Examensarbete vid Institutionen för Geovetenskaper
ISSN 1650-6553 Nr 82
An investigation of the surface
fluxes and other parameters in
the regional climate model RCA1
during ice conditions
Abstract
In this study data from the regional climate model RCA1 is compared to measured data to investigate how good the model is during ice conditions and mainly how well the turbulent surface fluxes are described by the model. Comparisons of the sensible heat flux and the momentum flux as well as mean parameters are included. The measured data used are from the Bothnian Bay measured during the BASIS field experiment in February to March 1998. RCA1 (Rossby Centre regional Atmospheric model) is a regional climate model for Northern Europe based on HIRLAM and forced by ERA-40 data. Two different grid points of the RCA1 44 km grid have been chosen with geographical coordinates as close as possible to the two measuring sites. The first site is a small peninsula south of the town Umeå at the east coast of Sweden and the second one is the ship R/V Aranda anchored in the sea ice outside the Finnish west coast.
The model presents generally too large negative (downward) sensible heat fluxes and too large momentum fluxes over ice. The largest difference between modelled and measured sensible heat fluxes are seen after warm front passages due to melting conditions. There are some uncertainties in the comparison of modelled fluxes and measured fluxes at Umeå due to a complex and varying ice cover around this measuring site.
The vertical structure in the atmosphere has also been studied and modelled temperature, wind and humidity profiles were compared to radiosondes at the Umeå site. Two periods with ice flow and off-ice flow was analysed and one of the main differences was that for the on-ice situation an internal boundary layer was built up over the on-ice. Modelled profiles are generally smoother then measured profiles and inversions and other small-scale phenomena like low-level jets are almost never described correctly by the model but there are tendencies of the phenomena to occur.
Sammanfattning av ”En undersökning i hur bra den regionala
klimatmodellen RCA1 beskriver olika parametrar över is, i huvudsak
de turbulenta flödena”.
I denna undersökning har den regionala klimatmodellen RCA1 jämförts med mätdata för att se hur bra modellen är över is och framförallt hur bra den beskriver de turbulenta flödena över is. RCA1 är en regional klimatmodell för norra Europa baserad på HIRLAM och som drivs av ERA-40 data. Modelldata har jämförts med mätdata från två platser i Bottenviken, en liten halvö utanför Umeå på den svenska östkusten och forskningsfartyget Aranda som var fastankrad i havsisen utanför Finlands västkust. Mätningarna gjordes under BASIS fältexperiment i februari till mars 1998.
Modellen ger generellt för stora negativa (nedåtriktade) sensibla värmeflöden över is och likaså för stora impulsflöden över is. Den största skillnaden mellan modellvärden och mätningar förekom efter det att varmfronter passerat och detta beror delvis på
smältförhållanden. I jämförelsen av de turbulenta flödena beräknade av RCA och de uppmäta turbulenta flödena från Umeå finns det vissa osäkerheter pga. den komplicerade issituationen runt denna mätplats med mycket varierande isförhållanden.
Den vertikala strukturen i atmosfären har också undersökts och temperatur, vind och fuktighetsprofiler har jämförts. Två perioder varav en där det blåste från havet mot isen (on-ice) och en där det blåste från isen mot havet (off-(on-ice) valdes ut och en av skillnaderna var att det bildades ett internt gränsskikt över isen i on-ice perioden. Modellen har en tendens att släta ut profilerna och småskaliga fenomen som inversionsskikt, inversionshöjder och low-level jets är nästan aldrig korrekt beskrivna i modellen men det finns tendenser till inversioner och vindmaximum.
Table of contents
1.
Introduction ... 5
2.
Model description... 6
2.1 A regional climate model ... 6
2.2 RCA1... 6
3.
Sites and measurements... 9
3.1 Umeå ... 9
3.2 R/V Aranda ... 10
4.
Description of the measuring period ... 11
4.1 Ice conditions ... 11
4.2 Weather Conditions... 13
5.
Results ... 14
5.1 Mean Parameters ... 14
5.2 Turbulent surface fluxes... 16
5.2.1 The sensible heat flux... 17
5.2.2 The momentum flux ... 20
5.3 Net radiation and Energy balance ... 22
5.3.1 Net short- and longwave radiation ... 22
5.3.2 Energy balance ... 24
5.4 Vertical structure in the atmosphere... 25
5.4.1 On-ice flow... 25
5.4.2 Off-ice flow ... 27
6.
Discussion... 29
7.
Summary and Conclusions... 30
8.
Acknowledgments ... 31
9.
References ... 32
1. Introduction
The aim of this study is to compare the regional climate model, RCA1, with measurements from the Bothnian Bay and analyse how good the model is during ice conditions and
especially how well the fluxes of momentum and heat are described by the model. Modelled time series of the sensible heat flux and the momentum flux as well as temperature, wind speed and net radiation have been compared with measured values. Vertical profiles of wind, temperature and relative humidity have also been studied. Two cases of on-ice air flow and off-ice air flow have been studied in more detail.
The Rossby Centre regional Atmospheric model, RCA1 (Rummukainen et al., 2001) is a regional climate model for northern Europe based on the High Resolution Limited Area Model (HIRLAM). At the boundaries RCA1 is forced by ERA data
(http://www.ecmwf.int/research/era/) and has a 44 km horizontal resolution. The model also includes components for land areas and inland lakes.
The measured data used for this study are from the northern Baltic Sea, measured during the BASIS project (Baltic Air-Sea-Ice Study) in February-March 1998. BASIS is a field
experiment of BALTEX (Baltic Sea Experiment) carried out by scientists from Finland, Sweden, Germany and Japan (Launiainen, 2001). The main goal of BASIS was to create and analyse an experimental data set for further understanding of energy- and water cycles during winter conditions and to improve coupled atmosphere-ice-ocean models. Two data sets from two different measuring sites have been used in this study. The first site is a small peninsula outside the Swedish east coast and the town Umeå and the second site is the ship R/V Aranda anchored in the sea ice outside the Finnish West Coast.
For the comparison between measured data and modelled data 2 grid points of the RCA1 44 km grid are chosen. These two grid points have geographical coordinates as close as possible to those of the meteorological stations.
The interaction between air and ice is a problem in many studies and there are difficulties with ice conditions and to measure fluxes over ice. The interaction between atmosphere and an inhomogeneous surface such as fractured sea-ice is complex and most climate and weather forecast models cannot resolve details between an inhomogeneous ice surface and the
atmosphere (Brümmer et al., 2002). Measured data over ice are few so data used for this study containing turbulent surface fluxes over ice is rather unique. It’s important to get correct surface fluxes in regional climate models due to the fluxes influence on the atmospheric boundary layer. A problem in many models is to correctly describe the stable stratification during winter conditions.
