A small-size meteorological mast, BEAR (Budget of Energy for Arctic regions) has been developed as a part of a new autonomous buoy for monitoring the sea ice mass balance. BEAR complements observations of the thickness and thermodynamic properties of the ice/snow pack determined by the so-called Ice-T (Ice-Thickness) buoy, giving access to bulk fluxes and energy budget at the surface, using meteorological measurements. The BEAR mast has been tested with success during ten days in April-May 2010 at Ny Alesund, in the Svalbard archipelago (Norway) showing that meteorological data were close to measurements at the same level of the Italian Climate Change Tower (CCT) from the ISAC-CNR. A discussion is undertaken on bulk fluxes determination and uncertainties. Particularly, the strategy to systematically use different relevant fluxes parameterizations is pointed out to explore flux range uncertainty before to analyze energy budget. Net radiation, bulk fluxes and energy budget are estimated using as average 10 minutes, 24 hours and the ten days of the experiment. The observation period was very short, but we observe a spring transition when the net radiation begins to warm the surface while the very small turbulent heat flux cools the surface.
The most conspicuous manifestations of the ongoing climate warming are found in the Arctic, where the multiyear ice decline is larger than expected from climatic scenarios (IPCC Fourth Assessment Report: Climate Change 2007 (AR4)). There is a real need to better document heat and fresh water exchange processes at the ocean-ice-atmosphere interfaces to understand energy budget variability and to improve process parameterization in models.
Recent studies have been concerned by sea ice evolution as an integrative indicator of global warming, and different processes (atmospheric or oceanic) have been analyzed experimentally or using models: see, for example, to quote a few, Sedlar et al. [
The OPTIMISM project (Observing dynamic and thermodynamic Processes involved in The sea Ice Mass balance from In Situ Measurement) is aimed at developing and deploying a small network of automated buoys system, built upon the “Ice-T” buoy prototype providing real-time measurement of sea-ice thickness and fluxes at the interfaces described in Vivier [
The “OPTIMISM” system consists of an ocean buoy to measure the physical properties of the sea and ice (currents, sea water temperature and salinity, sea ice thickness and temperature) and a meteorological mast upon it. The meteorological mast is the BEAR station concept (Budget of Energy for Arctic Region). For the OPTIMISM project, it has been tested during ten days in April-May 2010 at Ny Ålesund (Spitsbergen), in the island of Svalbard archipelago in Norway, to assess its performance under polar conditions.
Spring in Arctic region at the Svalbard latitude corresponds to a transition regime in the radiation fluxes since solar visible radiation becomes more and more intense. If one considers, for example, general conclusions on the Arctic energy budget regimes, (see, e.g., Maykut [
Figure
Sketch showing the evolution of the net radiation and turbulent fluxes along a year in Arctic regions. The ordinate is in arbitrary units to take into account a possible ratio between the different components. Typically, during SHEBA, maximum turbulent warming is close to 20 Wm−2, and 50 Wm−2 is the maximum infrared cooling during winter. Adopted from Maykut [
In the next section, we describe the BEAR instrumentation and report first comparisons with other measurements close to the BEAR location around Ny Ålesund. Then, we present projected analysis methods to estimate energy budget and bulk fluxes. In the following section, we examine this first BEAR dataset, by analyzing heat fluxes, total net radiation and energy budget observations during this short period of April-May. The conclusion discusses the measurement strategy taking into account the precision relative to the chosen measurement method.
The system has been designed to work in Arctic regions and uses available technology for cold environments. Wind speed and azimuth are obtained from a Young 05103-45 anemometer (instantaneous wind speed precision ±0.3 ms−1 and azimuth precision ±5.0 dg) designed for mountain and for polar regions. It is supposed to be less sensitive to frost than other wind measurement devices. For shade temperature and relative humidity, HMP 155 from Vaïsala is used with a naturally ventilated shade. The temperature and humidity instantaneous precisions are ±0.15°C and ±2%, respectively. The four radiation components (global shortwave and infrared radiations) are obtained from a CNR4 system (with two upward and downward pyranometers and pyrgeometers) manufactured by Kipp and Zonen. The spectral range for shortwave radiation is between 300 and 2800 nm while infrared radiation is between 4500 nm, and 42000 nm. The expected accuracy for daily total net shortwave and net infrared net radiation is close to 10%. A warming of the infrared pyrgeometer domes is performed to prevent from ice formation which demands substantial energy. A battery to be lodged in the buoy is dedicated to heating and is loaded with a Forgen 500 wind turbine whose rotation starts for a wind speed greater than about 5 m/s. The domes temperature is corrected for the infrared measurements.
