model for each reanalysis treats and parameterizes the microphysical processes in stratiform and convective clouds (Table 5). In these reanalyses, the partitioning of precipitation phase into rain and snow is treated differently and depends on the specific assumptions of mixed-phase cloud generation and dissipation, the formation of rain and snow precipitation through autoconversion and accretion processes, evaporation, sublimation, and the rain freezing and snow melting parameterizations. There are also different treatments of shallow and deep convection between these reanalyses, which will also affect the precipitation phase in convective situations (Table 5). The representation of mixed-phase cloud processes is complex, and many issues remain with simulating the correct cloud properties and phase due to many unknowns with Arctic cloud processes and properties (Morrison and Pinto 2006).
Breakdown of ERA-Interim, ERA5, and MERRA-2 cloud, microphysics, and convection schemes.
ERA-Interim’s cloud and precipitation scheme has only one prognostic variable for cloud condensate and diagnoses precipitation, so the phase of the cloud (water vs ice particles) and precipitation (rain vs snow) is determined by a diagnostic function of temperature (Tiedtke 1993). Clouds with temperatures < 250 K are treated as ice clouds, and ice crystals are allowed to grow via autoconversion into snowfall. Clouds with temperatures > 273 K are categorized as liquid clouds and droplets grow via autoconversion into rainfall. Clouds with temperatures of 250–273 K are categorized as mixed-phase clouds; condensate phase is partitioned as a function of temperature and the phase of the generated precipitation follows the same temperature-dependent function from all snow at 250 K, with increasing rain fraction with temperature, to all rain production at 273 K. There is no rain freezing process in ERA-Interim. This causes a greater fraction of rainfall to occur in ERA-Interim, in temperatures specifically for 250–273 K, when likely snowfall would have occurred (Fig. 14). This treatment of mixed-phase clouds impacts the amount of rainfall days and cloud liquid water and ice produced in the ERA-Interim reanalysis and the differences found in this study.
ERA5’s cloud and precipitation scheme has a more complex treatment of the mixed-phase with separate prognostic variables for cloud water, cloud ice, rain, and snow (Forbes and Tompkins 2011; Forbes et al. 2011). Whereas ERA-Interim has a fixed proportion of cloud liquid water versus ice at a given temperature, ERA5 represents the different microphysical processes that affect hydrometeor phase and allows for liquid and ice proportions to have significant variability at different temperatures (Forbes and Ahlgrimm 2014). The minimum temperature for supercooled liquid water occurrence was also changed from 250 to 235 K in closer agreement with observations (Hogan et al. 2004), but this is unlikely to have much impact on rain production as most of the cloud is in the ice-phase in this temperature range. The changes to the mixed-phase cloud and precipitation representation in the new scheme have likely improved the partitioning of precipitation of snowfall and rainfall in ERA5 and result in little to no rainfall occurring when temperatures are below freezing, similar to MERRA-2 (Fig. 14).
MERRA-2 produces two types of clouds: “anvil” via convection or “large scale,” which uses a PDF-based two-moment condensation calculation scheme (Molod 2012). Both cloud types lose moisture from four processes: 1) evaporation, 2) autoconversion of either liquid or mixed phase, 3) sedimentation of ice, and 4) accretion of falling precipitation. MERRA-2 treats liquid and ice particles separately and partitions condensate diagnostically by a linear ramp function on a temperature-dependent scale where the amount of ice particles increases linearly between 233 and 273 K (Barahona et al. 2014). MERRA-2 also uses a “Sundqvist-like” parameterization for autoconversion in mixed-phase clouds which is also temperature dependent (Rienecker et al. 2008). Ice condensate does not undergo autoconversion in the scheme, but transitions directly to settling and ice fall. For precipitation phase partitioning, precipitation formation at temperatures < 273 K is assumed to be frozen for the flux to the accretion portion of the scheme. In this scheme, precipitation melting can occur where anything above 278 K melts instantly and melting occurs at a time scale of 5000 seconds at temperatures between 273 and 278 K. Due to the partitioning of liquid and ice, and the temperature dependence, the rainfall occurrences in MERRA-2 behave similarly to ERA5, with very little rainfall occurring at below freezing temperatures and much less rainfall occurring compared to ERA-Interim.