2. Model description
2.1 A regional climate model
A climate model is in principle constructed like a weather forecast model but instead of describing the weather a few days ahead it describes past or future climate for decades or centuries. Climate models are also used to study the variability and physical processes of the climate system (www.met-office.gov.uk/research/hadleycentre/models/modeltypes.html, March 2004). The main difference is that a numerical weather forecast model is forced by observations on a regular basis while a climate model runs with very little external forcing. For climate prediction a global climate model (atmosphere general circulation model) is often coupled to an ocean general circulation model to get the information of changes in the ocean. These global models have a coarse resolution. Models with higher resolution cannot be used for climate simulations of long periods of time since they are very computer demanding. To get more detailed information for specific regions regional climate models have been developed and they are usually run for shorter periods (about 20 years).
In a global climate model the atmosphere and the oceans are divided into a three-dimensional grid where in each grid box prognostic variables (temperature, wind, humidity, clouds etc) gives a certain value. Changes are calculated in time based on physical processes in the atmosphere. Energy and mass are transported between the different grid boxes. The amounts of details that can be described depend on the size of the grid. A global climate model has a 200-300 km horizontal resolution that makes it difficult to describe regional and local variations. Chain of mountains and the distribution of land and sea are hard for the model to catch because of its coarse resolution.
Various techniques exist to add detail to global climate model results. This is called downscaling or regionalization (Rummukainen et al., 2001). There are two categories, statistical- or dynamical downscaling. The use of statistical techniques are limited by their link to past climate statistics but it doesn’t require a regional climate model. Dynamical downscaling involves the use of regional climate models with boundary conditions taken from global climate model results.
2.2 RCA1
The regional climate model used for this study is mainly the model RCA1 developed at the Rossby Centre, SMHI. It has a 44 km horizontal resolution. RCA1 is usually forced with a global climate model (for example ECHAM or Hadley). For investigation of the accuracy of the model or development of parameterisations in the model it is driven by analysed model data. In this study data from ERA-40 (ECMWF re-analysis) is used. RCA1 is version 1 (Rummukainen et al., 2001) of the Rossby Centre regional Atmospheric model (RCA1) developed from HIRLAM (High Resolution Limited Area Model). This is a regional climate model for northern Europe and the key tool in SWECLIM (Swedish Climate Modelling programme).
RCA1 is a hydrostatic, primitive equation grid point model. For each grid point, distributions of the prognostic variables temperature, specific humidity, horizontal wind, cloud water and surface pressure are calculated. The time resolution is 30 minutes. There are 24 vertical levels in the model domain and the model top is at 10 hPa. The parameterisations in RCA1 are
mostly based on HIRLAM. Only parameterisations and details of relevance to this study are mentioned in this paper.
The turbulent fluxes in RCA1 are computed from drag formulae (HIRLAM-5 Scientific Documentation, 2002): ' 'θ ρ c w H = ⋅ p⋅ (1) ' ' w u ⋅ − = ρ τ
( )
( )
Ri T U f z z z z k z z k c H z z z z k z z k C Ri f C C U T C w w c H m H M M M p H M M M HN m HN H H p ⋅ ∆ ⋅ ⋅ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ + ⋅ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ ⋅ = ⇒ ⎪ ⎪ ⎪ ⎪ ⎪ ⎭ ⎪⎪ ⎪ ⎪ ⎪ ⎬ ⎫ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ + ⋅ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ = ⋅ = ⋅ ∆ ⋅ = ⋅ ⋅ = 0 0 0 0 0 0 0 0 ln ln ln ln ln ln ' ' ' ' ρ θ θ ρ (2)( )
( )
2 2 0 2 0 2 ln ln ' ' ' ' U Ri f z z k z z k C Ri f C C U C w u w u h M M MN h MN M M ⋅ ⋅ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ ⋅ − = ⇒ ⎪ ⎪ ⎪ ⎪ ⎪ ⎭ ⎪⎪ ⎪ ⎪ ⎪ ⎬ ⎫ ⎟ ⎟ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎜ ⎜ ⎝ ⎛ = ⋅ = ⋅ = ⋅ − = ρ τ ρ τ (3)H and τ are the turbulent fluxes of sensible heat and momentum, w'θ' the kinematic heat flux and u' w' the kinematic flux of momentum. U is the wind speed and ∆T is the difference between surface and air temperature. The heat capacity for air per mass unit at constant pressure (cp) is 1004.71 J/kg·K and ρ is the density of air, 1.293 kg/m3. CH and CM are the bulk transfer coefficients for sensible heat and momentum. CHN and CMN are the bulk
coefficients for neutral stratification. Von Karman’s constant (k) is 0.4, z is the height above surface, z0M and z0H the roughness length for momentum and sensible heat and fm(Ri) and fh(Ri) are functions depending on the Richardson number for momentum and heat (HIRLAM-5 Scientific Documentation, 2002).
In a regional scale model there are always a number of processes that are parameterised and simplified. In RCA1 one of them is that roughness lengths (z0) for momentum, heat and moisture are the same over ice. The roughness length for ice is taken to be 0.03m (HIRLAM-5 Scientific Documentation, 2002). This value is much higher than values for a plane ice surface. Natural ice fields are irregular and to account for this the value has to be higher. The roughness lengths, surface exchanges and diagnostic near-surface variables are calculated separately for a land fraction and a water fraction in each grid box. The fraction of a grid box is treated as land when sea ice is present. The calculation of fluxes over ice is thus a part of the land-surface routine.
The heat fluxes are calculated separately for the fraction of the Baltic Sea that is free from ice and for the fraction that is covered with ice. A grid box in this version of RCA1 is given the value 1 or 0 where 1 means 100% ice and 0 means 100% open water. Water on ice in melting periods is calculated by the model as ice. When the model has ice and the grid box value is 1, correct fluxes can only be calculated for ice. The same yields for the fluxes over sea where the model only gives correct fluxes if the grid box value is 0.
In RCA1, the ice temperature is calculated prognostically for the Baltic Sea and the lakes in the Baltic Sea region (Rummukainen et al., 2001). To calculate the ice temperature in RCA1 a three layer model is used based on a one-dimensional equation for vertical heat diffusion with constant heat capacity and diffusivity in time and space (Källén, 1996). The snow thickness on ice is assumed to be 20 % of the ice thickness in the model. The fixed albedo value 0.2 is too low which gives a too early spring melt in the model according to analysis of the RCA1 runs (Rummukainen et al., 2001).
To compare modelled data with measured data in this study 2 grid points have been chosen of the RCA1 44 km grid. These two different grid points have geographical coordinates as close to those of the meteorological stations (Aranda and Umeå) as possible. Table 1 shows the geographical position of the measuring sites with corresponding model grid points and the fraction of land in each grid point.
Table 1. Geographical position of the measuring sites with corresponding RCA1 grid points.
