1. Introduction
Air temperature inversions—increasing temperatures with elevation—are present throughout the Arctic, covering a wide range of spatial and temporal domains. While these inversions can exist over a wide range of landscapes, and be the result of numerous processes and interactions, the inversion climatology literature mostly focuses on studies of single valley or basin locations (e.g., Putnins 1970; Kahl 1990; Kahl et al. 1992; Serreze et al. 1992; Kadygrov et al. 1999; Hansen et al. 2008). A realistic description of the spatiotemporal air temperature variation over complex topography influenced by air temperature inversions is essential for snow and ice melt calculations, glacier mass-balance estimates, river breakup simulations, ecological studies, water resource predictions, and for dispersion of pollutants in mountain and basin areas (e.g., Whiteman 1982; Chen et al. 1999; Singh 1999; Whiteman et al. 1999; Archer 2004; Lundquist and Cayan 2007; Kerminen et al. 2007; Barry 2008). Inversions occur when the coldest (and densest) air settles to the lowest topographic level and therefore temperatures increase with increasing elevation above the earth’s surface (Anquetin et al. 1998). In contrast, periods without inversions may typically experience the moist adiabatic temperature decrease with an altitude of −0.65°C (100 m)−1 (Marinec and Rango 1986; Oke 1987).
Air temperature inversions are present on local-to-regional scales, for example, during the passage of cold fronts, by radiative cooling of the surface, from advection, and from the invasion of cooler and increasingly moist onshore breezes (e.g., Streten et al. 1974; Oke 1987; Milionis and Davies 2008). Cold waters cause higher frequencies of inversion in high-latitude coastal regions, based on the local-to-regional wind systems (e.g., sea breezes) as a result of thermal differences between land and ocean (Hosler 1961; Milionis and Davies 2008). The height of the inversion base is attributed to the combined effects of topography (e.g., terrain-induced air motion), synoptic conditions (e.g., the thermal and dynamic structure of the atmosphere), and sea ice dynamics (e.g., Greenland 1979; Riordan et al. 1986; Hanna and Strimaitis 1990; Kahl 1990; Barry 2008; Milionis and Davies 2008).
Around Ammassalik Island in SE Greenland, summer inversions are common because of the effect of sea breezes. During winter at Ammassalik Island, both sea and land surfaces are typically visibly homogeneous with a covering of ice and snow, and snow, respectively: high albedo and low amounts of absorbed solar radiation therefore provide small differences in energy partitioning between the marine and terrestrial surfaces (Hansen et al. 2008). After terrestrial snowmelt in the spring, the land surface warms up, giving rise to a temperature gradient between the land and the still frozen sea. During daytime, cold and moist sea breezes affect the air temperature lapse rates in the coastal areas (e.g., Barry 2008; Mernild et al. 2008a). As spring and summer progress, the sea ice melts and the temperature differences between sea and land decrease, but the ocean temperatures still remain relatively low compared to the land. Because of the nearly continuous and relatively large solar radiation input during midsummer, the sea breeze can still exist, but possibly in a slightly weaker version. Sea breezes have been found all over the coastal Arctic [e.g., along the Alaskan Beaufort and Arctic Sea coast (Kozo 1982a,b; Kahl 1990), the Canadian Arctic Sea coast (Weick and Rouse 1991; Kahl et al. 1992), Disko Island in west Greenland (Hansen et al. 2005), and Zackenberg in NE Greenland (Hansen et al. 2008)].
Air temperature is an important climatological parameter—a key driver for melt in high-latitude polar regions—and therefore also the most studied parameter. Still, along the coast and in Greenland’s interior, because of the generally rough terrain, logistic constraints, and the remote locations, extensive air temperature observations have typically not been possible. Use of the few existing meteorological stations that measure air temperature and temperature of the atmosphere by radiosonde observations leads us to conclude that we have a limited knowledge about the distribution of air temperature and the effect of air temperature inversions on snow and ice cover throughout Greenland. The limited measurement of such a key climate system component is likely a serious impediment to hydrological research efforts. Thus, there is a clear need to explore issues associated with data sparseness and modeling capabilities.
The goal of this study is to understand the influence of temperature inversions on snow and glacier mass balance over the SW part of Ammassalik Island in SE Greenland, and to quantify their effects on snow and ice cover evolution. The aim of this study is to apply a well-tested approach and state-of-the-art modeling system, SnowModel (Liston and Elder 2006a; Mernild et al. 2006b; Liston et al. 2007), including its quasi-physically-based meteorological distribution model, MicroMet (Liston and Elder 2006b). We performed model simulations for an 8-yr period (1998/99–2005/06) with the following objectives: 1) to assess MicroMet–SnowModel meteorological driving data against independent observations; 2) to simulate the spatial air temperature distribution during periods without and with air temperature inversions; 3) to model the effect of inversions on winter snow accumulation, and summer ablation related to snowmelt and glacier-ice melt; and 4) to simulate the effect of air temperature inversions on the surface winter (defined herein to be September–May), summer (June–August), and annual mass balance for marginal glaciers near Greenland’s east coast.
2. Study area
a. Physical setting
Ammassalik Island (678 km2) is located in SE Greenland (65°N latitude; 37°W longitude) approximately 50 km east of the eastern margin of the Greenland Ice Sheet (GrIS), separated from the mainland to the west by the 10–15-km-wide Sermilik Fjord, to the north by the Ikaasartivaq Fjord, to the east by the Ammassalik Fjord, and to the south by the North Atlantic Ocean (Fig. 1a). The simulation domain (485 km2) covers the SW part of the island—the area of potential interest for water resources for the town Tasiilaq. The glacier and lake areas are 61.9 km2 (∼13% of the simulation domain) and 24.6 km2 (∼5%), respectively (Fig. 1b). The observed average winter, summer, and annual mass balances for 1998/99–2005/06 for the Mittivakkat Glacier observation area (outlined in Fig. 1c; 17.6 km2) are, respectively, 1100 ± 150, −2010 ± 410, and −770 ± 610 mm water equivalent (w.eq.). Since 1898, 89 out of 105 mass-balance years show a negative Mittivakkat Glacier annual mass balance (Mernild et al. 2008b). Strong alpine relief characterizes the SW part of Ammassalik Island, with elevations ranging to above 1093 m MSL at the highest peaks (Fig. 1a). Proglacier valleys west on the island have an E–W orientation toward Sermilik Fjord, whereas east of Mittivakkat Glacier, valleys have a N–S orientation toward Ammassalik Fjord.
b. Meteorological stations and climate
There are three meteorological stations within the simulation domain: Station Nunatak (65°42′N; 37°49′W; 515 m MSL) is representative of glacier conditions, located on a small nunatak (∼5 m from the glacier in the dominant wind directions) close to the equilibrium line altitude (ELA) on the NW part of Mittivakkat Glacier; Station Coast (65°41′N; 37°55′W; 25 m MSL) is representative of coastal/valley conditions located on a rock hill near Sermilik Fjord; and Station Tasiilaq (65°36′N; 37°38′W; 44 m MSL) is a standard synoptic World Meteorological Organization (WMO) meteorological station representative of coastal–fjord areas located on a hillside in the upper city limit of Tasiilaq (Fig. 1a) [technical specifications of Station Nunatak and Station Coast and the sensors can be found in Mernild et al. (2008a), and of Station Tasiilaq by contacting the Danish Meteorological Institute]. Ammassalik Island is considered to be low Arctic according to Born and Böcher (2001), and represents a relatively humid area of Greenland (Mernild et al. 2008a). Based on observed data from these stations, the mean annual air temperature (MAAT) (1998–2006) is −2.1°, −0.8°, and −0.1°C for Station Nunatak, Station Coast, and Station Tassilaq, respectively (Mernild et al. 2008a). The maximum monthly air temperature averaged across the three stations is 6.7°C in July and the minimum is −7.3°C in February. Mean annual relative humidity is 82%. The total annual precipitation (TAP) is 1851 mm w.eq. yr−1 at Station Nunatak, 1428 mm w.eq. yr−1 at Station Coast, and 1254 mm w.eq. yr−1 at Station Tasiilaq (Mernild et al. 2008a), indicating a positive orographic effect between the coastal stations and Station Nunatak. The mean annual wind speed is 3.8, 4.0, and 2.3 m s−1 for Station Nunatak, Station Coast, and Station Tasiilaq, respectively (Mernild et al. 2008a), while the average summer (June–August) and winter wind speed (September–May) across the three stations are 2.1 and 3.8 m s−1, respectively.
c. Air temperature inversion analysis
Air temperature inversions are common in the coastal areas around the Ammassalik Island, including the lower part of the Mittivakkat Glacier; these inversions affect air temperature lapse rates in the area and the associated snow and ice melt processes.