Despite producing the most rainfall, on average, ERA-Interim produces the most cloud ice water (least cloud liquid water) in the Arctic Ocean compared with ERA5 and MERRA-2 (Figs. 15a,b). Since ERA-Interim produces the most cloud ice and the least cloud liquid, likely because only cloud ice is able to exist at <250 K, whereas ERA5 and MERRA-2 can have cloud liquid between 235 (233) and 250 K, the ratio of cloud liquid to cloud ice is the smallest at roughly 50% compared to ERA5 and MERRA-2, with both having about 70% on average (Fig. 15c). Trends in this ratio of cloud liquid water to total cloud water show that the amount of cloud liquid water has become more predominant over the 1980–2016 time period for all reanalyses (Fig. 15d). ERA-Interim and MERRA-2 have both seen a larger increase in cloud liquid water (∼3% decade−1) compared to ERA5, especially in the central Arctic and in the Canadian Archipelago (for MERRA-2). For MERRA-2, these areas of large increasing trends are because there is a decrease in cloud ice in the Canadian Archipelago, along with an increase in cloud liquid water over the time period (not shown). This large trend seen in ERA-Interim and MERRA-2 could be related to the fact that in mixed-phase clouds, the phase in ERA-Interim is quadratic temperature dependent for 250–273 K and linear ramp temperature dependent for 233–273 K in MERRA-2 (Table 5), and thus there is more liquid at warmer temperatures, especially as temperatures warm with climate change. There is little change in this ratio and rainfall in ERA5 because mixed-phase clouds exist in the 235–273-K temperature range and liquid and ice are prognostic variables. MERRA-2 also is seeing a larger increase in the number of days above freezing compared to ERA5, especially in the Canadian Archipelago (Table 3, Fig. 7a) and this could explain why there is even more liquid cloud water being produced in MERRA-2 and a decrease in cloud ice. ERA-Interim is likely producing more rainfall because the number of days where the temperature falls between 250 and 273 K (mixed-phase clouds threshold) is increasing (∼+3 days yr−1) and temperatures > 273 K (liquid cloud threshold) are also increasing (∼+1.5 days yr−1), while days with temperatures < 250 K are decreasing at (∼−1 day yr−1) (Fig. 16). Thus, there is an increase in liquid and mixed-phase clouds, which produce only rainfall in ERA-Interim. This increase could be responsible for the increases in the rainfall occurrences in the central Arctic in the summer, which was not seen in the other reanalyses.
(a) Annual average vertically integrated cloud ice water (CIW), (b) annual average vertically integrated cloud liquid water (CLW), (c) annual average ratio of vertically integrated cloud liquid water to vertically integrated total cloud water, and (d) trend in the ratio of vertically integrated cloud liquid water to vertically integrated total cloud water from ERA-Interim, ERA5, and MERRA-2 over 1980–2016.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
ERA-Interim trends in the number of days annually when temperatures fall (left) below 250 K, (center) within 250–273 K, and (right) above 273 K.
Citation: Journal of Climate 36, 19; 10.1175/JCLI-D-22-0428.1
The relatively larger frequency of rainfall in ERA-Interim, specifically in the summer and autumn in the central Arctic, could be caused by warming temperatures, increasing the occurrence of mixed-phase clouds (250–273 K), and thus rainfall. When comparing the rainfall from ERA5 and ERA-Interim, it was found that ERA-Interim produces much more rainfall over the entire Arctic, further demonstrating the importance of representing mixed-phase cloud and precipitation processes in models with more appropriate complexity and realism, and is a crucial driving factor in the number of rainfall days in reanalysis datasets over the Arctic.
5. Conclusions
The results of this study demonstrate the following:
-
The number of rainfall days has increased in the Arctic during 1980–2016, most notably in the fall and in the North Atlantic region.
-
Rainfall is occurring earlier and later in the year at similar magnitudes than in the past, thus lengthening the rainy season.