Measuring site Geographical pos. RCA1 grid point Fraction of land
R/V Aranda 63.08 N, 21.14 E 63.17 N, 21.23 E 0.19
Umeå 63.45 N, 20.24 E 63.47 N, 20.18 E 0.28
The model gives vertical profiles in pressure coordinates and the pressure is converted into height by equation 4 (Rindert, 1993).
⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ ⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ − ⋅ = ⋅ g R V V V d p p T z / 0 0 1 γ γ (4)
All variables except for p0 and Tvo are estimated according to the standard atmosphere where Rd=287.06 J/kg·K, g=9.82 m/s2 and γv=0.65 °C/100m. Instead of the virtual temperature (usually 273.15°C in the standard atmosphere), the temperature at 2 m height is used. The same yields for p0, where instead of using pressure at sea level (1013.25 hPa in the standard atmosphere) model pressure at the ground is used.
3. Sites and measurements
Measurements are from the BASIS field campaign in the winter 1997/1998. The aim of BASIS was to improve coupled atmosphere-ice-ocean models and to create and analyse an experimental data set for further understanding of energy- and water cycles during winter conditions. Meteorological and hydrological data was gathered together with sea ice data. The data was then used to investigate the physical processes of energy and mass transfer between the atmosphere and the sea ice and between the atmosphere and the sea.
The main field campaign was carried out in the Bothnian Bay in February-March 1998. Meteorological measurements were made at six different sites in the Bothnian Bay. In this study, data from two of the six stations have been used. The first measuring site was located at Lövudden (63° 40.5’ N, 20° 24.0’ E), a small peninsula about 25 km south of the Swedish town Umeå at the Swedish east coast (Carlsson, 2000). The measuring period in Umeå was 14 February to 6 March (Julian day 45-65). The second measuring site used here was the ship R/V Aranda anchored in the sea ice outside the Finnish west coast (63° 08.12’ N, 21° 14.66’ E) near the Finnish town Vaasa. This data set is from 18 February to 7 March (Julian day 49-66).
Figure 1. Map of the measuring area during BASIS showing the two measuring sites Umeå and R/V Aranda.
3.1 Umeå
The mast at the Umeå site was 12 m high and placed about 20 km north of the ice edge. It was placed on the ice a few meters from the coastline and 20-30 m from the forest. The mast was equipped with instruments for temperature, wind and turbulence measurements. Wind and temperature where measured at heights 1, 3.5 and 11 m. The temperature at 2 m is calculated from the potential temperature at 3.5 and 1 m.
Temperature is assumed to vary linearly with height between 1 and 3.5 m and calculated by equation 5:
(
)
m m m m m T T T T z z T T 3.5 1 1 1 2 3.5−1 ⋅ 2−1 + − = + ∆ ⋅ ∆ ∆ = (5)A sonic anemometer at 10 m height measured the turbulent fluxes of heat and momentum. Radiation was also measured by radiometers. All data have been saved as 10 minute averages but in this study 1 hour averages will be used. The turbulence measurements are high
frequency measurements with sampling rates 20 Hz. Because of too large uncertainties in the measurement, wind speeds below 2 m/s are not used.
At the same location radiosonde soundings were made with a Vaisala RS80-15 sonde every sixth hour measuring temperature, humidity and pressure. Piball tracking were also performed every sixth hour and with this technique wind speed and wind direction are measured. For a more detailed description of the measurements at Umeå see Carlsson (2000).
No moisture data was taken during the measuring period so when the energy balance is calculated the latent heat flux is estimated by equation 6.
The Bowen ratio: E H =
β (6) Bowen ratio is the ratio of sensible to latent heat fluxes at the surface (Stull, 1988). This equation shows how dry a surface is. Over a moist surface where most of the energy goes into evaporation, β is close to zero or negative whereas over a dry surface where most of the energy goes into sensible heating β can be a large positive value. Bowen ratio over ice is estimated to 1 by dividing the sensible heat flux with the latent heat flux over ice calculated by the model. Modelled data contains both latent heat flux over ice and sensible heat flux over ice. The sensible heat flux over ice is simply divided by the latent heat flux and then a mean value is taken from the result. This gives a very rough value of β=1 over ice.
3.2 R/V
Aranda
The Finnish research icebreaker Aranda was anchored in the sea ice outside the Finnish West Coast. A 10 m high mast was placed about 300 m northwest of the ship (Brümmer et al., 2002). On this mast temperature, humidity, wind speed and wind direction were measured at different heights. The turbulent fluxes of heat and momentum were measured at a separate mast located 40 m from the other mast. Measurements were made with a sonic anemometer at 2.2 m height and calculated for 10 minutes intervals. This particular data set is used in this study and all parameters are measured by the sonic anemometer at 2.2 m height and transformed to one hour averages. The turbulence measurements are high frequency
measurements with sampling rates 20 Hz. Wind speeds below 2 m/s are not used because of too large uncertainties in measurements.
4. Description of the measuring period
4.1 Ice
conditions
The Baltic Sea is every winter more or less covered with ice. The winter of 1997/1998 was mild and in February-March only the Bothnian Bay, part of the Gulf of Finland and a small zone at the Finnish coast were covered with ice (Brümmer et al., 2002). The ice cover in the Baltic Sea was poor and unstationary, especially in the two first weeks of the BASIS
experiment (BASIS, 2001) and sometimes the ice was drifting fast. The poor ice cover especially at the Umeå site was a problem in this study.
There was a small island in sector 130°-195° which is not assumed to disturb the
measurements but apart from this the measuring site at Umeå had an open sea/ice fetch in the sector 45°-225°. In the sector 45°-135° the ice was thick, so for these wind directions solid ice conditions can be assumed. There was a 60 km ice fetch in the sector 135°-225°. Except for this ice fetch it is hard to say where and how much ice there really was. The ice in this sector was fast ice, open ice or very open ice and sometimes even open water. In sector 225°-45° there was woodland. This sector is not investigated and the data has been eliminated from the data set. Figure 2 shows the different wind sectors at the Umeå site.
Figure 2. Wind sectors at the Umeå measuring site. Sector 1 (45°-135°) has solid ice, sector 2 (135°-225°) various ice conditions and sector 3 (225°-45°) woodland.
The ice cover around R/V Aranda was more extended with mostly fast ice. The measuring site had an open ice fetch long enough to consider all wind directions to represent undisturbed ice conditions.
A problem that complicates this study even more is that the model doesn’t always have an ice fetch when the measurements have and vice versa. This is seen in figure 3 and 4 which show observed ice concentration and modelled ice concentration 23rd of February 1998.
Figure 3. The map shows the sea ice conditions according to a routine ice chart of FIMR (23 February, 1998) with some of the BASIS measuring sites marked (Launiainen and Vihma, 2001). Circles with cross show the BASIS-meteorological stations. R/V Aranda is marked with a cross and Falcon airbase with an F.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Figure 4. The ice concentration in the model at 12 UTC 23 February 1998 for the Umeå grid point. Ice=1 and no ice=0.