Twice-daily radiosonde observations recorded at 0000 and 1200 UTC time at Station Tasiilaq (1996–2005) were used to estimate the frequency of air temperature inversions (Table 1). For Station Tasiilaq, air temperature and elevation above sea level were recorded at different atmospheric pressure levels: at the terrain surface (52 m MSL), 925 hPa (average elevation and standard deviation: 631 ± 108 m MSL), 850 hPa (1329 ± 114 m MSL), and 700 hPa (2871 ± 141 m MSL). Because the 925-hPa level has an average elevation of 631 m MSL, many of the shallow coastal area inversions that actually occur are not captured by the radiosonde data. During summer (June–August), inversions were present an average of 30% of the time, and 20% of the time during the rest of the year (winter).
To improve our understanding of air temperature inversions in the Ammassalik Island area, observed climate data from Station Coast (2 m), Station Nunatak (2 m), and Station Tasiilaq (2 m) were analyzed during periods where Station Tasiilaq radiosonde observations (1995–2006) existed. No statistically significant difference occurred in observed surface relative humidity, air temperature, and wind speed at the three meteorological stations between the periods without or with radiosonde-observed inversion layers. This is likely because the 2-m climate on the Ammassalik Island is influenced by a combination of local-to-regional meteorological conditions: 1) local meteorological conditions based on variations in topography, the sea-breeze effect, and the presence of marginal glaciers producing katabatic winds, for example, on the Mittivakkat Glacier; and 2) overall meteorological conditions from the GrIS and the surrounding fjords/ocean. Even the difference in radiosonde wind speed data between the periods without or with radiosonde-observed inversion layers was insignificant (and around 1–2 m s−1) for the lower part of the atmosphere (below 925 hPa) (Table 2).
In addition, field observations were performed June–August 2005 and 2006, near Station Coast, to learn more about the inversions in this area (Fig. 2). Vertical hand-carried temperature observations near Station Coast (June–August 2005 and 2006) (Figs. 1a and 2) were used to estimate inversion frequency and elevation. These observations found a summer average inversion presence 84% of the time (Table 1), or 2.8 times the frequency captured by the radiosonde data. These observations also showed that the tops of the air temperature inversions were located at approximately 300 m MSL (Fig. 2). Clearly the radiosonde data are only capturing the deepest inversions, and there are many inversions that exist in the natural system that are not represented by the radiosonde observations. While we do not have the observations to quantify it, we expect that the winter radiosonde records also frequently miss the presence of low-level inversions.
In Fig. 2, the vertical temperature profiles show the summer variations and average distribution with elevation. The inversion frequency found near Station Coast is almost identical with values found other places in the Arctic. By comparison, frequencies of 85%–99% were found for Alaskan and Canadian Arctic stations (Kahl 1990; Kahl et al. 1992; Kadygrov et al. 1999), and 91% for the Arctic Ocean based on nearly 30 000 analyzed radiosonde temperature observations (Kahl et al. 1996). In winter, the occurrence of inversions at Station Tasiilaq was lower compared to summer conditions, probably because 1) the sea and land surfaces have much the same temperature since they are both covered more or less continuously in ice and snow and local temperature differences in the heating of sea and land is minimal; and 2) wind speeds are slightly higher in winter (between 1.3 and 2.9 m s−1) (Table 2). We also know that radiational cooling is greatest during winter, and this likely increase the chances of inversions and possibly increases the occurrence of relatively shallow inversions that are not detectable by the radiosonde observations.
3. SnowModel
a. SnowModel description
SnowModel (Liston and Elder 2006a) is a spatially distributed snowpack evolution modeling system specifically designed to be applicable over the wide range of snow landscapes and climates found around the world. It is made up of four submodels: MicroMet (a quasi-physically-based meteorological distribution model) defines the meteorological forcing conditions (Liston and Elder 2006b); EnBal calculates the surface energy exchanges, including melt (Liston 1995; Liston et al. 1999); SnowPack simulates heat and mass transfer processes, and snow depth and water equivalent evolution (Liston and Hall 1995); and SnowTran-3D is a blowing-snow model that accounts for snow redistribution by wind (Liston and Sturm 1998, 2002; Liston et al. 2007). SnowModel simulates snow-related physical processes at spatial scales ranging from 5 m to global and temporal scales ranging from 10 min to a whole season. SnowModel was originally developed for glacier-free landscapes. For glacier-surface mass-balance studies, SnowModel was modified to simulate glacier-ice melt after winter snow accumulation had ablated (Mernild et al. 2006b, 2007).
To solve the system of equations, SnowModel requires spatially distributed fields of topography and land cover types, and temporally varying meteorological data (air temperature, relative humidity, wind speed, wind direction, and precipitation) obtained from the two meteorological stations: Station Nunatak and Station Coast located within the simulation domain (Fig. 1a). Model simulations used a 100-m grid increment. The simulations were performed using a 1-day time step and spanned the 8-yr period from 1 September 1998 through 31 August 2006. Snow and ice melt and blowing snow are threshold processes and may not be accurately represented by a 1-day time step. Unfortunately, computational resources did not allow using a smaller time increment for the entire simulation domain. Therefore, daily simulated melt and blowing-snow processes were tested against hourly simulated values from the Mittivakkat Glacier subdomain (1998/99–2005/06) and remain significant ( p < 0.01, where p is the level of significance), with an average difference of 2%, 3%, and 8% for the winter, summer, and annual mass balance, respectively (Table 3).
Topographic data over the SW part of the Ammassalik Island were obtained from a digital elevation model (DEM) (100-m gridcell increment). Each grid cell was assigned a land cover type and classified as bedrock with a snow-holding depth (the snow depth that must be exceeded before snow can be transported by wind) of 0.50 m (Mernild et al. 2006b, 2007), as lake ice with a depth of 0.01 m, or as glacier with a snow-holding depth of 0.01 m (Fig. 1b; Table 4) (Liston and Sturm 2002; Mernild et al. 2006b). Albedo was assumed to be 0.8 for snow and 0.4 for ice. Realistically, snow and ice albedo change with time and surface characteristics (e.g., Pomeroy and Brun 2001). Model parameter values used in the simulations are provided in Table 4.
b. SnowModel modifications for air temperature inversions
c. SnowModel testing
Prior to performing SnowModel sensitivity simulations using the temperature inversion routines, test simulations were performed over our domain of interest using the original MicroMet temperature lapse-rate formulation. To assess the performance of the SnowModel–MicroMet distributed meteorological data, simulated meteorological data (without the inversion routines) were tested against independent observations not used in MicroMet. Station Tasiilaq (Fig. 1a) was used for comparisons spanning September 1998–August 2006. The validation station was located approximately 15 km from the stations used in MicroMet to drive SnowModel. In strong alpine terrain/mountainous areas air temperature variability is complicated because it encompasses such a broad range of temporal and spatial scales (Lundquist and Cayan 2007). However, validations of MicroMet-simulated meteorological data indicate substantial correlation with independent observed meteorological data from Station Tasiilaq (Fig. 3). MicroMet-generated air temperature values accounted for 87% of the variance in the observed 1998–2006 daily averaged dataset, indicating a significant ( p < 0.01) justification of the MicroMet temperature routines. Wind speed, precipitation, and relative humidity have less strong correlations, but remain significant ( p < 0.01) and respectable (around 50% variance) representations of Station Tasiilaq meteorological processes. This validation is limited because it employs only one independent station located near sea level, however, based on its previous applications and testing in areas having strong topographic relief (e.g., Liston and Elder 2006a; Mernild et al. 2008d), we assume that MicroMet satisfactorily represents the Ammassalik Island meteorological conditions.