The instrumentation is implemented on a 3 kg mast, and the whole system weight is lower than 9 kg. During the Ny Ålesund validation experiment, the BEAR system was fixed on a gauge of the Ice-T buoy. Note that the BEAR station needs ancillary data such as surface temperature, ice and snow depth, snow temperature, atmospheric pressure, provided by Ice-T system in the final configuration. During the Ny-Ålesund experiment, however, additional sensors were deployed around the BEAR mast.
Figure
Conceptual scheme of the BEAR station.
Panoramic photograph at 180 degrees of the site, showing the meteorological tower from Italian crew and the OPTIMISM BEAR mast.
Different views of the CCT and the BEAR sites. (a) Picture of the BEAR site. (b) Topography of the CCT site (with BEAR location). The different colored zones correspond to different types of topography. CCT tower and BEAR station are in the same relatively flat terrain which is hachured in the figure. (c) Sketch of the orientation of BEAR and surface temperature measurement (1 to 4) with distances indicated in metres.
This location was chosen by the ISAC-CNR team because it is well exposed to the dominant flow from north-east (along the main fjord orientation), with no building in the vicinity.
Around the site, mountains as high as 600 m are located west of the site (Figure
The BEAR system was installed from April 27 up to May 10, 2010, upon a snow cover. At this period, solar elevation is low and less than 17 degrees at 06:00 U.T. and 25 degrees between 09:00 and 13:00 U.T.
Orientation of the axis between CNR4 and Young has been optimized taking into account shadows effects of the different mountains. Figure
Temperature, humidity, and wind data at 2 m level and radiation measurements at 32 m height from the ISAC-CNR tower were used for comparison with BEAR measurements. We also benefited from AWI data (meteorological and radiation measurements, at 2 m above the surface), see
To distinguish the different stations, we hereafter note FR (FRench) for BEAR data, IT (ITalian) for ITalian tower data and G (German) for AWI data.
We compare here data at comparable levels for meteorological data, and we do not normalize these data at 10 m height because surface layers in these regions are often smaller than 10 m height. For radiation, we compare surface values using FR, IT, and, G, but, due to the 32 m height from IT radiation measurement, we expect some differences due to different footprints and attenuation.
Meteorological instruments used from IT are the same as BEAR instruments for wind temperature and humidity: a Young anemometer for the wind velocity and HMP45 probes for temperature and humidity. For the radiation measurement, IT tower used a CNR1 which is the old version of the CNR4 used for BEAR. For the G station, different instruments were used (a wind vane associated with a Thies system for the wind velocity, a PT100 for temperature, and a HMP 233 for humidity). The G radiation from AWI (Alfred Wegener Institute) is measured from different instruments: CM11 radiometers for shortwave upward and downward radiation and Eppley pyrgeometers for infrared radiation.
Table
Regression of FR basic measurements between G and IT. Bias at the origin, slope and correlation are presented.
Parameter | Regression between FR and G/IT | Bias at the origin | Slope | Correlation coefficient |
---|---|---|---|---|
Mean wind intensity (m/s) | G | 0.82 | 0.82 | 0.74 |
IT | −0.19 | 1.03 | 0.99 | |
Air temperature (°C) | G | −0.13 | 1.00 | 0.98 |
IT | 0.36 | 0.99 | 1.00 | |
Relative humidity (%) | G | 19.2 | 0.77 | 0.86 |
IT | 5.8 | 0.98 | 1.00 | |
Sw | G | −2.2 | 1.00 | 0.98 |
IT | 6 | 1.00 | 1.00 | |
Sw↑ (Wm−2) | G | 10.9 | 0.96 | 0.98 |
IT | 12.8 | 1.03 | 0.99 | |
Lw | G | −11.7 | 1.03 | 1.00 |
IT | −41.3 | 1.1 | 1.00 | |
Lw↑ (Wm−2) | G | −33.4 | 1.09 | 0.98 |
IT | −37.5 | 1.09 | 0.99 |
At least for air temperature, wind speed and air humidity bias are small between FR and IT. For the G measurements, as G corresponds to a different location, the local characteristics are different since it is more characteristic of Ny Alesund “city effect” as can be observed on the wind rose in Figure
Windroses during the experiment from April 28 to May 8 (2010) for FR (in blue), IT (in green), and G (in red). We use conventional meteorological orientation (0 and 360 degrees for wind blowing from the north, 90 degrees for wind blowing from the east).