As mentioned in the model description (section 2.2) a grid box in the model has either the value 1 or 0 where 1 means 100 % ice or 0 means 100 % open water. Areas with circles in figure 3 illustrate very open ice, areas with diagonal lines illustrate fast ice and areas with small vertical lines illustrate open ice (more info. see square in figure 3). This shows that the ice is mostly fast around R/V Aranda while there is some very open ice around the Umeå site this day.
RCA1 have mostly ice during the measuring period for both Aranda and Umeå grid points. For the Aranda grid point, model has ice except in Julian day 52.25-55 and 58.25-59.25. For the Umeå grid point, RCA has ice in periods 45-50, 51-52, 55-57.75 and 59.75-65. There are some small periods with sea in the ice periods and vice versa. For example period 50.5-50.75 shows open sea while the rest of the period 49-52.25 is ice for the Aranda grid point. These few values have been removed from the data set to make the already complex ice situation easier to analyse.
4.2 Weather
Conditions
The winter 1997/1998 was a mild one with variations both in the snow and ice thickness. There were frequently strong winds and large changes in temperature, especially for the first two weeks of the field experiment. The weather was not steady for much longer than one day and lows and highs, troughs and ridges as well as cold and warm fronts passed the area (Brümmer et al., 2002).
The temperature changed rapidly some days together with large changes in wind speed. These changes are not unusual because a large-scale north-south temperature gradient is always present in north eastern Scandinavia in winter (Brümmer et al., 2002). The strongest temperature drops occurred on 16, 19 and 23 February (Julian day 47, 50 and 54). The temperature dropped with up to 5°C on one hour. This happened after passing low-pressure systems. The highest wind speed, 18 m/s, were measured at Aranda after on of these low-pressure systems that passed 23rd of February.
5. Results
5.1 Mean
Parameters
In figure 5 to figure 8 measured temperatures and wind speeds at Umeå and Aranda are shown compared to modelled temperatures and wind speeds.
45 47 49 51 53 55 57 59 61 63 65 −20 −15 −10 −5 0 5 10 Julian Day 1998 Temperature ( ° C)
ice sea ice sea ice sea ice
temp. sector 1 Umeå temp. sector 2 Umeå temp. over ice RCA1 temp. over sea RCA1
Figure 5. Modelled temperature (thick and thin lines) compared with measured temperature (dots and circles) at 2m height. Thick lines show calculated temperature over ice by RCA and thin lines show the temperature over sea. Dots show measured temperature in sector 1 and circles show the temperature in sector 2 at the Umeå site (see section 4.1). Vertical broken lines show the different periods in the model with ice or open sea (see section 4.1). Figures 5 and 7 show that the model underestimates the temperature at both Aranda and Umeå. The model temperature is generally too low when T>0°C. For T<0°C modelled temperatures agree better with measured temperatures but modelled temperatures are generally lower here as well. In figures 5 and 7 measurements show a higher temperature in days 49, 51 and 57 than modelled temperatures and this is probably due to warm fronts that passed the measuring area these days. Both modelled and measured temperature dropped rapidly in period 54-56 due to cold snowstorms (Launiainen et al., 2001).
RCA generally underestimates the wind speed (see figure 6) especially when measured wind speeds are high. Modelled wind speeds agree better with measured wind speed at the Aranda site (see figure 8) but even here RCA generally presents too low wind speeds compared to measurements.
45 47 49 51 53 55 57 59 61 63 65 0 2 4 6 8 10 12 14 16 18 20 Julian Day 1998 Wind speed (m/s)
ice sea ice sea ice sea ice
wind sector 1 Umeå wind sector 2 Umeå wind over ice RCA1 wind over sea RCA1
Figure 6. Modelled wind speed (thick and thin lines) compared with measured (sonic) wind speed (dots and circles) at 10 m height. Symbols as in figure 5.
50 52 54 56 58 60 62 64 66 −20 −15 −10 −5 0 5 Julian Day 1998 Temperature ( ° C)
ice sea ice sea ice
temp. over ice Aranda temp. over ice RCA1 temp.over sea RCA1
Figure 7. Modelled temperature at 2 m (thick and thin lines) compared with measured temperature (dots) at 2.2 m at Aranda. Vertical broken lines show the different periods in the model with ice or open sea.
50 52 54 56 58 60 62 64 66 0 2 4 6 8 10 12 14 16 18 20 Julian Day 1998 Wind speed (m/s)
ice sea ice sea ice
wind over ice Aranda wind over ice RCA1 wind over sea RCA1
Figure 8. Modelled wind speed at 10 m (thick and thin lines) compared with measured wind speed (dots) at 2.2 m height at Aranda. Symbols as in figure 7.
5.2 Turbulent surface fluxes
The turbulent flux of sensible heat (H) can mainly be estimated by the temperature difference between air and surface, wind speed and stratification. Sensible heat fluxes are negative (downward) when the air temperature is higher than the surface temperature and positive (upward) when the air temperature is lower than the surface temperature.
The Baltic Sea is more or less covered with ice in the winter. Stable stratification is typical over an ice cover (Vihma and Brümmer, 2002) and sensible heat fluxes are downward. As long as the sea surface is free from ice the stratification is usually unstable during winter and heat fluxes are upward.
The turbulent flux of momentum (τ) depends mainly on variations in wind speed caused by synoptic variations. The momentum flux also depends on other parameters for example the stability and the roughness length.
5.2.1 The sensible heat flux 50 52 54 56 58 60 62 64 66 −150 −100 −50 0 50 100 Julian Day 1998
Sensible heat flux (W/m
2 )
ice sea ice sea ice
H over ice Aranda H over ice RCA1 H over sea RCA1
Figure 9. Modelled (thick and thin lines) surface fluxes of sensible heat compared with measured fluxes at R/V Aranda (dots). Thick lines show calculated H over ice and thin lines show calculated H over sea. Dots show measured H at Aranda. Vertical broken lines show the different periods in the model with ice or open sea.
The largest difference between the model and measurements at Aranda are shown in Julian days 49, 51 and 57 (see figure 5) after passage of warm fronts. Air temperatures for
measurements were mostly above zero and the ice was melting. Wind speeds were also mostly high. The difference between air temperature and surface temperature is larger for the model than for measurements because the melting ice and warm air advection gives a surface temperature that is lower for the model than for measurements. The model shows extremely large negative sensible heat fluxes these days of about –100 W/m2 or more.
In the later part of the measuring period, 59-66, model calculations agree better with measurements. During these days the temperature was below zero and wind speeds were lower. However, the model seems to give too large positive as well as negative peaks of sensible heat flux also during this period.