-
An increase in rainfall events appears to be tied to a warming Arctic. Earlier (later) above-freezing days in the year are correlated with earlier (later) rainfall events, but temperature is not the sole driver of rainfall occurrence.
-
ERA-Interim’s treatment of mixed-phase clouds and precipitation phase with a diagnostic temperature-dependent function is too simplistic, and produces too much rain from mixed-phase clouds between temperatures 250 and 273 K. This leads to an overestimation of rain at below zero temperatures compared to ERA5 and MERRA-2. Also, as temperatures warm, ERA-Interim is likely to show too high trends in rainfall increases where more temperatures fall within the mixed-phase cloud temperature regime.
Our analysis has shown that a warming Arctic is related to a rainier Arctic overall. The predominance of rainfall in total precipitation and the length of the rainy season are both increasing. As the Arctic warms and rainfall events become more prevalent, future research is needed to improve understanding of the effects of rainfall on albedo, the surface energy balance, and sea ice mass balance. Our analysis highlights that more modern reanalyses (e.g., ERA5 and MERRA-2), with more sophisticated parameterizations, produce rainfall patterns roughly consistent with each other and when compared with CloudSat observations. However, our findings demonstrate that reanalysis precipitation occurrence, phase, and trends are highly sensitive to model configuration and parameterizations. While there are substantial differences between methods for reanalysis and climate modeling, particularly in terms of data assimilation, the findings in this study underscore the importance of microphysics scheme and the temperature range threshold for accurately simulating mixed-phase clouds, as well as rainfall frequency and amount. Better understanding of Arctic cloud phase and microphysics is critical for improving climate models. In particular, improvements in the representation of phases of precipitation in the rapidly warming Arctic will facilitate accurate analyses of changing weather patterns over time.
Acknowledgments.
This work was funded by NASA ROSES Weather and Dynamics proposal 80NSSC20K0922. The authors thank the reviewers for their constructive criticism and feedback during the review process for this manuscript. The authors thank the reviewers for providing constructive and helpful feedback on this manuscript.
Data availability statement.
ERA-Interim data can be downloaded from https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era-interim. ERA5 data can be downloaded from https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5. MERRA-2 data were downloaded from the NASA GES DISC: Global Modeling and Assimilation Office (GMAO) (2015), MERRA-2 tavg1_2d_flx_Nx: 2d,1-Hourly, Time-Averaged, Single-Level Assimilation Surface Flux Diagnostics V5.12.4, Greenbelt, MD, USA, Goddard Earth 390 Sciences Data and Information Services Center (GES DISC), at https://doi.org/10.5067/7MCPBJ41Y0K6. Sea ice persistence data are taken from the National Snow and Ice Data Center at https://nsidc.org/data. CloudSat data can be downloaded from https://www.cloudsat.cira.colostate.edu/.
REFERENCES
Bacmeister, J. T., M. J. Suarez, and F. R. Robertson, 2006: Rain reevaporation, boundary layer–convection interactions, and Pacific rainfall patterns in the AGCM. J. Atmos. Sci., 63, 3383–3403, https://doi.org/10.1175/JAS3791.1.
Barahona, D., A. Molod, J. Bacmeister, A. Nenes, A. Gettelman, H. Morrison, V. Phillips, and A. Eichmann, 2014: Development of two-moment cloud microphysics for liquid and ice within the NASA Goddard Earth Observing System Model (GEOS-5). Geosci. Model Dev., 7, 1733–1766, https://doi.org/10.5194/gmd-7-1733-2014.
Barnes, E. A., and L. M. Polvani, 2015: CMIP5 projections of Arctic amplification, of the North American/North Atlantic circulation, and of their relationship. J. Climate, 28, 5254–5271, https://doi.org/10.1175/JCLI-D-14-00589.1.
Barrett, A. P., J. C. Stroeve, and M. C. Serreze, 2020: Arctic Ocean precipitation from atmospheric reanalyses and comparisons with North Pole drifting station records. J. Geophys. Res. Oceans, 125, e2019JC015415, https://doi.org/10.1029/2019JC015415.
