Pdf-File - The Mesoscale Alpine Programme - MeteoSwiss

Recent developments of VERA
Matthias Ratheiser
Reinhold Steinacker
Manfred Dorninger
Dept. of Meteorology and Geophysics, University of Vienna, Austria
Inga Groehn
Wolfgang Poettschacher
Andreas Schmölz
Christian Häberli MeteoSwiss, Zurich, Switzerland
Introduction
The aim of VERA was to develop an objective, automatic analysis procedure for meteorological
parameters over complex terrain. To resolve mesoscale structures caused by topography as well,
meteorological a priori knowledge (for hand analysis it is the experience of the synoptician) should be
included in the analysis. The available analysis scheme VERA fulfills these demands.
VERA has been developed at the Department of Meteorology at the University of Vienna since 1995.
The scheme may be used for both, error detection and correction, and interpolation of irregulary
distributed data to a regular grid. The emphasis is put on the transfer of information from data sparse
to data rich areas. For this purpose the so called “fingerprint technique” is used. It adjoins additional
orographic information to the measurements. The error detection mode checks the single
measurements concerning their spatial physical plausibility and calculates correction suggestions
where necessary. The method runs independently from any first guess or model field.
The analysis scheme
The basic approach of VERA is a variational algorithm that minimizes the squares of spatial
derivatives of the measured field. It is similar to the spline algorithm, but there are some major
differences. Let us assume a grid field whose points are partly filled with known values. The empty
ones should be filled up by interpolated values. For our purpose there are two postulations for the
interpolation function: First the function must exactly hit all given points. Second we are interested in
the function’s values on the grid points only, but not in the continuos function itself.
If we assume a one dimensional domain for simplicity (see fig.1), these postulations lead to the
minimum criterion for this domain:
values Ψ
2
1
Fig.1: 1D-domain to demonstrate the VERAscheme
0
0 1 2 3 4 5 6 7
spacial axis i
I
6
2
∆ Ψ
2
∆x
=
i=
2 i
2
→ Min
(1)
The ∆ symbols denote the discrete second order
derivatives. The dots in fig.1 denote points, where
observation values are known. (1) is derived for
each unknown value Ψ on the unknown points
and set equal to zero. Now there are as much
equationsasunknowns,hencethesystemmaybe
solved. All postulations are fulfilled: The values at the known points are not changed, so the
interpolation runs exactly through them. The function between the calculated points remains
unknown. These gaps may be filled by simply reducing ∆x. Minimizing the second derivatives means
minimum curvature, so the interpolated points may be connected with a smooth line. It is also
possible to minimize the first order derivatives or the the sum of first and second.
This concept can be easily adapted for a two dimensional field, for example for the analysis of surface
fields (see the applications of the method below). It may be run for vector quantities as well.
The fingerprint technique
The innovation of the fingerprint technique is to include a-priori knowledge to the analysis. It is a fact
that topography (e.g. the Alps) causes mesoscale structures in meteorological fields. There are thermal
(e.g. heat low in the pressure field) and dynamic (e.g blocking) effects. To strengthen these influences
for the analysis, the so called fingerprint techique (a procedure that is common in climatology) is

adapted. Let us assume that a scalar value on a grid point is expressed by the sum Ψo = Ψs +ctΨt +
cdΨd. Ψo is the observed value, Ψs the synoptic value (without topographic influences), Ψt the thermal
influenced part and Ψd the dynamic influenced part of Ψo. The two latter parts are superposed over the
synoptic signal. Synthetic pressure pattern for both the thermal and dynamic effects, so called
idealized fingerprints, are calculated from high resolution digital topography data (see fig.2).
Fig.2: Idealized thermal fingerprint for the Alpine region. Isolines denote relative change of 1
Then, together with the interpolation of the observational data, at every grid point the factors ct and cd
are calculated. They determine how strong the different fingerprints contribute to Ψo. Therefore(1)is
changed to
2
∆ Ψ
2
∆
x
2
→ Min
Ψ = Ψ − c Ψ − c Ψ
(2)
S
with S
T T D D
i i
Refering to the one dimensional example above we now have two additional unknowns, hence two
more variables for the derivations.
Using the fingerprint technique for real 2D-fields the dynamic influence is divided into four idealized
fingerprints and therefore four factors, one for every mean flow direction.
Application examples
The whole analysis scheme consists of an error correction module, an interpolation module and a
fingerprint module. All three parts were extensivly tested and utilized operationally by Austrian
weather services. VERA is also used by the Universities of Vienna and Innsbruck within an exercise
course for weather prediction. The application of VERA for a 3D or 4D domain is still in the
development stage. The scheme is currently running operationally for hourly analyis of surface
pressure, 10m wind vectors, pressure tendency, potential and equvivalent potential temperature, 12hprecipitation
sums and cloud cover. Fig 3 to 4 show selected cases where the abilities of VERA are
demonstrated.