The model has open sea in period 52-55 and 58-59 and modelled sensible heat fluxes agree rather well with measured sensible heat fluxes except in the end of day 54. These periods were warm with air temperatures above zero. A ridge, occlusions, a warm front and highs and lows passed the area in a fast sequence (Brümmer et al., 2002) which caused snow melt and melting ponds on the sea ice. The effect of this is that the surface behaves more as water. Measurements show large positive sensible heat fluxes in the end of day 54 when temperature dropped fast on the rear side of a passing low. Modelled temperature also dropped fast but this
happened a little later. Cold snowstorms occurred during days 54-56 (Launiainen et al., 2001) and the model seems to smooth out the highest measured fluxes these days.
45 47 49 51 53 55 57 59 61 63 65 −150 −100 −50 0 50 100 Julian Day 1998
Sensible heat flux (W/m
2 )
ice sea ice sea ice sea ice
H sector 1 Umeå H sector 2 Umeå H over ice RCA1 H over sea RCA1
Figure 10. Modelled (thick and thin lines) surface fluxes of sensible heat compared with measured fluxes at Umeå (dots and circles). Thick lines show calculated H over ice and thin lines show H calculated over sea. Dots show measured H in sector 1 and circles show measured H in sector 2 (see section 4.1). Vertical broken lines show the different periods in the model with ice or open sea.
The data from the Umeå site is more complex due to a more varying (and uncertain) ice-cover. This makes the analysis of these measured fluxes compared to modelled fluxes more complicated. In figure 10 circles show measured sensible heat fluxes in sector 2 (see section 4.1) where there is a 60 km ice fetch. At a further distance there is probably a fractured ice cover with very open ice so circles show either ice or open water. The large scatter depends on the uncertainties in sector 2 (circles) and whether or not there is ice or open sea in this sector. It can be assumed that there is ice conditions in this sector because modelled sensible heat fluxes and measured sensible heat fluxes agree well in some periods for example day 55-57. In day 63 the model gives mostly a smaller sensible heat flux than measurements due to lower modelled temperatures.
The data from figures 9 and 10 are shown as scatter-plots in figure 11. This figure illustrates how large the scatter is, how well modelled and measured fluxes are correlated and how much the model underestimates/overestimates the sensible heat flux. RCA1 are limited to the same wind directions as measurements to be sure that the ice fetch is the same in the comparisons.
−150 −100 −50 0 50 100 −150 −100 −50 0 50 100 H RCA1 (W/m 2 ) H Ar anda (W/m 2 ) −150 −100 −50 0 50 100 −150 −100 −50 0 50 100 H RCA1 (W/m 2 ) H Um eå (W/m 2 ) a) b)
Figure 11. Scatter-plots for the sensible heat flux when model and measurements have ice. In figure a stars show measured H in wind sector 2 (see section 4.1) at Umeå plotted against calculated H by RCA1 and dots show sector 1 plotted against RCA1. In figure b dots show measured H at Aranda plotted against calculated H by RCA1.
When model and measurements have the same ice fetch, RCA1 generally gives too large negative (downward) heat fluxes. The model gives on the average about 16 W/m2 too large negative sensible heat fluxes compared to measurements at the Aranda site. When comparing measured sensible heat fluxes at the Umeå site with modelled sensible heat fluxes the model also gives too large negative heat fluxes by 12 W/m2.
Results from scatter-plots of temperature, wind speed, sensible heat flux and momentum flux have been put together in tables 2 and 3. Mean absolute values for the sensible heat flux are included to see if the model overestimates the magnitude of the sensible heat flux.
Table 2. Mean values and correlation coefficients when both model and measurements (Umeå) have ice fetch. Values in brackets show the mean absolute value.
Parameters Mean value Umeå Mean value RCA1 Correlation-coefficient
Temperature (°C) -6.80 -9.34 0.31
Wind speed (m/s) 9.04 6.24 0.68
H (W/m2) -6.87 (19.62) -19.26 (23.72) 0.49
τ (N/m2) 0.102 0.128 0.12
Table 3. Mean values and correlation coefficients when both model and measurements (Aranda) have ice fetch. Values in brackets show the mean absolute value.
Parameters Mean value Aranda Mean value RCA1
The results from measurements at Aranda are easier to interpret than the results from Umeå due to the complicated ice situation at the Umeå site. Aranda has a larger number of
measurements over solid ice. RCA underestimates the temperature by about 3ºC compared to measurements at the Umeå site. Measured temperature and modelled temperature are poorly correlated (correlation coefficient 0.31) and the amounts of data in this scatter-plot are not enough to get a reliable result. According to measurements at Aranda the model presents in general temperatures that are 1ºC too low. Here both modelled and measured temperature is good correlated with correlation coefficient 0.88. Measurements from both Aranda and Umeå show that RCA underestimates the wind speed by 0.2 and 2.8 m/s respectively.
5.2.2 The momentum flux
50 52 54 56 58 60 62 64 66 −0.2 0 0.2 0.4 0.6 0.8 Julian Day 1998 Momentum flux (N/m 2 )
ice sea ice sea ice
τ over ice Aranda τ over ice RCA1 τ over sea RCA1
Figure 12. Modelled (lines) surface fluxes of momentum compared with measured fluxes at R/V Aranda (dots). Thick lines show calculated τ over ice and thin lines show τ over sea. Dots show measured τ at Aranda. Vertical broken lines show the different periods in the model with ice or open sea.
Figure 12 shows the measured and modelled momentum flux from the Aranda site. During days 57, 63 and 64 the largest difference between modelled and measured momentum fluxes at Aranda is 0.2 N/m2. RCA1 shows larger momentum fluxes these days compared to
measurements due to higher wind speeds.
There are also large differences between modelled momentum fluxes and measured
momentum fluxes day 49 and 51. As mentioned before warm fronts passed the measuring site these days. RCA has larger momentum fluxes these days probably due to the melting
value for measurements changes because of the melting ice and the stratification while the roughness length for the model is still the same (and larger) and this is why the model overestimates the momentum fluxes these days.
As for the sensible heat flux modelled momentum fluxes and measured momentum fluxes also agree rather well in period 52-55 (where the model has open sea conditions) except in the later part. The high measured values of τ in the end of day 54 are due to the passing cold front. Although wind speeds were almost the same on day 54 during the cold front passage and on day 56 during the warm front passage (Brümmer et al., 2002) the momentum fluxes were higher under the unstable stratification in the cold front (day 54).
Figure 13 show the measured and modelled momentum flux from the Umeå site.
45 47 49 51 53 55 57 59 61 63 65 −0.2 0 0.2 0.4 0.6 0.8 Julian Day 1998 Momentum flux (N/m 2 )
ice sea ice sea ice sea ice
τ sector 1 Umeå τ sector 2 Umeå τ over ice RCA1 τ over sea RCA1
Figure 13. Modelled (lines) surface fluxes of momentum compared with measured fluxes at Umeå (dots and circles). Thick lines show calculated τ over ice and thin lines show τ over sea. Dots show measured τ in sector 1 and circles show measured τ in sector 2 (see section 4.1). Vertical broken lines show the different periods in the model with ice or open sea.