To assess the performance of SnowModel accumulation and ablation routines, distributed observed point snow and SWE depths, time lapse photographs, and satellite images were used for validation with satisfactory results: a difference of a maximum 7% occurred between observed snow and SWE values and modeled values [for further information see Mernild et al. (2006b,a, 2007, 2008c–e, 2009)]. Further, winter, summer, and annual mass-balance observations from the Mittivakkat Glacier observation area (17.6 km2; Fig. 1c) were used to assess the model performance (without the inversion routines) for the end-of-winter accumulation (31 May) and the end-of-summer ablation (31 August) (Table 3). A split-sample test (e.g., Klemes 1985, 1986; Refsgaard and Knudsen 1996; Refsgaard 2000; Refsgaard and Henriksen 2004) was applied for calibration (1998/99–2001/02) and validation (2002/03–2005/06) of the simulated end-of-winter and summer mass balance. For the validation period a high degree of similarity between average modeled and observed winter, summer, and annual mass balance occurred (Table 3). For the winter, summer, and annual mass balance, an average difference of 20 (∼2%), 30 (∼1%), and 80 (∼10%) mm w.eq. yr−1 occurred, respectively. These differences are within the uncertainties of the observed winter, summer, and annual mass-balance values; still, the annual variation between the modeled and observed mass balance could be up to 420 mm w.eq. yr−1 (2002/03). For further information about the calibration and validation procedures see Mernild et al. (2006b, 2008e).
d. SnowModel simulations
To perform SnowModel simulations using the new temperature inversion representation, a methodology to define inversion presence, absence, height, and strength needed to be defined throughout the 8-yr simulation period. Unfortunately, our high-resolution inversion observations only exist during summer. To extend our summer observations into winter, and define a representative winter inversion frequency, we assumed, for the reasons given in section 2c, that inversions occurred during winter at the same frequency as they did during summer (84%). In addition, our radiosonde analysis and other published literature suggest that inversions are less likely to occur during strong winds and high-precipitation rates. High-precipitation rates are typically associated with higher wind speeds, humidity increases, and other factors that reduce the likelihood of an inversion (e.g., Oke 1987; Stull 1988). We also know it is possible to have precipitation during an inversion. In the model simulations that follow, we assumed inversions would be present during days with no precipitation, and also during days when the precipitation was < 10 mm w.eq. day−1 and/or when the wind speed was < 8 m s−1. These thresholds were defined by analyzing our Station Coast and Station Nunatak atmospheric forcing datasets during the 8-yr simulation period (Table 2). During 16% of the time no-inversion days occurred.
Because the available observations were not detailed enough to provide a full understanding of the inversion height and strength within and around Ammassalik Island, a sensitivity analysis was conducted on the Mittivakkat Glacier to see the influence of changing the inversion layer level in 100-m increments, from 100 through 500 m MSL (Table 5). The effect—per 100 m increase in elevation level—on the Mittivakkat Glacier winter, summer, and annual mass-balance averaged values was −50 mm w.eq. (100 m)−1 (5%), 88 mm w.eq. (100 m)−1 (5%), and 38 mm w.eq. (100 m)−1 (5%), respectively. Because of the small influence on the accumulation and ablation processes of the choice of the elevation level, we assume that the choice of the 300 m MSL level is appropriate given what we do know about inversions in this area. However, in the natural system we do expect variations in inversion level to occur under the influence of variations in topography and local climate (e.g., wind speed), and distance from the ocean and marginal glaciers.
Therefore, to determine the influence of temperature inversions on winter snow accumulation and summer ablation, and to simulate their effects on the surface winter, summer, and annual mass balance for the glaciers near Greenland’s east coast, two simulations were performed: the first was conducted without accounting for temperature inversions (Analysis 1), and the second (Analysis 2) assumed temperature inversions occurred up to a height of 300 m MSL and were assumed to be present 84% of the time during the simulation period. The no-inversion cases were defined by days with precipitation rates >10 mm w.eq. day−1 and/or winds >8 m s−1.
Monthly average air temperature lapse rates for Analysis 1 were estimated and used based on air temperature observations (2 m) from Station Coast (25 m MSL) and (2 m) from Station Nunatak [515 m MSL; see Mernild et al. (2006b) for further information]. For Analysis 1, a minimum monthly lapse rate of −0.51°C (100 m)−1 occurred for November and February, and a maximum monthly lapse rate of 0.33°C (100 m)−1 in June and July; these temperature increases with elevation are governed by summer sea breezes in daytime coming predominately from the S and SW (Mernild et al. 2006b, 2008a). Air temperature data from Station Coast and Station Nunatak indicated positive lapse rates for the ablation period (a temperature increase of approximately 1.5°C for the higher station). In reality, for Analysis 1, it seems unlikely that 2-m air temperatures would increase with elevation over the glacier itself (Mernild et al. 2006b).
For Analysis 2, lapse rates up to the fixed inversion layer (300 m MSL) and the 300 m MSL air temperature were calculated based on Station Coast observed air temperatures and Station Tasiilaq radiosonde 925-hPa (average elevation 631 m MSL) recorded air temperatures. Above the fixed inversion layer (300 m MSL) lapse rates were estimated based on the 300 m MSL inversion layer calculated air temperature and the Station Nunatak observed air temperature (see Fig. 4 for a schematic illustration of the Analysis 1 and 2 air temperature lapse rate routines and Table 6 for average monthly lapse rates). For January, February, October, and November the lapse rates above the fixed inversion layer are steeper than the adiabatic lapse rate of −0.98°C (100 m)−1 (Table 6). Realistically, this would represent unstable conditions that could not be maintained in the atmosphere. At present, no temperature observations are available at higher catchment elevations for validation of the estimated upper lapse rates. The model simulates up to ∼2.9°C colder conditions at peaks. Therefore, a sensitivity analysis was conducted illustrating <1% difference in glacier mass balance whether the Table 6 upper lapse rates or the adiabatic lapse rates were used for the months January, February, October, and November. Further, the use of different sites for the radiosonde and meteorological observations might also only create minor uncertainties.
4. Results and discussion
Figure 5a illustrates the spatial modeled daily mean air temperature distribution for the coldest day, 21 February 2002, for Analysis 1 of −20.3°C (without inversion) and 2 of −21.1°C, showing a spatial variation in air temperature of 6.1°C within the simulation domain for Analysis 1, ranging from −18.2° to −24.3°C. For Analysis 2 the variation was greater, 8.8°C, varying from −17.3° to −26.1°C. In the low-lying areas (e.g., below 453 m MSL for February), the Analysis 1 simulated air temperature was relatively low in comparison with the Analysis 2 simulated temperature, and vice versa. The level where simulated temperature from Analysis 1 and 2 is equal (e.g., at the 453 m MSL level for February) is called the “same temperature level” (STL). The temperature variations and the STL are clearly illustrated on the longitudinal temperature profiles in Fig. 5a, where the profiles cross each other. A similar spatial trend occurred for the warmest day, 13 July 2005, for Analysis 1 (19.8°C) and Analysis 2 (18.4°C) (Fig. 5b). For July, the STL was situated at 494 m MSL, almost equal to the February level. The STL was almost stable at the same altitude for winter and summer, but during breakup (April and May) and freeze-up (September and October) the STL and the temperature distribution appears to be more complex because of, for example, variations in air temperature lapse rates influenced by the presence of periodic snow cover at the meteorological stations and sea ice dynamics at the Sermilik Fjord near Station Coast (see Fig. 4b for April and October). On average, for the accumulation period (September–May), the spatially distributed simulation area air temperature was 0.5°C lower for Analysis 1 than Analysis 2. For the ablation period (June–August), the Analysis 1 modeled simulation area temperature averaged 0.6°C higher than Analysis 2 simulated values.