Bechtold, P., M. Köohler, T. Jung, F. Doblas-Reyes, M. Leutbecher, M. J. Rodwell, F. Vitart, and G. Balsamo, 2008: Advances in predicting atmospheric variability with the ECMWF model, 2008: From synoptic to decadal time-scales. Quart. J. Roy. Meteor. Soc., 134, 1337–1351, https://doi.org/10.1002/qj.289.
Bechtold, P., N. Semane, P. Lopez, J.-P. Chaboureau, A. Beljaars, and N. Bormann, 2014: Representing equilibrium and nonequilibrium convection in large-scale models. J. Atmos. Sci., 71, 734–753, https://doi.org/10.1175/JAS-D-13-0163.1.
Bintanja, R., 2018: The impact of Arctic warming on increased rainfall. Sci. Rep., 8, 16001, https://doi.org/10.1038/s41598-018-34450-3.
Bintanja, R., and F. M. Selten, 2014: Future increases in Arctic precipitation linked to local evaporation and sea-ice retreat. Nature, 509, 479–482, https://doi.org/10.1038/nature13259.
Bintanja, R., and O. Andry, 2017: Towards a rain-dominated Arctic. Nat. Climate Change, 7, 263–267, https://doi.org/10.1038/nclimate3240.
Bintanja, R., K. van der Wiel, E. C. Van der Linden, J. Reusen, L. Bogerd, F. Krikken, and F. M. Selten, 2020: Strong future increases in Arctic precipitation variability linked to poleward moisture transport. Sci. Adv., 6, eaax6869, https://doi.org/10.1126/sciadv.aax6869.
Boisvert, L. N., and J. C. Stroeve, 2015: The Arctic is becoming warmer and wetter as revealed by the atmospheric infrared sounder. Geophys. Res. Lett., 42, 4439–4446, https://doi.org/10.1002/2015GL063775.
Boisvert, L. N., D. L. Wu, and C.-L. Shie, 2015: Increasing evaporation amounts seen in the Arctic between 2003 and 2013 from AIRS data. J. Geophys. Res. Atmos., 120, 6865–6881, https://doi.org/10.1002/2015JD023258.
Boisvert, L. N., M. A. Webster, A. A. Petty, T. Markus, D. H. Bromwich, and R. I. Cullather, 2018: Intercomparison of precipitation estimates over the Arctic Ocean and its peripheral seas from reanalyses. J. Climate, 31, 8441–8462, https://doi.org/10.1175/JCLI-D-18-0125.1.
Boisvert, L. N., M. A. Webster, A. A. Petty, T. Markus, D. H. Bromwich, and R. I. Cullather, 2020: Intercomparison of precipitation estimates over the Southern Ocean from atmospheric reanalyses. J. Climate, 33, 10 627–10 651, https://doi.org/10.1175/JCLI-D-20-0044.1.
Boisvert, L. N., M. Grecu, and C.-L. Shie, 2021: Investigating wintertime GPM-IMERG precipitation in the North Atlantic. Geophys. Res. Lett., 48, e2021GL095391, https://doi.org/10.1029/2021GL095391.
Boisvert, L. N., R. C. Boeke, P. C. Taylor, and C. L. Parker, 2022: Constraining Arctic climate projections of wintertime warming with surface turbulent flux observations and representation of surface-atmosphere coupling. Front. Earth Sci., 10, 765304, https://doi.org/10.3389/feart.2022.765304.
Bosilovich, M. G., F. R. Robertson, L. Takacs, A. Molod, and D. Mocko, 2017: Atmospheric water balance and variability in the MERRA-2 reanalysis. J. Climate, 30, 1177–1196, https://doi.org/10.1175/JCLI-D-16-0338.1.
Cabaj, A., P. J. Kushner, C. G. Fletcher, S. Howell, and A. A. Petty, 2020: Constraining reanalysis snowfall over the Arctic Ocean using CloudSat observations. Geophys. Res. Lett., 47, e2019GL086426, https://doi.org/10.1029/2019GL086426.