Fig.3: 12h-precipitation sum (coloured areas) an potential level of snowfall (isolines)
Fig.3 shows an example for a heavy precipitation event in the southern Alps on November 6 th, 2000,
18 UTC. It was followed by severe flooding in northern Italy and southern Switzerland. Note the large
area where the values of 12 hour precipitation exeeds 50 mm (red colour). The blocking effect of the
Alp’s main ridge can be seen best in Tyrol and in Salzburg. The sharp gradient is due to the fingerprint
technique. The potential level of snowfall is the height where snow turns into rain depending on the
values of wet bulb temperature. The distribution of potential level of snowfall shows the warm and
moist southwesterly flow that advects the rainfall against the Alps. Note the low levels (down to 600m)
in southern Switzerland. This effect may be due to cooling of the trapped air volume in the valleys by
melting of the heavy precipitation.
Other parameters that are analyzed with the VERA algorithm are cloud cover and lifting condensation
level. See Fig.4 on the next page for an example. This “pseudo-satellite image seen from below the
clouds” shows low ceiling and a window possibly due to föhn over the border area Bavaria/Salzburg.
The lifting condensation level is calculatd from the spread between temperature and dewpoint
temperature at the stations. The values may be interpreted as the lowest condensation level possible
for the measured surface values of dewpoint and temperature. They will be quite accurate if the
surface air is lifted dynamically or on days with good vertical mixing of the atmosphere.
Literature about VERA
Articles in journals
Steinacker, R., W. Pöttschacher, M. Dorninger, 1997: Enhanced Resolution Analysis of the
Atmosphere over the Alps Using the Fingerprint Technique. Annalen der Meteorologie, 35, 235- 237
Steinacker, R., W. Pöttschacher, M. Dorninger, 1997: An Enhanced Resolution Analysis Scheme for
theAtmosphereoverComplexTerrain.Annales Geophysicae, 15,Suppl.II,Europ.Geoph.Soc.,C426
Steinacker et al., 2000: A Transparent Method for the Analysis and Quality Evaluation of Irregularly
Distributed and Noisy Observational Data.Monthly Weather Review Vol.128, No.7, pp 2303-2316

Contributions to the MAP Newsletter
Steinacker, R., M. Dorninger, W. Pöttschacher, 1995: Meteorological applications of high resolution
topographic data. MAP Newsletter, 3, 73-74
Pöttschacher, W, R Steinacker and M Dorninger, 1996: VERA - A high resolution analysis scheme for
the atmosphere over complex terrain. MAP Newsletter, 5, 64-65
Pöttschacher, W, R Steinacker, M Dorninger, 1997: Heavy Snow Fall in the Viennese Region Analysed
with VERA. MAP Newsletter, 7, 32-33
Pöttschacher, W, Ch. Häberli, R Steinacker, 1997: The Error Detection and Correction Module of
CALRAS and VERA. MAP Newsletter, 7, 88-89
Groehn, I., R. Steinacker, Ch. Häberli, W.Pöttschacher, M. Dorninger, 1998: Results of DAQUAMAP:
Characteristics of Humidity Observations of Alpine Synoptic Network, MAP Newsletter NR. 9, p. 54-55
Häberli, Christian and Reinhold Steinacker, 1998: Quality Assessment of Upper Air Soundings in
MAP, MAP Newsletter NR. 9, p. 68-69
Groehn, I., R. Steinacker, Ch. Häberli, W.Pöttschacher, M. Dorninger, 1999: DAQUAMAP- results and
response of providers, MAP Newsletter Nr. 11, p. 76
Groehn, I., R. Steinacker, Ch. Häberli, W.Pöttschacher, M. Dorninger, 2000: DAQUAMAP after SOP,
MAP Newsletter Nr. 12, p. 22
Groehn, I., R. Steinacker, Ch. Häberli, W.Pöttschacher, M. Dorninger, 2000: DAQUAMAP- results and
response of providers, MAP Newsletter Nr. 13, pp. 6-8
Groehn, I., et al, 2000 : Data Quality Control of MAP (DAQUAMAP), MAP Newsletter 13
Contributions to meeting proceedings
Steinacker, R., M. Dorninger, W. Pöttschacher, 1996: A high resolution analysis scheme for the
atmosphere over complex terrain: a contribution to MAP. Proceedings of the 24th International
Conference on Alpine Meteorology 1996, Bled, Slovenia, 9. - 13.9.1996, 106-111
Pöttschacher, W, R Steinacker and M Dorninger, 1996: VERA A high resolution analysis scheme for
the atmosphere over complex terrain. Preprints of the 7th Conference on Mesoscale Processes,
Reading, UK, 9. - 13.9.1996, 245-247
Steinacker, R., Ch. Häberli, I. Groehn, W. Pöttschacher, M. Dorninger, 1998: DAQUAMAP - An
Initiative to Assess the Performance of Surface Observations in MAP Using a Model Independent
Method. 8th Int. Conf. on Mountain Meteorology, Flagstaff, Amer. Meteor. Soc., pp 113-118.
Fig.4:Cloud cover (white shaded areas) and lifting condensation level (isolines)