The effect of Analysis 1 (routines without inversion) and 2 (with inversion) distributed air temperature is shown in Fig. 6 for three randomly chosen snow cover depletion areas at the Mittivakkat Glacier: an area representative of the meteorological coastal–valley conditions that spans the elevation from 4 to 200 m MSL, an area representative of glacier conditions (from 489 to 615 m MSL), and an area representative of the mountain-peak conditions (from 679 to 941 m MSL). The snow cover extent is a product of both snow accumulation and ablation processes (phase change processes) of evaporation–sublimation and melting, which are strongly influenced by the air temperature distribution. Within SnowModel, SnowTran-3D simulates spatial blowing-snow deposition patterns in response to erosion and deposition, and EnBal calculates energy flux available for snowmelt. In Fig. 6, the 2003 (the year with the lowest ablation) Analysis 1 modeled valley snowmelt (ablation) started on day of year (DOY) 151 (31 May)—12 days later than the Analysis 2 estimated start of snowmelt. The 12-days earlier snowmelt modeled by Analysis 2 captured the effect of representing inversions in the simulations. Further, for Analysis 1, 25% of the snow cover extent was melted away at DOY 175, 50% at DOY 181, and 75% at DOY 187. For Analysis 2, this occurred on DOY 169, 176, and 184, respectively, indicating a 3–6-day faster snowmelt for the valley (low-lying areas) under conditions of air temperature inversions. For 2005 (the year with the greatest ablation) the trend was similar. However, the Analysis 2 snowmelt occurred 1–3 days before, indicating a reduced effect of the inversions on melting during years with high ablation. For the glacier depletion area the trend is opposite: Analysis 1 snowmelt occurred earlier for both 2003 and 2005. For 2003 (Analysis 1), 25% of the snow cover extent was melted away at DOY 155, 50% at DOY 162, and 75% at DOY 207. For Analysis 2, the DOY are 162, 173, and 225, respectively. On average, throughout the ablation period, an approximately 1–2-week later Analysis 2 snowmelt occurred. For 2005 the trend was similar, however a later snowmelt extent of 1–4 days is indicated. For the peak depletion area, the melt trend equals the one from the glacier area; however, melt occurs later during the ablation season because of the higher elevations. Analysis 1 snowmelt occurred earlier for both 2003 and 2005. For 2003 (Analysis 1), 25% of the snow cover extent was melted away at DOY 191, 50% at DOY 228, and 75% at DOY 232. For Analysis 2, only 25% of melted snow cover extent was reached (DOY 220) within the period from May to the end of August. On average, a 3–5-week later and slower snowmelt occurred for Analysis 2. For 2005, the peak area trend was similar; however a later and slower Analysis 2 snowmelt extent of between 2 and 4 days occurred. In general, for both Analysis 1 and 2, the start date of continuous snowmelt was delayed with increasing elevation. However, for the glacier depletion area, 25% of the snow cover was melted away approximately 2–3 weeks before the low-lying coastal–valley area. This is because the coastal area is influenced by cold sea breezes during the ablation period. Furthermore, in the early ablation period (June), runoff (including high spring flow rates) is mainly controlled by snowmelt whereas later in the season (July and August) when the snow cover is largely gone, runoff is dominated by perennial snow patches, rain events, and glacier-ice melt from marginal glaciers like the Mittivakkat Glacier. On average (1999–2004) the change (loss) in, for example, the Mittivakkat Glacier snow and ice storage explains between 30% and 60% of the runoff (Mernild 2006).
The influence of distributed air temperature representation on average SWE depth, for all three depletion cover areas from May through August for 2003 and 2005, is further illustrated in Fig. 6. The difference in average SWE depth varied up to 53 mm w.eq. between Analysis 1 and 2. However, for the 2003 peak area, the difference was 294 mm w.eq. in terms of average SWE depth. In the coastal–valley area, Analysis 2 modeled SWE depth was less than Analysis 1, and vice versa for the high elevated areas (the glacier and the peak areas). This is due to the temporal and spatial air temperature distributions shown in Figs. 5a and 5b.
The spatial variations in modeled snow cover extent, and whether snow cover is present or absent on Ammassalik Island, is illustrated as an example from 15 May through 15 June 2005 (the year with the greatest ablation) (Fig. 7). Modeled snow cover extent indicates that snow is present on glaciers, mainly on leeside south-facing slopes, and mostly in the valleys, because of the blowing-snow redistribution since the majority of snow-transporting winds are from the NE, N, and SE. For 15 May, Analysis 1 indicates a snow cover extent of 84% (406 km2), and an approximately 7% (∼35 km2) spatial discrepancy between Analysis 1 and 2 (Fig. 7), a discrepancy mainly pronounced in the northwest-, north-, and northeast-facing slopes. This is probably due to the shadow effect of the mountains, and on south-facing slopes according to snow erosion, accumulation, and ablation. For 1 and 15 June, the snow cover extent was 7% (34 km2) and <1% (4 km2), and the spatial discrepancy 8% (41 km2) and 3% (11 km2), respectively. Figure 7 illustrates that the spatial snow cover extent for the entire SW part of Ammassalik Island is nearly identical with the variations in snow cover extent illustrated at the three randomly chosen depletion curve examples in Fig. 6.
During blowing-snow events, sublimation of wind transported snow can play an important role in the high-latitude hydrological cycle. During the investigation period 1998/99–2005/06, modeled annual sublimation for the Mittivakkat Glacier averaged 10% (approximately 135 mm w.eq.) of the solid precipitation inputs for Analysis 1, and 11% (approximately 147 mm w.eq.) for Analysis 2. For the entire simulation domain it is approximately 11% of the solid precipitation for both Analysis 1 and 2. The sublimation losses are low at Ammassalik Island relative to many previous studies in Arctic North America and Greenland (e.g., Pomeroy and Gray 1995; Pomeroy et al. 1997; Liston and Sturm 1998; Essery et al. 1999; Pomeroy and Essery 1999; Liston and Sturm 2004), where approximately 5%–50% of the annual solid precipitation was returned to the atmosphere by sublimation. Blowing snow sublimation rates are mainly dependent upon air temperature, humidity deficit, wind speed, and particle size distribution (e.g., Schmidt 1972, 1982; Tabler 1975; Pomeroy and Gray 1995; Liston and Sturm 2002). In our coastal domain, high wind speeds are generally coincident with high relative humidity, and therefore, sublimation has played a lesser role in the snow and glacier mass balance budget, whether it is modeled by Analysis 1 or 2.
In Table 7 the average winter, summer, and annual mass balances for the glaciers in the SW part of the Ammassalik are shown for both Analysis 1 and 2 (1998/99–2005/06). The glaciers are located above the inversion layer, mainly from 300 to 800 m MSL (for additional glacier information see Table 8). The average modeled winter mass balance is 1130 ± 240 (Analysis 1) and 1010 ± 270 mm w.eq. (Analysis 2), indicating a significant difference (97.5% quantile) of 120 mm w.eq. (or 12%). The difference in average modeled winter mass balance is sensitive to changes in temperature, since colder air can carry less precipitable moisture. For the glaciers lying above the STL level, the Analysis 1 simulated average winter air temperature is relatively higher than the average winter temperature simulated by Analysis 2. The interannual variation in average winter balance for both Analysis 1 and 2 from 1998/99 through 2005/06 was almost similar (Fig. 8a), illustrating a nonsignificant decreasing trend of −14 and −11 mm w.eq. yr−1, respectively. The average modeled summer mass balances for Analysis 1 and 2 were −2010 ± 480 and −1730 ± 480 mm w.eq., respectively (Table 7), indicating a significant difference (97.5% quantile) of 280 mm w.eq. (16%). The interannual variation in average summer mass balance throughout the simulation period indicated a nonsignificant increasing loss of 50 mm w.eq. yr−1 (Analysis 1) and 63 mm w.eq. yr−1 (Analysis 2) (Fig. 8a). Based on simulated winter and summer balance values, the average annual glacier mass balance was −880 ± 620 and −720 ± 620 mm w.eq. yr−1 for Analysis 1 and 2, respectively, indicating a significant difference (97.5% quantile) of 160 mm w.eq. (22%) (Table 7). Throughout the simulation period the average annual mass balance indicated a nonsignificant increasing loss of 64 and 74 mm w.eq. yr−1 for Analysis 1 and 2, respectively (Fig. 8a), a loss related to the increased MAAT of the area of 0.09°C yr−1 (1998–2006). An example of the spatial modeled negative annual mass balance is shown in Fig. 9 for a collection of glaciers varying in size, elevation, aspect, and location for the year 2003/04 for both Analysis 1 and 2; they indicate the highest annual mass-loss rates at low elevations and, in general, for Analysis 1. The spatial variation in average annual mass balance (1998/99–2005/06) is shown in Tables 7 and 9. The Itsaqjivit Glacier (Glacier 10) had on average the lowest annual mass balance of −1800 ± 550 mm w.eq. yr−1 (Analysis 1), while the Vegas Fjeld Glacier (Glacier 17) had the highest annual mass balance of −300 ± 530 mm w.eq. yr−1 (Fig. 1c). These mass-balance variations are primarily related to the differences in elevation and aspect. For Analysis 2 the values were, respectively −1190 ± 700 and −90 ± 660 mm w.eq. yr−1 (Table 9), indicating the average annual mass balances were negative for all glaciers in both Analysis 1 and 2. In Fig. 8b, the average negative glacier annual mass balance is shown in relation to the average glacier elevation. This indicates that the average ELA (the ELA is defined as the elevation where the annual mass balance is zero, or where accumulation equals ablation) for the glaciers is located at 900 m MSL for Analysis 1 and at 875 m MSL for Analysis 2, showing a change in location of ELA of 25 m MSL. This is mainly due to changes in ablation of 280 mm w.eq. rather than changes in accumulation of 120 mm w.eq. (Table 7). Annual mass-balance observations from the Mittivakkat Glacier show that the average ELA was close at 700 m MSL, varying in elevation between 400 m MSL and up to more than 900 m MSL through the period from 1995/96 to 2005/06 (Knudsen and Hasholt 2008). This is in the range of simulated ELA elevation estimated by Analysis 1 and 2.
