Print Version

Chapter IV: Forecast

Table of Contents

Accuracy of the Forecast

The extreme snowfall was partly forecasted by the models. The comparison between the ALADIN model and the ECMWF- analysis should show the discrepancy between the forecasted values within the models and the realities. The actual measurements should verify the estimates of the forecast.

The ALADIN model has a resolution of 9 x 9 km. The ECMWF model has a resolution of 25 x 25 km. For the comparison of the snowfall, a greater extract of the ECMWF for Austria is used. (initial position of both models: Feb, 10th 2010 / 00 UTC)

4.1.1 February, 10th 2010

Figure 4.1.1.1: RR - 6 hourly, 10 February 2010 1200UTC; left: ALADIN, right: ECMWF

According to ALADIN the precipitation in the south of Austria started in the morning of Feb, 10th. ALADIN forecasted between 06 and 12 UTC about 1 cm snowfall. ECMWF calculated values between 0.5 cm and 3 cm, partly at the border between Austria and Slovenia up to 8 cm.

Figure 4.1.1.2: RR - 6 hourly, 10 February 2010 1800UTC; left: ALADIN, right: ECMWF

ALADIN computed within six hours 2 cm and 5 cm, especially more snowfall in some regions in Styria. ECMWF values ranged from 4 cm to 9 cm.

Figure 4.1.1.3: RR - 6 hourly, 11 February 2010 0000UTC; left: ALADIN, right: ECMWF

In the first part of the night, according to ALADIN the snowfall ranged between 2 cm to 11 cm in the south of Styria. ECMWF had similar values.

ALADIN provided about 9 - 10 cm for Graz and 5 - 10 cm for Klagenfurt within 24 hours. The ECMWF forecasted about 12 - 15 cm for Graz and 12 - 17 cm for Klagenfurt.

Figure 4.1.1.4: INCA Meteogram - accumulated precipitation analysis [mm], 10 February 2010 00:00 - 23:45 UTC; left: Graz, right: Klagenfurt

The INCA Meteogram shows that the snowfall in Graz started after 07 UTC and continued the whole day, although it weakened in the afternoon. The almost 24 hours (the last 15 min are missing) snowfall amount accounted between 9 and 10 cm. ALADIN forecasted the amount better than ECMWF, which values were slightly too high.

The snowfall in Klagenfurt began at the same time and lasted the whole day. But this time ECMWF had more exact forecast data than ALADIN. There were measurements up to 13 cm in Klagenfurt within 24 hours. The initiation time of the snowfall was computed correctly by both models.

February, 11th 2010

Figure 4.1.2.1: RR 6 hourly, 11 February 2010 0600UTC; left: ALADIN, right: ECMWF

In the next six hours the snowfall weakened and obtained in both models 2 cm to 7 cm.

Figure 4.1.2.2: RR 6 hourly, 11 February 2010 1200UTC; left: ALADIN, right: ECMWF

Figure 4.1.2.3: RR 6 hourly, 11 February 2010 1800UTC; left: ALADIN, right: ECMWF

Over the day, the snowfall amount decreased to about 2 cm/12 hours in ALADIN and 3 cm/12 hours in ECMWF.

Figure 4.1.2.4: RR 6 hourly, 12 February 2010 0000UTC; left: ALADIN, right: ECMWF

During the night and the next day the snowfall was becoming less. The snowfall amount by ALADIN during Feb, 11th 00 UTC and Feb, 12th 00 UTC amounted 2 - 3 cm in Graz and about 2 cm in Klagenfurt (ECMWF provided values of 4 - 5 cm for both cities).

Figure 4.1.2.5: INCA Meteogram - accumulated precipitation analysis [mm], 11 February 2010 00:00 - 23:45 UTC; left: Graz, right: Klagenfurt

In Graz and Klagenfurt the snowfall lasted the whole night. It stopped in Graz in the early morning hours, but started at 12 UTC again and took about 9 hours till it was finally over. The 24 hours snowfall amount was nearly 4 cm. The ECMWF model computed the more exact values. In Klagenfurt the precipitation lasted until noon. Both models underestimated the snowfall amount for Klagenfurt. There were nearly 7 cm in 24 hours.