In day 63 RCA gives higher momentum fluxes than measurements and the difference between modelled momentum fluxes and measured momentum fluxes is 0.3-0.4 N/m2. Measured wind speeds are mostly larger than modelled wind speeds this day so the larger modelled
momentum flux is probably due to the stability and/or the roughness length. In period 55-57 momentum fluxes calculated by the model agree well with measured momentum fluxes.
The measured and modelled data from figure 12 and 13 are illustrated as scatter-plots in figure 14. This figure shows how large the scatter is, how well modelled and measured momentum fluxes are correlated and how much the model underestimates/overestimates the momentum flux. 0 0.2 0.4 0.6 −0.1 0 0.1 0.2 0.3 0.4 0.5 τRCA1 (N/m2) τ Ar anda (N/m 2 ) 0 0.2 0.4 0.6 −0.1 0 0.1 0.2 0.3 0.4 0.5 τRCA1 (N/m2) τ Um eå (N/m 2 ) a) b)
Figure 14. Scatter-plots for the momentum flux when model and measurements have ice. In figure a stars show measured τ in wind sector 2 (see section 4.1) at Umeå plotted against calculated τ by RCA1 and dots show sector 1 plotted against RCA1. In figure b dots show measured τ at Aranda plotted against calculated τ by RCA1.
Momentum fluxes (τ) calculated by the model are generally too high over ice compared to measurements. The model overestimates the momentum flux by about 0.03 N/m2 compared to measured momentum fluxes at the Umeå site. According to the time series (figures 12 and 13), the low value of the correlation coefficient (0.12) and the small amount of data in this scatter-plot (see figure 14.a) RCA1 probably overestimates the momentum flux by more than 0.03 N/m2. RCA1 gives 0.1 N/m2 to high momentum fluxes in general compared to Aranda measurements in spite of too low wind speeds. This indicates a wrong roughness length.
5.3 Net radiation and Energy balance
5.3.1 Net short- and longwave radiation
Longwave and shortwave radiation both downward and upward were measured during BASIS (Cheng et al., 2001). Radiation data is missing for some days due to bad weather conditions. The data set from Umeå only contains radiation measurements in period 48-57 and 62-65. The longwave upward radiation (L↑) measurements at Umeå are incorrect. The measurements are too low, thus L↑ have been corrected by +15 W/m2 (Brümmer et al., 2002).
Figure 15 shows net shortwave and net longwave radiation from the Umeå site.
45 47 49 51 53 55 57 59 61 63 65 −250 −200 −150 −100 −50 0 50 SW (W/m 2 ) a) SW RCA1 SW Umeå 45 47 49 51 53 55 57 59 61 63 65 −25 0 25 50 75 100 125 Julian Day 1998 LW (W/m 2 )
ice s i sea ice sea ice
b) LW RCA1
LW Umeå
Figure 15. Modelled net shortwave (a) and longwave (b) radiation compared with measured net shortwave and longwave radiation at Umeå. Vertical broken lines show the different periods in the model with ice or open sea.
The model gives in general too large negative shortwave and positive longwave net radiation. RCA1 overestimates the net (negative) shortwave radiation by 10 W/m2 and overestimates the net longwave radiation by 7 W/m2. Figure 16 illustrates this.
−300 −200 −100 0 100 −300 −200 −100 0 100 SW RCA1 (W/m 2 ) SW UM EÅ (W/m 2 ) −50 0 50 100 150 −50 0 50 100 150 LW RCA1 (W/m 2 ) LW UM EÅ (W/m 2 ) a) b)
Figure 16. Scatter-plots for net shortwave radiation (a) and net longwave radiation (b) over ice. Dots show measured net short- and longwave radiation at Umeå plotted against calculated net short- and longwave radiation by RCA1.
5.3.2 Energy balance
The energy balance at the surface is the sum of all turbulent fluxes and the radiation. They add up to the ground heat flux (directed down to/up from to the ice) Ground heat flux is calculated by G = H + E + SW + LW. This is done for both measurements and model to see how well RCA1 describes this balance. The latent heat flux is not measured directly, but is estimated by the Bowen ratio (see section 3.1).
Figure 17 shows the modelled and measured energy balance from the Umeå site.
48 49 50 51 52 53 54 55 56 57 −200 −100 0 100 200 300 W/m 2 Umeå a) H E SW LW 48 49 50 51 52 53 54 55 56 57 −200 −100 0 100 200 300 W/m 2 RCA1 b) H E SW LW 48 49 50 51 52 53 54 55 56 57 −200 −100 0 100 200 300 Julian Day 1998 W/m 2
Ground heat flux
c) G Umeå
G RCA1
Figure 17. The energy balance at the Umeå site (a), modelled energy balance (b) and the ground heat flux (c). Thick lines show net shortwave radiation (SW) and net longwave radiation (LW). Thin lines show the sensible heat flux and dotted line the latent heat flux.
The model overestimates the net shortwave radiation, the net longwave radiation and the sensible heat flux (as is seen from the previous sections).
This gives a significantly too large ground heat flux in the model. However, this result is quite uncertain due to uncertainties in determination of measured latent heat flux by the Bowen ratio method.
5.4 Vertical structure in the atmosphere
The vertical structure in the atmosphere have been analysed at the Umeå site for two periods with different features. One situation with on-ice flow on 27 February 00 UTC-18 UTC and one with off-ice flow on 4 March 00 UTC-18 UTC. Off-ice winds bring cold air (Vihma et al., 2002) from the sea ice over the open sea and on-ice winds bring warm, moist air from the open sea over the ice. These two different air flow situations typically occur in wintertime over the ice edge zone in the Baltic Sea.
Modelled vertical profiles of temperature, humidity and wind speed are compared with vertical profiles from radiosondes and piball tracking at Umeå. Temperature, humidity and pressure are measured with Vaisala RS80-15 sondes every sixth hour. These sondes do not have a GPS (Global Positioning System), so wind speed and wind direction are measured with a piball tracking technique. The piball tracking files contain one to five balloons and a mean value is calculated if there is more than one balloon.
5.4.1 On-ice flow
The wind came from south to southwest and temperatures were above zero. An extensive low-pressure system caused this flow and its centre was near the Lofote Islands (outside the coast of northern Norway). The Bothnian Bay was mostly covered by sea ice during this day (80-100 %) but the ice cover was fractured (Vihma and Brümmer, 2002).