Figure
Although the wind roses are very similar for FR and IT, we notice a 10-degree difference which is probably due to a misorientation of BEAR. While FR and IT reach 11 m/s close to 210 degrees azimuth sector, the maximum G wind speed is close to 5 m/s. On average, the mean wind speed for G is 20% smaller than for FR and IT.
Regarding the radiation and especially long wave radiation, although the regression slope and correlation coefficients are very good, biases are high. Bias between FR and IT for long wave radiation can be partly explained by the difference of the measurement height (32 m for IT and 1.7 m for G) since IT measurements are not height corrected for upward and downward signal and there is also a different footprint due an imperfect cosine response. Moreover, instruments are different (a CNR1 is used for the tower), and, due to a long duration of use, radiation calibration of the tower can be also affected by some unknown drift.
Figure
Total net radiation scatter plot (FR and IT on the left, FR and G, on the right), from April 28 to May 9, 2010.
A view of net radiation fluxes in Figure
Evolution of net radiation during the experiment (FR, G and IT) from April 28 to May 9, 2010.
The drastic amplitude reduction as of May 4th corresponds to a period with more clouds during which net radiation becomes fully positive.
Estimation of surface temperature using SST (Snow Surface Temperature) inversion from infrared radiation is proposed by Claffey et al. [
This estimation is compared with the four temperatures probes at 1.5 cm below the snow surface cm. Figure
Scatter plot of surface temperature estimated from infrared radiation and the four temperature probes at 1.5 cm below the snow surface for a 0.99 surface emissivity.
The surface temperature can therefore be rather properly estimated using infrared radiation measurements.
The energy budget at the surface can be expressed in a generic fashion, regardless of the surface type (snow, ice, water).
The total net radiation is
The net radiation is also
We take here the meteorological convention in which
Usually,
To compute the surface energy budget, it remains to estimate
We propose to use a simple method to estimate turbulent flux called bulk method. Such methods are usually used in neutral or unstable situations. However, in Arctic regions where the surface temperature (ice or snow) can be lower than the temperature of the lowest atmospheric layers, stably stratified layers are frequently observed and bulk methods must be adapted. Recently, Persson et al. [
Other parameterizations, following, for example, Beljaars and Holtslag [
However, as these parameterizations have been found during different atmospheric contexts and regions, we propose to test these different parameterizations and to analyze the flux range as limits in the flux estimates. It can be noticed that uncertainty on estimated bulk flux is more related to parameterization uncertainty than to the accuracy on mean variable estimates. If parameterizations were indeed really universal, they would be more precise than direct eddy flux measurement (Weill [
The first step to estimate bulk flux at the surface (snow or ice) is to check that the measurement height
All the parameterizations assume that Monin-Obukhov similarity (M.O.) can be applied (see Businger et al. [
The functions for wind, temperature, and humidity, respectively, are said as “universal,” but some differences between authors are observed.