A negative glacier annual mass balance may be important from a water resource perspective. For the town Tasiilaq (Fig. 1a) the simulation domain is of potential interest for water resources. In Table 9 and Fig. 8c, the absolute values/percentages of the glacier mass loss (mm w.eq. yr−1 and m3 y−1) (Analysis 1) indicates that Glacier 2 (Glacier 783), 6 (Glacier 930), and 18 (Mittivakkat Glacier) are the main contributors with values around 11.9 × 106 m3 yr−1 (∼19% of total loss), 12.6 × 106 m3 yr−1 (∼20%), and 27.9 × 106 m3 yr−1 (∼43%), respectively. For Analysis 2, the values are less and around 8.7 × 106 m3 yr−1 (∼17%), 9.7 × 106 m3 yr−1 (∼19%), and 22.3 × 106 m3 yr−1 (∼44%), respectively. Modeled changes in glacier annual mass balance for Analysis 1 and 2 were −64.4 × 106 m3 yr−1 and −50.7 × 106 m3 yr−1, respectively, indicating a maximum difference of ∼21% (Table 9) for conditions with and without air temperature inversions.
Air temperature inversions are commonplace throughout the Arctic. Therefore, a pilot study like this can be used as a guide to emphasize the importance of 1) including air temperature inversion routines in climate models to improve snow and ice melt calculations, and to improve, more specifically, GrIS climate change simulations; and 2) detailed radiosonde observations in space and time, for example, around the GrIS to collect comprehensive information about temperature distributions, including inversion, since inversion according to Huybrechts et al. (1991) depends on latitude. As a first step, these model estimates show, for example, an up to ∼21% difference in glacier annual mass balance for simulations without or with air temperature inversions. This quantity, when projected over the entire GrIS ablation zone, is expected to affect GrIS surface mass-balance, freshwater runoff, and water resource predictions in important ways, and advocates the need to represent air temperature inversions in GrIS mass balance calculations and predictions.
5. Summary and conclusions
In Arctic coastal areas, air temperature inversions are a common feature. The physically based snow evolution modeling system (SnowModel) was modified with routines to account for the spatial distribution of air temperature inversions, and to describe the subsequent effect on snow accumulation, and snow and glacier ice ablation over the SW part of the Ammassalik Island in East Greenland. Based on vertical air temperature observations, the top of the temperature inversion layer was found to be approximately 300 m MSL, and was present in 84% of the summer observations. The observations were not detailed enough to provide a full annual understanding of the inversion height, strength, and thickness within and around all of Ammassalik Island. But, using a combination of observational datasets and physical understanding of the natural system, we justifiably assumed inversions were present 84% of the time with an inversion level of 300 m MSL throughout the 8-yr simulation period. Inversion absence was assumed when wind speeds were greater than 8 m s−1 and/or precipitation rates were greater than 10 mm w.eq. day−1. Sensitivity simulations were performed that changed the inversion level by 100 m over a 100–500-m range. These yielded uncertainties of 5% in both winter and summer glacier mass balance, and further justified our choice of an observation-based 300-m inversion level for the rest of our model simulations. These Ammassalik Island simulations showed that ignoring routines for air temperature inversion in Arctic coastal climate, snow, ice, and runoff modeling studies could create errors in snowmelt date, indicating an earlier or later snowmelt date by up to 1–3 weeks. The model simulations (1998/99–2005/06) yielded useful insights into the average glacier annual mass-balance consequences from a water resource perspective, representing a ∼21% difference in mass-balance loss without or with inversion routines of 64.4 × 106 and 50.7 × 106 m3 yr−1, respectively. Therefore, in many Arctic coastal landscapes we expect that a realistic description of temperature inversion is essential for accurate snow and glacier ice melt and glacier mass-balance simulations.
Acknowledgments
We extend a very special thanks to Jessica D. Lundquist, University of Washington, and the two anonymous reviewers for their insightful critique of this article. Thanks are given to the Cooperative Institute for Research in the Atmosphere, Colorado State University, for hosting the first author during October 2007 and February 2008, and the Faculty of Science and Institute of Low Temperature Science, Hokkaido University, Japan, for hosting the first author from April through July 2008. Thanks are also given to the Department of Geography and Geology, University of Copenhagen, for providing the input data, and to the Danish Meteorological Institute for providing WMO synoptic meteorological data from the station in Tasiilaq. This work was supported by grants from the University of Alaska Presidential IPY Postdoctoral Foundation, the University of Alaska Fairbanks (UAF) Office of the Vice Chancellor for Research, Office of the Director at the International Arctic Research Center, UAF, and the Water Environmental Research Center, UAF, and was carried out during the first author’s IPY Post-Doctoral Program at the UAF.
REFERENCES
Anquetin, S., C. Guilbaud, and J-P. Chollet, 1998: The formation and destruction of inversion layers within a deep valley. J. Appl. Meteor., 37 , 1547–1560.
Archer, D., 2004: Hydrological implications of spatial and altitudinal variation in temperature in the upper Indus basin. Nord. Hydrol., 35 , 209–222.
Barry, R. G., 2008: Mountain Weather and Climate. 3rd ed. Cambridge University Press, 506 pp.
Born, E. W., and J. Böcher, 2001: The Ecology of Greenland. Ministry of Environment and Natural Resources, 429 pp.
Chen, J., S. C. Sauners, T. R. Crow, R. J. Naiman, K. D. Brosofske, G. D. Morz, B. L. Brookshire, and J. F. Franklin, 1999: Microclimate in forest ecosystem and landscape ecology. Bioscience, 49 , 288–297.
Dodson, R., and D. Marks, 1997: Daily air temperature interpolation at high spatial resolution over a large mountainous region. Climate Res., 8 , 1–20.
Essery, R. L. H., L. Li, and J. W. Pomeroy, 1999: A distributed model of blowing snow over complex terrain. Hydrol. Processes, 13 , 2423–2438.
Greenland, D. E., 1979: Modelling air pollution potential for mountain resorts. Institute of Arctic and Alpine Research, Occasional Paper 32, Boulder, CO, University of Colorado, 96 pp.
Hanna, S. R., and D. G. Strimaitis, 1990: Rugged terrain effects on diffusion. Atmospheric Processes over Complex Terrain, Meteor. Monogr., No. 45, Amer. Meteor. Soc., 109–143.
Hansen, B. U., B. Elberling, O. Humlum, and N. Nielsen, 2005: Meteorological trends (1991–2004) at Arctic Station, Central West Greenland (69°15′N) in a 130 years perspective. Dan. J. Geogr., 106 (1) 45–56.
Hansen, B. U., and Coauthors, 2008: Present day climate at Zackenberg. Adv. Ecol. Res., 40 , 115–153.
Hosler, C. R., 1961: Low level inversion frequency in the contiguous United States. Mon. Wea. Rev., 89 , 319–339.
Huybrechts, P., A. Letreguilly, and N. Reeh, 1991: The Greenland ice sheet and greenhouse warming. Palaeogeogr. Palaeoclimatol. Palaeoecol., 89 , 399–412.
Kadygrov, E. N., A. S. Viazankin, E. R. Westwater, and K. B. Widener, 1999: Characteristics of the low-level temperature inversion at the North Slope of Alaska on the base of microwave remote sensing data. Proc. Ninth ARM Science Team Meeting, San Antonio, TX, Dept. of Energy, 1–7.
Kahl, J. D., 1990: Characteristic of the low-level temperature inversion along the Alaska Arctic Coast. Int. J. Climatol., 10 , 537–548.
Kahl, J. D., M. D. Serreze, and R. C. Schnell, 1992: Tropopsheric low-level temperature inversions in the Canadian Arctic. Atmos.–Ocean, 30 , 511–529.
Kahl, J. D., D. A. Martinez, and N. A. Zaitseva, 1996: Long-term variability in the low-level inversion layer over the Arctic Ocean. Int. J. Climatol., 16 , 1297–1313.
Kerminen, V., and Coauthors, 2007: Development of particle number size distribution near a major road in Helsinki during an episodic inversion situation. Atmos. Environ., 41 , 1759–1767.
Klemes, V., 1985: Sensitivity of water resources systems to climate variations. WPC Rep. 98, World Meteorological Organization, 56 pp.