0 5 10 15 0 200 400 600 800 1000 1200 1400 1600 1800 2000 θ (°C) Heigth (m) 27 February 1998 Umeå 20 60 100 0 200 400 600 800 1000 1200 1400 1600 1800 2000 RH (%) 5 15 25 0 200 400 600 800 1000 1200 1400 1600 1800 2000 U (m/s) 150 200 250 300 0 200 400 600 800 1000 1200 1400 1600 1800 2000 DD (°) 00 06 12 18
Figure 18. Vertical profiles of potential temperature (θ), relative humidity (RH), wind speed (U) and wind direction (DD) measured by radiosondes at Umeå on 27 February 1998. The figure shows profiles at 00 UTC (thin solid line), 06 UTC (dotted line), 12 UTC (thick solid line) and 18 UTC (dashed line).
0 5 10 15 0 200 400 600 800 1000 1200 1400 1600 1800 2000 θ (°C) Height (m) 27 February 1998 RCA1 20 60 100 0 200 400 600 800 1000 1200 1400 1600 1800 2000 RH (%) 5 15 25 0 200 400 600 800 1000 1200 1400 1600 1800 2000 U (m/s) 150 200 250 300 0 200 400 600 800 1000 1200 1400 1600 1800 2000 DD (°) 00 06 12 18
Figure 19. Modelled vertical profiles of potential temperature (θ), relative humidity (RH), wind speed (U) and wind direction (DD) on 27 February 1998. Lines as in figure 18.
Figures 18 to 19 show measured and modelled vertical profiles of potential temperature, relative humidity, wind speed and wind direction. The profiles of wind speed and wind direction at 00 UTC are not shown in figure 18 because measurements are probably incorrect. Measurements show (see figure 18) one inversion at about 200 m. There is a moist layer below the inversion and a significantly dryer layer above. This inversion is rising to about 300 m and smooth out during the day while the moist layer (near ground) rises and becomes moister. Another inversion is shown at 600 m and this layer is also seen in the relative humidity profiles. The height of this inversion also increases during the day. A low-level jet (with maximum wind speed around 25 m/s) was observed at the top of the stable layer at 200 m. This was also seen by the Falcon research aircraft near Aranda at 250 m in the top of the stable boundary layer (Vihma and Brümmer, 2002). This low-level jet is developed at 06 UTC but disappear later in the day.
The model has a tendency of an inversion up to about 400 m where also the wind is increasing significant. This weak inversion is also smoothing out during the day and the relative
humidity is increasing about the same way as for measurements. Temperature profiles don’t change much during the day. Modelled temperature profiles show some similarities to measured profiles but are much smoother and at 12 UTC the profiles are very similar after 800 m. The low-level jet is not reproduced correctly in RCA but there is a wind maximum at about 500-900 m. The model has a wind speed around 10 m/s that is almost constant the first 200 meters. The profile of wind speed does agree well with measurements after about 400 meters.
5.4.2 Off-ice flow
A low-pressure system has passed over the Bothnian Bay on 4 March and cold air advection was developed on the rear side of it. Winds were north-easterly and temperatures mostly -10°C to -5ºC. −100 −5 0 5 200 400 600 800 1000 1200 1400 1600 1800 2000 T (°C) Height (m) 4 March 1998 Umeå 50 80 110 0 200 400 600 800 1000 1200 1400 1600 1800 2000 RH (%) 5 10 15 20 0 200 400 600 800 1000 1200 1400 1600 1800 2000 U (m/s) −50 0 50 100 0 200 400 600 800 1000 1200 1400 1600 1800 2000 DD (°) 00 12 18
Figure 20. Vertical profiles of potential temperature (θ), relative humidity (RH), wind speed (U) and wind direction (DD) measured by radiosondes at Umeå on 4 March 1998. The figure shows profiles at 00 UTC (thin solid line), 12 UTC (thick solid line) and 18 UTC (dashed line).
Figure 20 shows measured vertical profiles of temperature, relative humidity, wind speed and wind direction for the off-ice flow situation.
Unfortunately the wind speed and the wind direction profiles only reach 400 m. This is probably due to cloudiness because the relative humidity is 90-100 % at that height. Because the lack of a GPS, wind speed and wind direction cannot be measured after a balloon has disappeared into clouds. Measurements show an inversion layer at about 400 m which rises to about 1000 m later in the day. This layer is seen even at 18 UTC when it has reached 1200 m. It is moist during most of the day and the relative humidity is increasing to about 400 m in the morning where there probably are clouds.
Modelled vertical profiles of temperature, relative humidity, wind speed and wind direction for the off-ice flow situation can be seen in figure 21.
−100 −5 0 5 200 400 600 800 1000 1200 1400 1600 1800 2000 θ (°C) Height (m) 4 March 1998 RCA1 50 80 110 0 200 400 600 800 1000 1200 1400 1600 1800 2000 RH (%) 5 10 15 20 0 200 400 600 800 1000 1200 1400 1600 1800 2000 U (m/s) −50 0 50 100 0 200 400 600 800 1000 1200 1400 1600 1800 2000 DD (°) 00 12 18
Figure 21. Modelled vertical profiles of potential temperature (θ), relative humidity (RH), wind speed (U) and wind direction (DD) on 4 March 1998. Lines as in figure 20.
Modelled temperature profiles and measured temperature profiles are similar this day but model temperatures are lower and RCA cannot describe the inversion correctly but there is a small tendency. The wind speed is increasing with height and reaches a maximum at the top of the inversion layer. Even this day modelled profiles are smoother than measured profiles, its most clear in the profiles of relative humidity.
One large difference between these two situations with on-ice and off-ice flow is that during the off-ice day the relative humidity is high and temperature is significantly lower. The high relative humidity could be explained by the cold air. In the on-ice flow situation the wind is blowing from the sea over the ice and the warm, moist air is cooling when entering the ice. An internal boundary layer up to about 200 m is built up over the ice. This internal boundary layer rises to about 300 m at 18 UTC due to lower wind speeds. An internal boundary layer is a layer within the boundary layer which is formed when air is advected over a discontinuity in surface properties (Carlsson, 2000).
6. Discussion
The problem in this study is to make sure that RCA1 and measurements have an ice fetch during the comparisons of mean parameters, surface fluxes and vertical profiles. As
mentioned earlier in this paper this is not always the case because sometimes the model shows open sea when measurements shows ice and vice versa. The fact that RCA1 only separates between one ice type and open sea while the ice cover around the measuring sites is varying and fractured makes a comparison difficult. The description of the surface types is a great problem especially during melting conditions, where we have incorrect surface fluxes as well as other parameters in the model.
Climate models do not resolve all details of the interaction between the atmosphere and an inhomogeneous ice surface. RCA1 can only calculate fluxes for either fast, thick ice or open water. Melting ice or fast drifting ice as well as fractured ice can result in incorrect surface fluxes. Water on the ice is calculated by the model as ice. This problem is clearly seen when comparing modelled fluxes with measured fluxes at the Umeå site where the ice cover was fractured and varying with not only one ice type. In an ideal case, to analyse how well the surface fluxes are described in the model during ice conditions, the model and measurements would have the same ice fetch with the same kind of ice types but in reality this is not the case. Another problem in this kind of investigations is that data containing measured fluxes over ice are few so data used in this investigation is rather unique.