Cullather, R. I., D. H. Bromwich, and M. C. Serreze, 2000: The atmospheric hydrologic cycle over the Arctic basin from reanalyses. Part I: Comparison with observations and previous studies. J. Climate, 13, 923–937, https://doi.org/10.1175/1520-0442(2000)013<0923:TAHCOT>2.0.CO;2.
Dai, A., 2006: Recent climatology, variability, and trends in global surface humidity. J. Climate, 19, 3589–3606, https://doi.org/10.1175/JCLI3816.1.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Donlon, C. J., M. Martin, J. Stark, J. Roberts-Jones, E. Fiedler, and W. Wimmer, 2012: The Operational Sea Surface Temperature and Sea Ice Analysis (OSTIA) system. Remote Sens. Environ., 116, 140–158, https://doi.org/10.1016/j.rse.2010.10.017.
Dou, T. F., S. F. Pan, R. Bintanja, and C. D. Xiao, 2022: More frequent, intense, and extensive rainfall events in a strongly warming Arctic. Earth’s Future, 10, e2021EF002378, https://doi.org/10.1029/2021EF002378.
Eastwood, S., T. Lavergne, and R. Tonboe, 2014: Algorithm theoretical basis document for the OSI SAF global reprocessed sea ice concentration product. EUMETSAT Network Satellite Application Facilities, 28 pp., https://osisaf-hl.met.no/sites/osisaf-hl/files/baseline_document/osisaf_cdop2_ss2_atbd_sea-ice-conc-reproc_v1p1_0.pdf.
Edel, L., C. Claud, C. Genthon, C. Palerme, N. Wood, T. L’ecuyer, and D. Bromwich, 2020: Arctic snowfall from CloudSat observations and reanalyses. J. Climate, 33, 2093–2109, https://doi.org/10.1175/JCLI-D-19-0105.1.
Forbes, R. M., and A. M. Tompkins, 2011: An improved representation of cloud and precipitation. ECMWF Newsletter, No. 129, ECMWF, Reading, United Kingdom, 13–18, https://www.ecmwf.int/en/elibrary/17431-improved-representation-cloud-and-precipitation.
Forbes, R. M., and M. Ahlgrimm, 2014: On the representation of high-latitude boundary-layer mixed-phase cloud in the ECMWF global model. Mon. Wea. Rev., 142, 3425–3445, https://doi.org/10.1175/MWR-D-13-00325.1.
Forbes, R. M., and A. M. Tompkins, and A. Untch, 2011: A new prognostic bulk microphysics scheme for the IFS. ECMWF Tech. Memo. 649, 30 pp., https://www.ecmwf.int/sites/default/files/elibrary/2011/9441-new-prognostic-bulk-microphysics-scheme-ifs.pdf.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
Harvey, B. J., L. C. Shaffrey, and T. J. Woollings, 2014: Equator-to-pole temperature differences and the extra-tropical storm track responses of the CMIP5 climate models. Climate Dyn., 43, 1171–1182, https://doi.org/10.1007/s00382-013-1883-9.
Haynes, J. M., T. S. L’Ecuyer, G. L. Stephens, S. D. Miller, C. Mitrescu, N. B. Wood, and S. Tanelli, 2009: Rainfall retrieval over the ocean with spaceborne W-band radar. J. Geophys. Res., 114, D00A22, https://doi.org/10.1029/2008JD009973.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hibler, W. D., III, and J. Zhang, 1995: On the effect of sea-ice dynamics on oceanic thermohaline circulation. Ann. Glaciol., 21, 361–368, https://doi.org/10.3189/S0260305500016074.
Hogan, R. J., and A. Battaglia, 2008: Fast lidar and radar multiple-scattering models. Part II: Wide-angle scattering using the time-dependent two-stream approximation. J. Atmos. Sci., 65, 3636–3651, https://doi.org/10.1175/2008JAS2643.1.