Klemes, V., 1986: Operational testing of hydrological simulation models. Hydrol. Sci. J., 31 , 13–24.
Knudsen, N. T., and B. Hasholt, 2008: Mass balance observations of the Mittivakkat Glacier, Ammassalik Island, Southeast Greenland 1995–2006. Dan. J. Geogr., 108 (1) 111–120.
Kozo, T. L., 1982a: An observational study of sea breezes along the Alaskan Beaufort Sea Coast: Part I. J. Appl. Meteor., 21 , 891–905.
Kozo, T. L., 1982b: A mathematical model of sea breezes along the Alaskan Beaufort Sea Coast: Part II. J. Appl. Meteor., 21 , 906–924.
Kunkel, K. E., 1989: Simple procedures for extrapolation of humidity variables in the mountainous western United States. J. Climate, 2 , 656–669.
Liston, G. E., 1995: Local advection of momentum, heat, and moisture during the melt of patchy snow covers. J. Appl. Meteor., 34 , 1705–1715.
Liston, G. E., and D. K. Hall, 1995: An energy-balance model of lake-ice evolution. J. Glaciol., 41 , 373–382.
Liston, G. E., and M. Sturm, 1998: A snow-transport model for complex terrain. J. Glaciol., 44 , 498–516.
Liston, G. E., and M. Sturm, 2002: Winter precipitation patterns in arctic Alaska determined from a blowing-snow model and snow-depth observations. J. Hydrometeor., 3 , 646–659.
Liston, G. E., and M. Sturm, 2004: The role of winter sublimation in the Arctic moisture budget. Nord. Hydrol., 35 , 325–334.
Liston, G. E., and K. Elder, 2006a: A meteorological distribution system for high-resolution terrestrial modelling (MicroMet). J. Hydrometeor., 7 , 217–234.
Liston, G. E., and K. Elder, 2006b: A distributed snow evolution modeling system (SnowModel). J. Hydrometeor., 7 , 1259–1276.
Liston, G. E., J-G. Winther, O. Bruland, H. Elvehøy, and K. Sand, 1999: Below surface ice melt on the coastal Antarctic ice sheet. J. Glaciol., 45 (150) 273–285.
Liston, G. E., R. B. Haehnel, M. Sturm, C. A. Hiemstra, S. Berezovskaya, and R. D. Tabler, 2007: Simulating complex snow distributions in windy environments using SnowTran-3D. J. Glaciol., 53 , 241–256.
Lundquist, L. D., and D. R. Cayan, 2007: Surface temperature patterns in complex terrain: Daily variations and long-term change in the central Sierra Nevada, California. J. Geophys. Res., 112 , D11124. doi:10.1029/2006JD007561.
Marinec, J., and A. Rango, 1986: Parameter values for snowmelt runoff modelling. J. Hydrol., 84 , 197–219.
Mernild, S. H., 2006: Freshwater discharge from the coastal area outside the Greenland Ice Sheet, East Greenland, with focus on Mittivakkat, Ammassalik Island (low Arctic) and Zackenberg (high Arctic). Ph.D. dissertation, Institute of Geography, Faculty of Science, University of Copenhagen, 394 pp.
Mernild, S. H., B. Hasholt, and G. E. Liston, 2006a: Water flow through Mittivakkat Glacier, Ammassalik Island, SE Greenland. Dan. J. Geogr., 106 (1) 25–43.
Mernild, S. H., G. E. Liston, B. Hasholt, and N. T. Knudsen, 2006b: Snow-distribution and melt modeling for Mittivakkat Glacier, Ammassalik Island, SE Greenland. J. Hydrometeor., 7 , 808–824.
Mernild, S. H., G. E. Liston, and B. Hasholt, 2007: Snow-distribution and melt modeling for glaciers in Zackenberg River drainage basin, NE Greenland. Hydrol. Processes, 21 , 3249–3263. doi:10.1002/hyp.6500.
Mernild, S. H., B. U. Hansen, B. H. Jakobsen, and B. Hasholt, 2008a: Climatic conditions at the Mittivakkat Glacier catchment (1994–2006), Ammassalik Island, SE Greenland, and in a 109 years term perspective (1898–2006). Dan. J. Geogr., 108 (1) 49–70.
Mernild, S. H., D. L. Kane, B. U. Hansen, B. H. Jakobsen, B. Hasholt, and N. T. Knudsen, 2008b: Climate, glacier mass balance, and runoff (1993–2005) Mittivakkat Glacier catchment, Ammassalik Island, SE Greenland, and in a long term perspective (1898–1993). Hydrol. Res., 39 (4) 239–256.
Mernild, S. H., G. E. Liston, and B. Hasholt, 2008c: East Greenland freshwater runoff to the North Atlantic Ocean 1999–2004 and 2071–2100. Hydrol. Processes, 22 , 4571–4586. doi:10.1002/hyp.7061.
Mernild, S. H., G. E. Liston, C. A. Hiemstra, and K. Steffen, 2008d: Surface melt area and water balance modeling on the Greenland Ice Sheet 1995–2005. J. Hydrometeor., 9 , 1191–1211.
Mernild, S. H., G. E. Liston, D. L. Kane, B. Hasholt, and N. T. Knudsen, 2008e: Spatial snow distribution, runoff, and mass balance modelling for entire Mittivakkat Glacier (1998–2006), Ammassalik Island, SE Greenland. Dan. J. Geogr., 108 (1) 125–140.
Mernild, S. H., G. E. Liston, C. A. Hiemstra, and K. Steffen, 2009: Record 2007 Greenland Ice Sheet surface melt extent and runoff. Eos, Trans. Amer. Geophys. Union, 90 .doi:10.1029/2009EO020002.
Milionis, A. E., and T. D. Davies, 2008: A comparison of temperature inversion statistics at a coastal and a non-coastal location influenced by the same synoptic regime. Theor. Appl. Climatol., 94 , 225–239.
Oke, T. R., 1987: Boundary Layer Climates. 2nd ed. Routledge, 435 pp.
Pomeroy, J. W., and D. M. Gray, 1995: Snowcover accumulation, relocation and management. National Hydrology Research Institute Science Rep. 7, Hydrological Science Division, University of Saskatchewan, 144 pp.
Pomeroy, J. W., and R. L. H. Essery, 1999: Turbulent fluxes during blowing snow: Field test of model sublimation predictions. Hydrol. Processes, 13 , 2963–2975.
Pomeroy, J. W., and E. Brun, 2001: Physical properties of snow. Snow Ecology: An Interdisciplinary Examination of Snow-Covered Ecosystems, H. G. Jones et al., Eds., Cambridge University Press, 45–126.
Pomeroy, J. W., P. Marsh, and D. M. Gray, 1997: Application of distributed blowing snow model to the Arctic. Hydrol. Processes, 11 , 1450–1465.
Putnins, P., 1970: The climate of Greenland. Climates of the Polar Regions, S. Orvig, Ed., Vol. 12, World Survey of Climatology, Elsevier, 3–112.
Refsgaard, J. C., 2000: Towards a formal approach to calibration and validation of models using spatial data. Spatial Patterns in Catchment Hydrology: Observations and Modelling, R. Gryson and G. Bloschl, Eds., Cambridge University Press, 329–354.
Refsgaard, J. C., and J. Knudsen, 1996: Operational validation and intercomparison of different types of hydrological models. Water Resour. Res., 32 , 2189–2202.
Refsgaard, J. C., and H. J. Henriksen, 2004: Modelling guidelines—terminology and guiding principles. Adv. Water Resour., 27 , 71–82.
Riordan, A. J., J. M. Davis, and R. B. Kiess, 1986: The morning inversion near the ground and its daytime transition at two rural sites in the Cerolinas. J. Climate Appl. Meteor., 25 , 239–256.
Schmidt, R. A., 1972: Sublimation of wind-transported snow—A model. Research Paper RM-90, Rocky Mountain Forest and Range Experiment Station, Forest Service, U.S. Department of Agriculture, Fort Collins, CO, 24 pp.
Schmidt, R. A., 1982: Vertical profiles of wind speed, snow concentration, and humidity in blowing snow. Bound.-Layer Meteor., 23 , 223–246.
Serreze, M. C., J. D. Kahl, and R. C. Schnell, 1992: Low-level temperature inversions of the Eurasian Arctic and comparisons with Soviet drifting station data. J. Climate, 5 , 615–629.
Singh, P., 1999: A temperature lapse rate study in Western Himalayas. Hydrol. J. Indian Assoc. Hydrol., 14 , 156–163.