In order to represent correct area averages (Brümmer et al., 2002) of heat fluxes in a grid box, the model cannot simply separate between one ice type and water. In reality there are all kinds of ice types present at the same time. Variations between open water and sea ice is often much smaller than the size of a grid box and the surface may change faster in reality than in the model. The parameterization of surface fluxes over a broken ice cover is a problem for most climate models (Vihma et al., BASIS 2001).
Another problem for many climate models is the ice temperature. The ice surface temperature is in some climate models prescribed by climatological values (Brümmer et al., 2002). Due to this, warm and cold air advections are in most cases not accounted for properly and the model presents incorrect surface fluxes. In RCA1 the ice temperature depends roughly on the air temperature and a climatological temperature and can to a certain extent account for warm and cold air advection. Many models of this kind also have problems with the stable stratification during winter conditions and to properly account for this.
7. Summary and Conclusions
The model gives too large negative (downward) sensible heat fluxes over ice compared to measurements at both Aranda and Umeå sites. RCA1 overestimates on the average the negative sensible heat flux over ice by 16 W/m2. The largest difference between modelled sensible heat fluxes and measured sensible heat fluxes are seen after passage of warm fronts due to warm air advection and melting ice conditions. Melting ice can lead to incorrect
modelled fluxes over ice due to larger temperature difference in the model between the air and the surface and to the fact that the model calculates water on ice in melting periods as ice. RCA can only calculate correct fluxes for either ice or open sea and the surface variations are often smaller and changes faster in reality than the size of a grid box.
The momentum fluxes over ice are also overestimated by the model and RCA1 gives in general 0.1 N/m2 too high values in spite of the underestimated wind speeds. The roughness length is probably too large in the model for the ice conditions in this study and this explains why the model overestimates the momentum flux although the wind speed is underestimated. The roughness lengths for momentum (z0M) and heat (z0H) are the same in RCA1 as
mentioned in the model description (section 2.2) but to get more correct heat fluxes as well as momentum fluxes the roughness length for heat should probably have another value not equal to the roughness length for momentum. The assumption that the roughness lengths for heat and momentum are equal is therefore quite unrealistic.
Modelled sensible heat fluxes and momentum fluxes agree rather well with measured fluxes at Aranda when the model shows open sea. There are some uncertainties in the comparison of modelled fluxes and measured fluxes at Umeå because of the variations in ice cover and that some wind directions are disturbed. The main conclusion concerning surface fluxes is that the model needs various ice types to better describe different ice conditions (for example, melting ice, fast ice, fractured ice, very rough ice).
RCA1 underestimates in general the wind speed over ice by 0.2-3 m/s and the temperature by 1°C. Modelled net radiation both downward shortwave and upward longwave are too large over ice. Net shortwave radiation is on the average 10 W/m2 too large and net longwave radiation is 7 W/m2 too large compared to measurements.
Modelled vertical profiles of temperature, humidity and wind speed show generally less variation under 24 hours compared to measured profiles. RCA1 also generally tend to smooth out the profiles due to too low vertical resolution. Temperature profiles are those who agree best with measurements. The model cannot produce scale inversion heights or small-scale inversion layers. So inversions especially those near the ground are almost never described correctly by the model but sometimes at a higher level there can be a tendency to a temperature increase. A small-scale phenomenon like low-level jets is not produced correctly in the model but there are tendencies to wind maximum. This is not surprising since the vertical resolution in the model is too coarse for these kinds of phenomena.
One of the main differences between the analysed off-ice and on-ice flow situations was that for the on-ice situation an internal boundary layer was built up over the ice. The relative humidity was higher in the off-ice day due to lower air temperatures.
The data set from Umeå contains unfortunately relatively few measurements over solid ice; the ice fetch is too short. For further analysis of RCA1 and how well different parameters are described by the model under ice conditions it would be interesting to include more data. For example from the other measuring sites in the BASIS project (Vihma and Brümmer, 2002). In forthcoming studies another interesting thing would be to examine how well the fluxes and other parameters are described by RCA1 over open sea.
8. Acknowledgments
First of all I would like to thank my supervisor Anna Rutgersson for all help and
encouragement. I would also like to thank Timo Vihma and Jouko Launiainen for access to the BASIS data from Aranda. I am also grateful to Markku Rummukainen and Ralf Döscher for access to the model results from the RCA model. Thanks also to all people at MIUU and to all my fellow students for making my time at Uppsala University memorable.
9. References
Brümmer B, Schröder D, Launiainen J, Vihma T, Smedman A-S, Magnusson M, (2002). Temporal and spatial variability of surface fluxes over the ice edge zone in the northern Baltic Sea. Journal of geophysical research, vol. 107, no. C8, 10.1029/2001JC000884, 2002.
Rummukainen M, Räisänen J, Bringfelt B, Ullerstig A, Omstedt A, Willén U, Hansson U, Jones C (2001). A regional climate model for northern Europe: model description and results from the downscaling of two GCM control simulations. Climate Dynamics (2001) 17: 339-359.
Stull R (1988). `An introduction to boundary layer meteorology´., Kluwer Academic Publishers, 666 pp.
Cheng B, Launiainen J, Vihma T, Uotila J (2001). Modelling sea-ice thermodynamics in BALTEX-BASIS. Annals of Glaciology 33 2001 page 243-247.
Launiainen J, Cheng B, Uotila J, Vihma T (2001). Turbulent surfaces fluxes and air-ice coupling in the Baltic Air-Sea-Ice Study (BASIS). Annals of Glaciology 33 2001 page 237-242.
Carlsson M (2000). ‘The stable boundary layer over the ice covered Bothnian Bay’. Uppsala University, Dept. of Meteorology, Sweden.
Undén P et al. (2002). HIRLAM-5 Scientific Documentation Dec. 2002. SMHI Norrköping, Sweden, 119pp.
Rindert B (1993). ‘Termodynamik, hydrostatik och molnfysik’. Uppsala University, Dept. of Meteorology, Sweden, 147pp.
Vihma T, Brümmer B (2002). Observations and modelling of the on-ice and off-ice air flow over the Northern Baltic Sea. Boundary-Layer Meteorology 103: 1-27, 2002.
Vihma T, Schröder D, Kerber A, Brümmer B. On the parameterization of turbulent surface fluxes over broken sea ice. International BALTEX Secretariat, publication no. 19, May 2001, page 82.
Vihma T and Launianen J. BALTEX-BASIS Final Report MAS3-CT97-0117 Contract with EC. International BALTEX Secretariat, publication no. 19, May 2001
Picture at the front page is taken from BALTEX home page
(http://www.gkss.de/baltex/baltex_frame_builder.html), May 2004.
