Section 7.5: Comparing Observed and GCM Climates
133
7.5.3 Relationships Between Circulation Frequencies
and Temperature and Precipitation
Tables 7.3 and 7.4 summarize the relationships between the seasonal frequeneies of particular circulation types and mean temperature and total rainfall
in the equivalent season. The relationships, quantified in terms of simple
correlation coeffieients, are shown for the real-world (observational) data and
the model-generated eirculation types, temperature and preeipitation. The
relationships are not always consistent in the different seasons (see also Table
7.1). S days are generally warm in all seasons. A-type days are, however,
warm in summer but cool in winter. Preeipitation variability is dominated
by the frequency of synoptic types in all seasons; it being generally wet on C
days and dry on A days.
The sign of the temperature relationships are, on the whole, correctly simulated by the models, though performance in summer is inferior to winter,
when both models reproduce the correct sign and strength of relationship
between temperature and A, E and W types (and MPI correctly simulates
a negative correlation with N). Only UKHI captures the correct spring relationships with W, N and E, though C is incorrectly warm. Both models
correctly give cold N in winter but MPI has spurious cold A and warm C
relationships. Importantly, neither model generates the correct (but weak)
warm S relationship apparent in the observed data in all seasons.
As for the preeipitation relationships, both models reproduce realistic correlations with A and C types though the magnitude of the model-related
correlations is less than in the observed data and MPI does not reproduce
the positive relationship (i.e. more rain) with C in summer while UKHI likewise does not produce wet C days in autumn and winter. The model data
demonstrate overly strong relationships between preeipitation and directional
types, especially in UKHI. Whereas the observed relationships between precipitation and W types are significant (but still quite low at 0.31) only in
winter, UKHI simulates a much stronger relationship (0.77) in winter and in
spring and autumn. SimilarlY' MPI shows a very strong (0.80) but unrealistic
relationship in summer (remember that MPI also produced too few W days
in summer and too little precipitation - see Figures 7.5 and 7.6).
7.5.4 Weather-Type SpeIl Lengths
and Storm Frequencies
Two final brief illustrations of ways in which the model data can be compared
with observed data are shown in Figures 7.7 and 7.8. The first examines the
lengths of C and A speIls in the sample model output. UKHI tends to underestimate short and medium C speIls (up to about 6 days) and overestimate
133
7.5.3 Relationships Between Circulation Frequencies
and Temperature and Precipitation
Tables 7.3 and 7.4 summarize the relationships between the seasonal frequeneies of particular circulation types and mean temperature and total rainfall
in the equivalent season. The relationships, quantified in terms of simple
correlation coeffieients, are shown for the real-world (observational) data and
the model-generated eirculation types, temperature and preeipitation. The
relationships are not always consistent in the different seasons (see also Table
7.1). S days are generally warm in all seasons. A-type days are, however,
warm in summer but cool in winter. Preeipitation variability is dominated
by the frequency of synoptic types in all seasons; it being generally wet on C
days and dry on A days.
The sign of the temperature relationships are, on the whole, correctly simulated by the models, though performance in summer is inferior to winter,
when both models reproduce the correct sign and strength of relationship
between temperature and A, E and W types (and MPI correctly simulates
a negative correlation with N). Only UKHI captures the correct spring relationships with W, N and E, though C is incorrectly warm. Both models
correctly give cold N in winter but MPI has spurious cold A and warm C
relationships. Importantly, neither model generates the correct (but weak)
warm S relationship apparent in the observed data in all seasons.
As for the preeipitation relationships, both models reproduce realistic correlations with A and C types though the magnitude of the model-related
correlations is less than in the observed data and MPI does not reproduce
the positive relationship (i.e. more rain) with C in summer while UKHI likewise does not produce wet C days in autumn and winter. The model data
demonstrate overly strong relationships between preeipitation and directional
types, especially in UKHI. Whereas the observed relationships between precipitation and W types are significant (but still quite low at 0.31) only in
winter, UKHI simulates a much stronger relationship (0.77) in winter and in
spring and autumn. SimilarlY' MPI shows a very strong (0.80) but unrealistic
relationship in summer (remember that MPI also produced too few W days
in summer and too little precipitation - see Figures 7.5 and 7.6).
7.5.4 Weather-Type SpeIl Lengths
and Storm Frequencies
Two final brief illustrations of ways in which the model data can be compared
with observed data are shown in Figures 7.7 and 7.8. The first examines the
lengths of C and A speIls in the sample model output. UKHI tends to underestimate short and medium C speIls (up to about 6 days) and overestimate