Streten, N. A., N. Ishikawa, and G. Wendler, 1974: Some observations of the local wind regime on an Alaskan Arctic Glacier. Arch. Meteor. Geophys. Bioklimatol. Ser. B, 22 , 337–350.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic, 666 pp.
Tabler, R. D., 1975: Estimating the transport and evaporation of blowing snow. Proc. Snow Management on the Great Plains Symp., Publication 73, Bismarck, ND, Great Plains Agricultural Council, 85–104.
Walcek, C. J., 1994: Cloud cover and its relationship to relative humidity during a spring midlatitude cyclone. Mon. Wea. Rev., 122 , 1021–1035.
Weick, E. J., and W. R. Rouse, 1991: Advection in the coastal Hudson Bay Lowlands, Canada. II. Impact of atmospheric divergence on the surface energy balance. Arct. Alp. Res., 23 , 338–348.
Whiteman, C. D., 1982: Breakup at temperature inversions in deep mountain valleys: Part I. Observations. J. Appl. Meteor., 21 , 270–289.
Whiteman, C. D., X. Bian, and S. Zhong, 1999: Wintertime evolution of the temperature inversion in the Colorado Plateau basin. J. Appl. Meteor., 38 , 1103–1117.
Southwest Ammassalik Island simulation domain: (a) topography (gray shades, 100-m contour interval), the entire Mittivakkat Glacier complex outlined by the black line, and the three meteorological tower stations: Station Nunatak (515 m MSL), Station Coast (25 m MSL), and Station Tasiilaq (44 m MSL; a standard synoptic WMO meteorological station operated by the Danish Meteorological Institute); (b) surface characteristics including inversion height; (c) glacier numbers and glacier observation area at the Mittivakkat Glacier (17.6 km2); and (d) location of the snow cover depletion areas and longitudinal profiles A–B and C–D. The inset figure in (a) indicates the general location of the Mittivakkat Glacier in eastern Greenland. The domain coordinates can be converted to UTM by adding 548 km to the west–east origin (easting) and 7272 km to the south–north origin (northing) and converting to meters.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
Southwest Ammassalik Island simulation domain: (a) topography (gray shades, 100-m contour interval), the entire Mittivakkat Glacier complex outlined by the black line, and the three meteorological tower stations: Station Nunatak (515 m MSL), Station Coast (25 m MSL), and Station Tasiilaq (44 m MSL; a standard synoptic WMO meteorological station operated by the Danish Meteorological Institute); (b) surface characteristics including inversion height; (c) glacier numbers and glacier observation area at the Mittivakkat Glacier (17.6 km2); and (d) location of the snow cover depletion areas and longitudinal profiles A–B and C–D. The inset figure in (a) indicates the general location of the Mittivakkat Glacier in eastern Greenland. The domain coordinates can be converted to UTM by adding 548 km to the west–east origin (easting) and 7272 km to the south–north origin (northing) and converting to meters.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
Southwest Ammassalik Island simulation domain: (a) topography (gray shades, 100-m contour interval), the entire Mittivakkat Glacier complex outlined by the black line, and the three meteorological tower stations: Station Nunatak (515 m MSL), Station Coast (25 m MSL), and Station Tasiilaq (44 m MSL; a standard synoptic WMO meteorological station operated by the Danish Meteorological Institute); (b) surface characteristics including inversion height; (c) glacier numbers and glacier observation area at the Mittivakkat Glacier (17.6 km2); and (d) location of the snow cover depletion areas and longitudinal profiles A–B and C–D. The inset figure in (a) indicates the general location of the Mittivakkat Glacier in eastern Greenland. The domain coordinates can be converted to UTM by adding 548 km to the west–east origin (easting) and 7272 km to the south–north origin (northing) and converting to meters.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
Vertical observed air temperature distribution near Station Coast during daytime with different weather conditions: sunny, foggy, cloudy, and rainy conditions from June through August 2005 and 2006. Air temperature was measured at approximately every 20 m of elevation. The dotted line indicates the top of the observed temperature inversion level at 300 m MSL used for the model simulations.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
Vertical observed air temperature distribution near Station Coast during daytime with different weather conditions: sunny, foggy, cloudy, and rainy conditions from June through August 2005 and 2006. Air temperature was measured at approximately every 20 m of elevation. The dotted line indicates the top of the observed temperature inversion level at 300 m MSL used for the model simulations.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
Vertical observed air temperature distribution near Station Coast during daytime with different weather conditions: sunny, foggy, cloudy, and rainy conditions from June through August 2005 and 2006. Air temperature was measured at approximately every 20 m of elevation. The dotted line indicates the top of the observed temperature inversion level at 300 m MSL used for the model simulations.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
A comparison between mean daily-observed meteorological data (a) wind speed, (b) air temperature, (c) relative humidity, and (d) precipitation, and mean daily SnowModel–MicroMet Analysis 1 simulated meteorological data for Station Tasiilaq (1998–2006) (for station location see Fig. 1a). Only precipitation values >1 mm w.eq. were included.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
A comparison between mean daily-observed meteorological data (a) wind speed, (b) air temperature, (c) relative humidity, and (d) precipitation, and mean daily SnowModel–MicroMet Analysis 1 simulated meteorological data for Station Tasiilaq (1998–2006) (for station location see Fig. 1a). Only precipitation values >1 mm w.eq. were included.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
A comparison between mean daily-observed meteorological data (a) wind speed, (b) air temperature, (c) relative humidity, and (d) precipitation, and mean daily SnowModel–MicroMet Analysis 1 simulated meteorological data for Station Tasiilaq (1998–2006) (for station location see Fig. 1a). Only precipitation values >1 mm w.eq. were included.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Schematic illustration of the Analysis (left) 1 and (right) 2 air temperature lapse-rate routines. For Analysis 2, the air temperature inversion level was set at 300 m MSL according to observations (see Fig. 2). In general, for Analysis 2, the air temperature increased with height up to the inversion level, and above the air temperature decreased. (b) An example of the average air temperature distribution with increasing elevation for Analysis 1 and 2 for the months of January, April, July, and October 2004. A negative difference between Analysis 1 and 2 indicates that Analysis 1 estimated air temperature is below Analysis 2 estimated values. Inversion-level and STL is illustrated. Further, a thick line is shown on the ordinate to illustrate the main glacier elevations (ranging from 300 to 800 m MSL).
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Schematic illustration of the Analysis (left) 1 and (right) 2 air temperature lapse-rate routines. For Analysis 2, the air temperature inversion level was set at 300 m MSL according to observations (see Fig. 2). In general, for Analysis 2, the air temperature increased with height up to the inversion level, and above the air temperature decreased. (b) An example of the average air temperature distribution with increasing elevation for Analysis 1 and 2 for the months of January, April, July, and October 2004. A negative difference between Analysis 1 and 2 indicates that Analysis 1 estimated air temperature is below Analysis 2 estimated values. Inversion-level and STL is illustrated. Further, a thick line is shown on the ordinate to illustrate the main glacier elevations (ranging from 300 to 800 m MSL).
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Schematic illustration of the Analysis (left) 1 and (right) 2 air temperature lapse-rate routines. For Analysis 2, the air temperature inversion level was set at 300 m MSL according to observations (see Fig. 2). In general, for Analysis 2, the air temperature increased with height up to the inversion level, and above the air temperature decreased. (b) An example of the average air temperature distribution with increasing elevation for Analysis 1 and 2 for the months of January, April, July, and October 2004. A negative difference between Analysis 1 and 2 indicates that Analysis 1 estimated air temperature is below Analysis 2 estimated values. Inversion-level and STL is illustrated. Further, a thick line is shown on the ordinate to illustrate the main glacier elevations (ranging from 300 to 800 m MSL).