Hogan, R. J., M. D. Behera, E. J. O’Connor, and A. J. Illingworth, 2004: Estimate of the global distribution of stratiform supercooled liquid water clouds using the LITE lidar. Geophys. Res. Lett., 31, L05106, https://doi.org/10.1029/2003GL018977.
Holland, M. M., J. Finnis, A. P. Barrett, and M. C. Serreze, 2007: Projected changes in Arctic Ocean freshwater budgets. J. Geophys. Res., 112, G04S55, https://doi.org/10.1029/2006JG000354.
IPCC, 2021: Summary for policymakers. Climate Change 2021: The Physical Science Basis, Cambridge University Press, 3–21, https://doi.org/10.1017/9781009157896.001.
Kulie, M. S., and R. Bennartz, 2009: Utilizing space-borne radars to retrieve dry snowfall. J. Appl. Meteor. Climatol., 48, 2564–2580, https://doi.org/10.1175/2009JAMC2193.1.
Landrum, L., and M. M. Holland, 2020: Extremes become routine in an emerging new Arctic. Nat. Climate Change, 10, 1108–1115, https://doi.org/10.1038/s41558-020-0892-z.
L’Ecuyer, T. S., and G. L. Stephens, 2002: An estimation-based precipitation retrieval algorithm for attenuating radars. J. Appl. Meteor., 41, 272–285, https://doi.org/10.1175/1520-0450(2002)041<0272:AEBPRA>2.0.CO;2.
Lindsay, R., M. Wensnahan, A. Schweiger, and J. Zhang, 2014: Evaluation of seven different atmospheric reanalysis products in the Arctic. J. Climate, 27, 2588–2606, https://doi.org/10.1175/JCLI-D-13-00014.1.
Lique, C., M. M. Holland, Y. B. Dibike, D. M. Lawrence, and J. A. Screen, 2016: Modeling the Arctic freshwater system and its integration in the global system: Lessons learned and future challenges. J. Geophys. Res. Biogeosci., 121, 540–566, https://doi.org/10.1002/2015JG003120.
Manabe, S., and R. J. Stouffer, 1980: Sensitivity of a global climate model to an increase of CO2 concentration in the atmosphere. J. Geophys. Res., 85, 5529–5554, https://doi.org/10.1029/JC085iC10p05529.
Marcovecchio, A., A. Behrangi, X. Dong, B. Xi, and Y. Huang, 2022: Precipitation influence on and response to early and late Arctic sea ice melt onset during the melt season. Int. J. Climatol., 42, 81–96, https://doi.org/10.1002/joc.7233.
McCrystall, M. R., J. Stroeve, M. Serreze, B. C. Forbes, and J. A. Screen, 2021: New climate models reveal faster and larger increases in Arctic precipitation than previously projected. Nat. Commun., 12, 6765, https://doi.org/10.1038/s41467-021-27031-y.
Min, S. K., X. B. Zhang, and F. Zwiers, 2008: Human-induced Arctic moistening. Science, 320, 518–520, https://doi.org/10.1126/science.1153468.
Molod, A., 2012: Constraints on the profiles of total water PDF in AGCMs from AIRS and a high-resolution model. J. Climate, 25, 8341–8352, https://doi.org/10.1175/JCLI-D-11-00412.1.
Molod, A., L. Takacs, M. Suarez, and J. Bacmeister, 2015: Development of the GEOS-5 atmospheric general circulation model: Evolution from MERRA to MERRA2. Geosci. Model Dev., 8, 1339–1356, https://doi.org/10.5194/gmd-8-1339-2015.
Moorthi, S., and M. J. Suarez, 1992: Relaxed Arakawa-Schubert. A parameterization of moist convection for general circulation models. Mon. Wea. Rev., 120, 978–1002, https://doi.org/10.1175/1520-0493(1992)120<0978:RASAPO>2.0.CO;2.
Morrison, H., and J. O. Pinto, 2006: Intercomparison of bulk cloud microphysics schemes in mesoscale simulations of springtime Arctic mixed-phase stratiform clouds. Mon. Wea. Rev., 134, 1880–1900, https://doi.org/10.1175/MWR3154.1.
