Weather type-dependent verification of vertical profiles

Andrea M. Rossa and Marco Arpagaus, MeteoSwiss
 

Outline

  • Motivation
  • Alpine weather statistics
  • Scores for selected regimes
  • Summary

    • Motivation

    The limitations of monthly, seasonally, and yearly statistical verifications are well known, in that model performance is judged over the whole spectrum of weather types the atmosphere can produce, and is, therefore, not able to highlight weather type-specific model deficiencies. In this paper we present first results of a weather type-dependent verification of the Swiss Model's (SM) vertical profiles based on the Schüepp (Schüepp 1979) classification, which has been derived from the synoptic climatology of the Alpine region.

    Alpine weather statistics

    The main elements of the Schüepp (Schüepp 1979) classification are as follows:
  • 40 classes divided into convective, advective, and mixed synoptic situations;
  • over 50 years of weather classification for the Alpine region (~220km radius);
  • partly subjective based on five parameters: surface pressure gradient, surface wind direction, and wind and geopotential height on 500hPa;
  • systematic for 12UTC every day.
    • The evaluation of the Schüepp classes for the meteorological year 1999 is as follows:
     
     
    		 DJF MAM JJA SON Total
    advective west
    6
    11
    8
    1
    26
    advective north
    12
    12
    2
    12
    38
    advective east
    3
    4
    3
    1
    11
    advective south
    5
    4
    3
    11
    23
    convective high
    16
    13
    14
    19
    62
    convective low
    1
    7
    3
    7
    18
    convective flat
    16
    33
    51
    26
    126
    mixed jet
    30
    8
    8
    13
    59
    mixed mix
    1
    0
    0
    1
    2

    Scores for selected regimes

    As already shown by Binder and Rossa (1995) the quality of the SM's vertical profiles is generally very good. Characteristic error structures have been found but the magnitude of the model errors are of the same order of magnitude as the observational errors. Therefore, for the present study we concentrate on the most apparent features of the weather type-dependent verification.

    As an example, bias and standard deviation of the geopotential height and the temperature profiles are shown in figures 1 and 2 for sounding stations in the Alpine region (Payerne, Lyon, Muenchen, Innsbruck, Udine and Milano). Note that warm/cold biases result in positive/negative biases increasing with height for the geopotential height.

    For the meteorological year 1999 we consider the advective, convective and jet classes. As may be expected the standard deviation for the more dynamic synoptic situations (advective/jet) is larger than for the convective situations characterized by generally weaker horizontal gradients.


     

    Figure 1: Verification (mean error and standard deviation) of geopotential height (forecast time +48 hours) for advective, convective and jet classes.

     

     
     
     
     
     


     

    Figure 2: Verification (mean error and standard deviation) of temperature (forecast time +48 hours) for advective, convective and jet classes.

     

     
     
     
     
     

    Concentrating on the convective subclasses high and low, characterized by dry, fair weather and rainy weather, respectively, we note a stronger warm bias from the 900 up to the 350hPa level for the low class at forecast time +48h (fig. 4), which is not present at analysis time (fig. 3). Also, the low class features a drier bias in a substantial portion of the troposphere (c.f. figures 6 and 5, respectively). This is consistent with the hypothesis that the model tends to overestimate moist diabatic processes, that would lead to the warm and dry bias. This in turn would leads to an overestimation of precipitation, which should be apparent as a positive bias in the surface precipitation. Plans are made to extend the weather type-dependent verification to the surface parameters.


     

    Figure 3: Verification (mean error and standard deviation) of temperature for convective classes high and low and forecast time +00 hours.

     

     
     
     
     
     


     

    Figure 4: Verification (mean error and standard deviation) of temperature for convective classes high and low and forecast time +48 hours.

     

     
     
     
     
     


     

    Figure 5: Verification (mean error and standard deviation) of relative humidity for convective classes high and low for forecast time +00 hours.

     

     
     
     
     
     


     

    Figure 6: Verification (mean error and standard deviation) of relative humidity for convective classes high and low for forecast time +48 hours.

     

    Summary

    A weather type-dependent verification of the vertical structure of the SM has been presented. The overall quality of the profiles is very good, as the model errors are of the same order of magnitude as the observational errors. Characteristic differences in the error structure have been identified between some of the synoptic weather classes, a finding that may help to pinpoint specific problems in the model formulation. Plans are made to extend the weather type-dependent verification to the surface parameters.

    References

    Schüepp, M., 1979. Witterungsklimatologie. Beiheft zu den Annalen der Schweizerischen Meteorologischen Anstalt (Jahrgang 1978), 93pp.
    Binder, P., and A. M. Rossa, 1995. Verifikation der Vertikalstruktur des SM. DWD-SMA Hochauflösendes Regionalmodell Rundbrief Nr. 10, Teil III.