The roughness length
Bulk flux is then derived from (
In the same way are computed
Classical bulk methods generally include iterations on the unknown parameter
The experiment took place during a spring period during which the atmospheric surface layer is generally thin. It is difficult to observe this layer using data from the ISAC-CNR tower since the second level is at 4 m height. During the observation period, several snow falls were observed, but the surface level remains unchanged as shown by the ruler at the foot of BEAR. As radiation net fluxes are positive on average, this means a direct warming of the surface associated with the visible radiation. Moreover, camera observations of the surface show a snow surface structure corresponding to patches of icy snow and dry snow as observed on photograph of Figure
Before analyzing bulk fluxes obtained during the experiment, we find it important to briefly discuss the flux uncertainties due to an imperfect knowledge of
We examine the errors on the neutral transfer coefficient
We first note that a 5 cm error on
If
Figure
Estimates of the sensible heat flux
However, if one computes on average the absolute value of the ratio between
We consider now what kind of flux difference is associated with a different roughness length parameterization. For that purpose, we adopt roughness length from [
We have applied this parameterization to FR, IT, and G data but effects do not depend on the sites. This parameterization on average increases latent heat flux by 24% and sensible heat flux by 31%, (see Figure
Comparisons between BEAR fluxes using a constant roughness length and [
We also took into account effects of thermal roughness length parameterization as suggested by [
We now only consider here BEAR measurements, since we have already analyzed basic uncertainties related to the different meteorological stations in Section
Are discussed here uncertainties which are inherent to the used methods. In the case of April-May 2010, the heat flux is indeed small which justifies that the net radiation is the most important part of the budget: see Figure
Temporal evolution of FR net radiation and turbulent bulk fluxes during the experiment.
Although
The 10-minute average chosen is useful to follow the evolution of atmospheric events and their variability. It will help to discriminate the relationships between atmospheric events and fluxes in relation with ice-snow Ice-T profiles.
To go a little further into the energy budget from a more integrated point of view, we have computed 24-hour average for each day of the experiment.
Figure
Daily averaged FR energy budget during the experiment, with each short-wave and long-wave net radiation components.
We notice that
If we now consider averages over the 10 days of the experiment, we notice that total net radiation reaches 10 Wm−2 on average while
Uncertainties on heat flux during this transition period do not seem a major issue here since
During a short experiment in Ny Ålesund, Svalbard, BEAR station data have been successfully compared with those of the Italian meteorological tower and the AWI German station. This gives some confidence in the use of the BEAR meteorological and radiation data.
We have tested long-wave radiation inversion to get the surface temperature and have found that, during the conditions of the experiment, the surface emissivity of 0.99 chosen by [
We have tested bulk methods from [
Having tested parameterization from [
By analyzing the daily energy, we have remarked that except for periods of low net radiation, we were able to estimate the residual energy flux transmitted to the snow/ice surface. However, as uncertainty can be large, it warrants to be analyzed jointly with the snow/ice information to be delivered by the Ice-T buoy. The low-radiation periods which can occur during winter are important and need to be more documented. The OPTIMISM buoy seems well adapted for this purpose.
Several concluding remarks and outlooks can be drawn from these analyses. In the transition period (see the schema representation in Figure Figure To the question “are we able to analyze the surface energy budget and what can be the precision of estimates,” we think that a 35% or more uncertainty is inherent to the bulk method. However, a simultaneous examination of the energy budget evolution and of the colocated ice thermodynamics, which is the scope of the OPTIMISM project, should at least give redundant information to determine the most probable energy budget. The uncertainty on fluxes will make difficult to obtain the annual energy budget, which has been found to be of a few Wm−2 from SHEBA experiment [
Therefore, we rather suggest to explore events in the surface evolution, looking at the meteorological variables, fluxes (with their uncertainties ranges due to parameterizations), and to analyze these events impacts inside the ice-snow layer using Ice-T buoy. For that purpose, roughness lengths values impacts on radiation components and turbulent fluxes have also to be considered in relation with surface observations, because, as shown by Held et al. [
Another point which warrants to be examined in the future is the frost formation since it can contaminate drastically the measurements. During this campaign, the BEAR station was always operating without rime or frost due to low humidity but frost has been observed during a preliminary test in the French Alps, Loisil et al. [
The OPTIMISM project has received support from ANR (ANR-09 BLANC 022701) and IPEV (programme 1015). It has received incitation support from LEFE (INSU/CNRS). The authors have greatly appreciated fruitful help from ISAC (CNR) and AWI (Germany) for data access during April-May, 2010 in Ny Ålesund. The authors also want to thank very much their reviewers for their very useful remarks which have helped to improve this paper.