Parker, C. L., P. A. Mooney, M. A. Webster, and L. N. Boisvert, 2022: The influence of recent and future climate change on spring Arctic cyclones. Nat. Commun., 13, 6514, https://doi.org/10.1038/s41467-022-34126-7.
Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, and W. Wang, 2002: An improved in situ and satellite SST analysis for climate. J. Climate, 15, 1609–1625, https://doi.org/10.1175/1520-0442(2002)015<1609:AIISAS>2.0.CO;2.
Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 5473–5496, https://doi.org/10.1175/2007JCLI1824.1.
Rienecker, M. M., and Coauthors, 2008: The GEOS-5 data assimilation system—Documentation of versions 5.0.1, 5.1.0, and 5.2.0. NASA Tech. Rep. NASA/TM-2008-104606, Vol. 27, 118 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Rienecker369.pdf.
Serreze, M. C., and C. M. Hurst, 2000: Representation of mean Arctic precipitation from NCEP–NCAR and ERA reanalyses. J. Climate, 13, 182–201, https://doi.org/10.1175/1520-0442(2000)013<0182:ROMAPF>2.0.CO;2.
Serreze, M. C., A. P. Barrett, and F. Lo, 2005: Northern high-latitude precipitation as depicted by atmospheric reanalyses and satellite retrievals. Mon. Wea. Rev., 133, 3407–3430, https://doi.org/10.1175/MWR3047.1.
Serreze, M. C., A. P. Barrett, J. C. Stroeve, D. N. Kindig, and M. M. Holland, 2009: The emergence of surface-based Arctic amplification. Cryosphere, 3, 11–19, https://doi.org/10.5194/tc-3-11-2009.
Song, Y., A. Behrangi, and E. Blanchard-Wrigglesworth, 2020: Assessment of satellite and reanalysis cold season snowfall estimates over Arctic Sea ice. Geophys. Res. Lett., 47, e2020GL088970, https://doi.org/10.1029/2020GL088970.
Stammerjohn, S., and T. Maksym, 2017: Gaining (and losing) Antarctic sea ice: Variability, trends and mechanisms. Sea Ice, 3rd ed. D. N. Thomas, Ed., John Wiley and Sons, 261–289.
Stephens, G. L., and Coauthors, 2008: CloudSat mission: Performance and early science after the first year of operation. J. Geophys. Res., 113, D00A18, https://doi.org/10.1029/2008JD009982.
Stephens, G. L., and Coauthors, 2010: Dreary state of precipitation in global models. J. Geophys. Res., 115, D24211, https://doi.org/10.1029/2010JD014532.
Sturm, M., D. K. Perovich, and J. Holmgren, 2002: Thermal conductivity and heat transfer through the snow on the ice of the Beaufort Sea. J. Geophys. Res., 107, 8043, https://doi.org/10.1029/2000JC000409.
Sundqvist, H., E. Berge, and J. E. Kristjánsson, 1989: Condensation and cloud parameterization studies with a mesoscale numerical weather prediction model. Mon. Wea. Rev., 117, 1641–1657, https://doi.org/10.1175/1520-0493(1989)117<1641:CACPSW>2.0.CO;2.
Takacs, L. L., M. J. Suárez, and R. Todling, 2016: Maintaining atmospheric mass and water balance in reanalyses. Quart. J. Roy. Meteor. Soc., 142, 1565–1573, https://doi.org/10.1002/qj.2763.
Tanelli, S., S. L. Durden, E. Im, K. S. Pak, D. G. Reinke, P. Partain, J. M. Haynes, and R. T. Marchand, 2008: CloudSat’s cloud profiling radar after two years in orbit: Performance, calibration, and processing. IEEE Trans. Geosci. Remote Sens., 46, 3560–3573, https://doi.org/10.1109/TGRS.2008.2002030.
Taylor, P. C., B. M. Hegyi, R. C. Boeke, and L. N. Boisvert, 2018: On the increasing importance of air–sea energy exchanges in a thawing Arctic: A review. Atmosphere, 9, 41, https://doi.org/10.3390/atmos9020041.