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) SnowModel–MicroMet spatial simulated 2-m air temperature for the Ammassalik Island, for the coldest day during the simulation period (21 Feb 2002): (top left) Analysis 1 simulated spatial air temperature; (top right) the spatial difference in air temperature between Analysis 1 and 2, where negative temperatures indicate Analysis 1 estimated values are below Analysis 2 estimated values; (middle) the Analysis 1 and 2 modeled temperatures at the A–B longitudinal profile; (bottom) the Analysis 1 and 2 modeled temperatures at the C–D longitudinal profile. At the longitudinal profiles, the STL (553 m MSL) is shown. For the profile locations see Fig. 1d. (b) As in (a), but for the warmest day during the simulation period (13 Jul 2005). At the longitudinal profiles, the STL (494 m MSL) is shown. For the profile locations see Fig. 1d.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) SnowModel–MicroMet spatial simulated 2-m air temperature for the Ammassalik Island, for the coldest day during the simulation period (21 Feb 2002): (top left) Analysis 1 simulated spatial air temperature; (top right) the spatial difference in air temperature between Analysis 1 and 2, where negative temperatures indicate Analysis 1 estimated values are below Analysis 2 estimated values; (middle) the Analysis 1 and 2 modeled temperatures at the A–B longitudinal profile; (bottom) the Analysis 1 and 2 modeled temperatures at the C–D longitudinal profile. At the longitudinal profiles, the STL (553 m MSL) is shown. For the profile locations see Fig. 1d. (b) As in (a), but for the warmest day during the simulation period (13 Jul 2005). At the longitudinal profiles, the STL (494 m MSL) is shown. For the profile locations see Fig. 1d.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) SnowModel–MicroMet spatial simulated 2-m air temperature for the Ammassalik Island, for the coldest day during the simulation period (21 Feb 2002): (top left) Analysis 1 simulated spatial air temperature; (top right) the spatial difference in air temperature between Analysis 1 and 2, where negative temperatures indicate Analysis 1 estimated values are below Analysis 2 estimated values; (middle) the Analysis 1 and 2 modeled temperatures at the A–B longitudinal profile; (bottom) the Analysis 1 and 2 modeled temperatures at the C–D longitudinal profile. At the longitudinal profiles, the STL (553 m MSL) is shown. For the profile locations see Fig. 1d. (b) As in (a), but for the warmest day during the simulation period (13 Jul 2005). At the longitudinal profiles, the STL (494 m MSL) is shown. For the profile locations see Fig. 1d.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Digital elevation for the three snow cover depletion areas (1.1 km × 1.5 km; 1.65 km2): the (from left to right) coastal–valley area, glacier area, and peak area. (b) For the same areas, snow cover depletion and SWE depth curves for (top) 2003 (the year with the lowest ablation) and (bottom) 2005 (the year with the greatest ablation) both for Analysis 1 and 2, for the period 1 May (DOY 121)–31 Aug (DOY 243). For location of the three depletion areas see Fig. 1d.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Digital elevation for the three snow cover depletion areas (1.1 km × 1.5 km; 1.65 km2): the (from left to right) coastal–valley area, glacier area, and peak area. (b) For the same areas, snow cover depletion and SWE depth curves for (top) 2003 (the year with the lowest ablation) and (bottom) 2005 (the year with the greatest ablation) both for Analysis 1 and 2, for the period 1 May (DOY 121)–31 Aug (DOY 243). For location of the three depletion areas see Fig. 1d.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Digital elevation for the three snow cover depletion areas (1.1 km × 1.5 km; 1.65 km2): the (from left to right) coastal–valley area, glacier area, and peak area. (b) For the same areas, snow cover depletion and SWE depth curves for (top) 2003 (the year with the lowest ablation) and (bottom) 2005 (the year with the greatest ablation) both for Analysis 1 and 2, for the period 1 May (DOY 121)–31 Aug (DOY 243). For location of the three depletion areas see Fig. 1d.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
SnowModel spatial simulated snow cover extent for the SW part of the Ammassalik Island for (top) 15 May, (middle) 1 Jun, and (bottom) 15 Jun 2005 for (left) Analysis 1 and (right) the difference between Analysis 1 and 2. The percentage and area of snow cover extent is also shown for each time step.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
SnowModel spatial simulated snow cover extent for the SW part of the Ammassalik Island for (top) 15 May, (middle) 1 Jun, and (bottom) 15 Jun 2005 for (left) Analysis 1 and (right) the difference between Analysis 1 and 2. The percentage and area of snow cover extent is also shown for each time step.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
SnowModel spatial simulated snow cover extent for the SW part of the Ammassalik Island for (top) 15 May, (middle) 1 Jun, and (bottom) 15 Jun 2005 for (left) Analysis 1 and (right) the difference between Analysis 1 and 2. The percentage and area of snow cover extent is also shown for each time step.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Analysis 1 and 2 SnowModel simulated average winter, summer, and annual mass balance for all glaciers in the simulation domain from 1998/99 through 2005/06, except for glacier 2 (Glacier 783), which is part of a larger glacier complex ranging outside the domain; (b) Analysis 1 and 2 average modeled glacier annual mass balance plotted against the average glacier elevation; (c) Analysis 1 and 2 change in average glacier storage (%).
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Analysis 1 and 2 SnowModel simulated average winter, summer, and annual mass balance for all glaciers in the simulation domain from 1998/99 through 2005/06, except for glacier 2 (Glacier 783), which is part of a larger glacier complex ranging outside the domain; (b) Analysis 1 and 2 average modeled glacier annual mass balance plotted against the average glacier elevation; (c) Analysis 1 and 2 change in average glacier storage (%).
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
(a) Analysis 1 and 2 SnowModel simulated average winter, summer, and annual mass balance for all glaciers in the simulation domain from 1998/99 through 2005/06, except for glacier 2 (Glacier 783), which is part of a larger glacier complex ranging outside the domain; (b) Analysis 1 and 2 average modeled glacier annual mass balance plotted against the average glacier elevation; (c) Analysis 1 and 2 change in average glacier storage (%).
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
SnowModel simulated spatial annual mass balance for different glaciers on Ammassalik Island for the period 2003/04 based on Analysis 1 and 2. The glaciers refer to the numbers in Table 8 and Fig. 1c.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
SnowModel simulated spatial annual mass balance for different glaciers on Ammassalik Island for the period 2003/04 based on Analysis 1 and 2. The glaciers refer to the numbers in Table 8 and Fig. 1c.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
SnowModel simulated spatial annual mass balance for different glaciers on Ammassalik Island for the period 2003/04 based on Analysis 1 and 2. The glaciers refer to the numbers in Table 8 and Fig. 1c.
Citation: Journal of Applied Meteorology and Climatology 49, 1; 10.1175/2009JAMC2065.1
The average monthly occurrence of air temperature inversions at Station Tasiilaq (1996–2005) and near Station Coast (June–August 2005 and 2006). Winter observations near Station Coast were not conducted because of the harsh climatic conditions and logistical constraints.
Station Coast and Station Nunatak observed daily wind speed (Ws) and precipitation (P), and Station Tasiilaq radiosonde observed wind speed. The table shows the average and frequency of wind speed below and above 8 m s−1 for the three stations, and precipitation below and above 10 mm w.eq. for Station Coast and Station Nunatak. Further, average summer (June–August) and winter (September–May) wind speed and average wind speed during periods with radiosonde observed inversion and no inversion are shown. The radiosonde observations are conducted at different pressure levels (1996–2005): terrain surface (52 m MSL), 925 hPa (631 ± 108 m MSL), 850 hPa (1329 ± 114 m MSL), and 700 hPa (2871 ± 141 m MSL).
Observed and SnowModel Analysis 1 simulated daily and hourly winter, summer, and annual mass balance for the Mittivakkat Glacier observation area (17.6 km2) (1998/99–2005/06). Winter mass-balance observations were carried out in late May and in early June, summer mass-balance observations in late August, modeled winter mass balance 31 May, and modeled summer mass balance 31 Aug. The annual mass-balance calculations span the period from 1 Sep through 31 Aug of the next year. Observed data are based on Knudsen and Hasholt (2008) and Mernild et al. (2008e).
User-defined constants used in the SnowModel simulations [see Liston and Sturm (1998) for parameter definitions].
SnowModel Analysis 2 simulated winter, summer, and annual mass balance for the Mittivakkat Glacier observation area (17.6 km2) for different inversion elevations. Simulations were conducted for inversions at 100-m intervals going from 0 through 500 m MSL. The simulations span the period from 1 Sep 2000 through 31 Aug 2001.
Average monthly lapse rates for Analysis 1 and 2. For Analysis 2, lapse rates are defined both below (ΔTlapsebelow) and above (ΔTlapseabove) the top of the air temperature inversion layer. During Analysis 2 simulation periods with no inversion, Analysis 1 lapse rates were used. Lapse rates are based on air temperature data for the period 1999–2004 (see Fig. 4 for a schematic illustration).
Analysis 1 and 2 SnowModel simulated average winter, summer, and annual mass balance for glaciers on the SW part of the Ammassalik Island for the 8-yr period 1998/99–2005/06. For glacier information see Table 8.
A description of the glaciers in the simulation domain by number, name, area, and elevation based on the 100-m gridcell increment. The Glacier numbers refers to Fig. 1c.
