In the meantime, the BELARE 2010-2011 campaign has come to an end. The last team left on 24 February Utsteinen. They prepared the station for the winter conditions. Two of my instruments, the aethalometer and the TEOM-FDMS were left in operation mode in the southern scientific shelter. This is kind of an exciting test. First, there is no communication connection to this shelter yet, so we do not know really what is happening in there and if the two instruments are working. Second, aerosol measurements like this in Antarctica are usually done at locations where year-round maintenance is possible. Third, everybody is hoping that the power will be available the whole winter through. Taking all this into account, it will be a unique dataset if everything works out well and if we can retrieve the winter data beginning of next summer season.
A very good estimation if everything is ok is the knowledge about the power management system and if other instruments at the station, which are already connected to the Internet, are still working. If there is enough power available at the station, there is also power at the southern shelter as it is on the same power priority level. And as the aethalometer and TEOM-FDMS are robust standard monitoring instruments, there is no need for an increased concern. Until today there has not been a problem yet with the power provision at the station. However, as wind has been low at many days, apparently the diesel generators took over during periods of low winds.
At our institute, Andy has written a small programme that creates a so-called Ensemble Prediction System (EPS) forecast for Utsteinen (better: the nearest model grid point). Such an EPS takes meteorological analysis data from the European Centre for Medium-Range Weather Forecast and makes 10-days forecasts (see image above). The specialty of the EPS forecast is, to run the same model 50 times, but each with slightly different initial conditions, representing model uncertainties. The result is a probability distribution for the meteorological forecast values, giving medians, percentiles and extreme values. A narrow distribution for a parameter shows very good confidence in the value, and a broad distribution illustrates higher uncertainty.
We compared these EPS forecasts to measured values from the automatic weather station and it appeared that the tendencies for wind speed and temperature are very good. Wind speed is most often a bit overestimated and temperature mostly too cold – probably due to the unique setting of the station protected by the Sor Rondane Mountains, which cannot be resolved by the model. Although clouds and precipitation are very difficult to simulate, the EPS forecasts are useful the first few days for cloud coverage. Precipitation is most often largely overestimated. When I was at the station, during most predicted snowfall events, precipitation was detected by the instruments of the HYDRANT project able to measure cloud height and falling crystals or droplets. However, this precipitation mostly never reached the ground or it was much less than predicted. This is a phenomenon we want to investigate further, e.g., with additional monitoring instruments or with profile information on temperature, humidity and wind.
A very good estimation if everything is ok is the knowledge about the power management system and if other instruments at the station, which are already connected to the Internet, are still working. If there is enough power available at the station, there is also power at the southern shelter as it is on the same power priority level. And as the aethalometer and TEOM-FDMS are robust standard monitoring instruments, there is no need for an increased concern. Until today there has not been a problem yet with the power provision at the station. However, as wind has been low at many days, apparently the diesel generators took over during periods of low winds.
At our institute, Andy has written a small programme that creates a so-called Ensemble Prediction System (EPS) forecast for Utsteinen (better: the nearest model grid point). Such an EPS takes meteorological analysis data from the European Centre for Medium-Range Weather Forecast and makes 10-days forecasts (see image above). The specialty of the EPS forecast is, to run the same model 50 times, but each with slightly different initial conditions, representing model uncertainties. The result is a probability distribution for the meteorological forecast values, giving medians, percentiles and extreme values. A narrow distribution for a parameter shows very good confidence in the value, and a broad distribution illustrates higher uncertainty.
We compared these EPS forecasts to measured values from the automatic weather station and it appeared that the tendencies for wind speed and temperature are very good. Wind speed is most often a bit overestimated and temperature mostly too cold – probably due to the unique setting of the station protected by the Sor Rondane Mountains, which cannot be resolved by the model. Although clouds and precipitation are very difficult to simulate, the EPS forecasts are useful the first few days for cloud coverage. Precipitation is most often largely overestimated. When I was at the station, during most predicted snowfall events, precipitation was detected by the instruments of the HYDRANT project able to measure cloud height and falling crystals or droplets. However, this precipitation mostly never reached the ground or it was much less than predicted. This is a phenomenon we want to investigate further, e.g., with additional monitoring instruments or with profile information on temperature, humidity and wind.