Taylor, P. C., and Coauthors, 2022: Process drivers, inter-model spread, and the path forward: A review of amplified Arctic warming. Front. Earth Sci., 9, 758361, https://doi.org/10.3389/feart.2021.758361.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800, https://doi.org/10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev., 121, 3040–3061, https://doi.org/10.1175/1520-0493(1993)121<3040:ROCILS>2.0.CO;2.
Titchner, H. A., and N. A. Rayner, 2014: The met office Hadley Centre sea ice and sea surface temperature data set, version 2: 1. Sea ice concentrations. J. Geophys. Res. Atmos., 119, 2864–2889, https://doi.org/10.1002/2013JD020316.
Trenberth, K. E., 2011: Changes in precipitation with climate change. Climate Res., 47, 123–138, https://doi.org/10.3354/cr00953.
Vihma, T., 2014: Effects of Arctic sea ice decline on weather and climate: A review. Surv. Geophys., 35, 1175–1214, https://doi.org/10.1007/s10712-014-9284-0.
Vihma, T., M. Mech, A. Rinke, D. Zhang, M. Lauer, A. Radovan, I. Gorodetskaya, and S. Crewell, 2016: The atmospheric role in the Arctic water cycle: A review on processes, past and future changes, and their impacts. J. Geophys. Res. Biogeosci., 121, 586–620, https://doi.org/10.1002/2015JG003132.
von Lerber, A., and Coauthors, 2022: Evaluating seasonal and regional distribution of snowfall in regional climate model simulations in the Arctic. Atmos. Chem. Phys., 22, 7287–7317, https://doi.org/10.5194/acp-22-7287-2022.
Walsh, J. E., V. Kattsov, D. Portis, and V. Meleshko, 1998: Arctic precipitation and evaporation: Model results and observational estimates. J. Climate, 11, 72–87, https://doi.org/10.1175/1520-0442(1998)011<0072:APAEMR>2.0.CO;2.
Walsh, J. E., T. J. Ballinger, E. S. Euskirchen, E. Hanna, J. Mård, J. E. Overland, H. Tangen, and T. Vihma, 2020: Extreme weather and climate events in northern areas: A review. Earth-Sci. Rev., 209, 103324, https://doi.org/10.1016/j.earscirev.2020.103324.
Webster, M., and Coauthors, 2018: Snow in the changing sea-ice systems. Nat. Climate Change, 8, 946–953, https://doi.org/10.1038/s41558-018-0286-7.
Webster, M., A. K. DuVivier, M. M. Holland, and D. A. Bailey, 2021: Snow on Arctic sea ice in a warming climate as simulated in CESM. J. Geophys. Res. Oceans, 126, e2020JC016308, https://doi.org/10.1029/2020JC016308.
Webster, P. J., 1994: The role of hydrological processes in ocean-atmosphere interactions. Rev. Geophys., 32, 427–476, https://doi.org/10.1029/94RG01873.
Wood, N. B., T. S. L’Ecuyer, D. G. Vane, G. L. Stephens, and P. Partian, 2013a: Level 2C snow-profile process description and interface control document, algorithm version P_R04. Accessed 3 December 2018, https://www.cloudsat.cira.colostate.edu/cloudsat-static/info/dl/2c-snow-profile/2C-SNOW-PROFILE_PDICD.P_R04.20130210.pdf.
Wood, N. B., T. S. L’Ecuyer, F. L. Bliven, and G. L. Stephens, 2013b: Characterization of video disdrometer uncertainties and impacts on estimates of snowfall rate and radar reflectivity. Atmos. Meas. Tech., 6, 3635–3648, https://doi.org/10.5194/amt-6-3635-2013.
Wood, N. B., T. S. L’Ecuyer, A. J. Heymsfield, G. L. Stephens, D. R. Hudak, and P. Rodrigues, 2014: Estimating snow microphysical properties using collocated multisensor observations. J. Geophys. Res. Atmos., 119, 8941–8961, https://doi.org/10.1002/2013JD021303.