Figure 4.1.2.6: Winter service forecast for Graz from 10 February 2010 by ZAMG Styria for 10 - 12 February 2010 and cut-out of snow depth - total/new ( 10 - 11 February 2010) [cm], 11 February 2010 0600UTC

The beginning and the ending of the snowfall was well forecasted. The new snow amount with nearly 20 cm between Feb, 10th 06 UTC and Feb, 11th 06 UTC was a bit overestimated. It accounted 15 cm in Graz. But it is important to mention, that at station in some kilometres away the new snow amount reached values up to 30 cm. This discrepancy shows the difficulty to make a local forecast.

Figure 4.1.2.7: Winter service forecast for Carinthia/Klagenfurt from 10 February 2010 by ZAMG Carinthia for 10 February 2010 and cut-out of snow depth - total/new ( 10 -11 February 2010) [cm], 11 February 2010 0600UTC

Figure 4.1.2.8: Winter service forecast for Carinthia/Klagenfurt from 11 February 2010 by ZAMG Carinthia for 11 February 2010 and cut-out of snow depth - total/new ( 11 - 12 February 2010) [cm], 12 February 2010 0600UTC

The beginning of the snowfall was well forecasted, but it lasted only to the early afternoon on Feb, 11th, not till the night like estimated.

To define the forecast of the new snow amount is a little bit difficult, because there were no snowfall values forecasted between 01 UTC and 07 UTC, so six hours were missing. But especially during the night there was a lot of snowfall. Therefore it is complicated to compare the forecasted snowfall amount to the real values. The forecasted new snowfall amount from Feb, 10th 07 UTC to Feb, 11th 01 UTC ranged between 15 cm to 20 cm for Klagenfurt.

Difficulties of the Forecast

First of all, in spite of everything the models are still too rough in resolution and the Alps are too shallow represented. The models often give a maximum for a region, but if this maximum is located some kilometres away, the whole forecast can be wrong and the precipitation over- or underestimated for this area. Therefore it is very complex to make a local forecast; hence you can only compute a mean for a geographical market. But especially for business, like snow clearance, it is necessary to estimate the exact precipitation amount, because there is always a financial question. The models also have their difficulties to illustrate small valleys in the mountain. Especially in small and large valleys there are often inversions, which the models can't compute. Due to this weather condition fog/low stratus will be formed. Because of the "Seeder-Feeder" mechanism, which the model doesn't consider, the computed precipitation values can be too low. Especially because of high reaching upward slidings, like in the described weather situation from the south, and trapped low level clouds in valleys, locally more precipitation can be produced.

It is also important to look at the local circumstances and topography. Due to the inflow on the Alps orographic precipitation on the windward side of the mountain can be generated and the precipitation can be more than expected.

To make a precipitation forecast the forecasters use different models, like ALADIN, ECMWF or INCA. They are examining if there is a possibility of precipitation. Due to the temperature the snowfall limit and the appearance of the precipitation, as rain or snowfall, can be determined. One possibility is to use the "normal" temperature - values below 2°C indicate snowfall - or the wet-bulb temperature, where values less than 1.5°C are necessary for snow. The difference between the corresponding snowfall limits lies up to 100 m.

Another difficulty is the translation of mm precipitation in cm snowfall. Most of the models compute the precipitation in mm but wanted is the snow amount in cm. The translation depends on temperature, humidity and density. A tough assumption is that 1 mm precipitation is equivalent to 1 cm snowfall. If it is more moist, than the translation 1 mm precipitation is 0.5 cm snowfall can be used.

Figure 4.2.1: Conversion table: mm precipitation in cm snowfall (estimation)

An also important problem is the measurement of snow amount and snow depth. Because of temperature variability the snow becomes drier or more moist, which can influence the snow depth. Due to snow shipping with heavy wind the measured snow amount values can be false.

There are few problems of precipitation forecast, but if the strengths and the weaknesses of the used models are considered and more experience is accumulated by the forecaster, the forcasts become better and better. Also important is to monitor the weather development to be aware of changes thus to modify the forecast.

Warning situation

MeteoAlarm warned Austria of risk of severe weather events (alert level 3) by snowfall and ice from Feb, 10th at 00 UTC to Feb, 11th at 18 UTC. On Feb, 12th 2010 the alert level was decreased to level two - attention to weather activity - and at 18 UTC the warning was cancelled.

Figure 4.3.1: Left: Warning chart for 10 February 2010; Middle: Warning chart for 11 February 2010; Right: Warning chart for 12 February 2010