(more details, C. Rada, A. Sima, M. Caian, NIMH)
A new function has been tested to express the climatic vertical profile of the ozone in the Aladin model, for integrations over Romanian domain.
For that 10 years of SBUV/2 NOAA11 data and daily measured data at Bucharest have been fitted by a function, similar to the Arpege one but with one additional degree of freedom, a variable exponent.
Monthly triplet of constants gave the new monthly variable profile function, as well as daily triplets fitted the daily measured values. For a period of time, in parallel to the operational run, two more runs were executed: a run with the monthly climatic profile and another with the daily fitted profile. The assumption made was the constancy of the profile in the integration domain: studies based on measured data show, usually, a slow latitudinal variation over Romania territory. A variation with latitude and longitude has been finally tested but fitting satellite date, due to the fact that one single DOBSON spectrometer UMKEHR measurement data, at Bucharest, is available. The impact of the ozone vertical profile on the forecast has been tested.
In the Aladin model, the integrated ozone is a function of pressure, being defined by the following expression : (1)
Using Umkehr profiles provided by SBUV data selected for the geographical position of Bucharest, monthly climatic vertical profiles have been constructed based on a 10 years interval. For these profiles a new function has been fitted, with small differences comparing with the one used in the Aladin model, having the following form: (2)
The monthly variation of the triplet (a, b, c) used in the numerical experiments for the new function of the vertically integrated ozone amount are presented in the following table: (Table 1)
Table 1 | |||
Month |
A [Pa] |
B[Pa] |
C |
January |
0.070335 |
4084.541 |
2.687082 |
February |
0.068064 |
3817.875 |
2.627732 |
March |
0.064599 |
3569.678 |
2.648133 |
April |
0.067411 |
3844.202 |
2.650851 |
May |
0.073363 |
4034.560 |
2.544536 |
June |
0.072327 |
3909.156 |
2.578798 |
July |
0.068094 |
3644.002 |
2.608211 |
August |
0.063252 |
3309.811 |
2.651979 |
September |
0.060063 |
3174.439 |
2.673819 |
October |
0.060502 |
3282.954 |
2.683656 |
November |
0.059083 |
3326.518 |
2.703777 |
December |
0.063446 |
3676.920 |
2.700567 |
Aladin |
0.060120 |
3166.000 |
3.000000 |
In the Aladin model the function is used with a constant exponent (c) equal to 3 and a, b given in the previous table. The c variable was introduced to better fit the maxima localization on altitude.
Yearly differences between the Aladin total ozone and the climatic one are shown (Fig. 1), for a reference pressure of 1013 hPa. The bigger differences are from january to august with a maximum in April.
Fig. 1 Climatic integrated ozone at Bucharest for a reference pressure and Fig. 2 satellite measured ozone the 5-th of july 2000Two other methods have been tested: (M1) fitting the same function based on integrated amount and (M2) using the empirical function of Lacis and Hansen (1974), both fitted to a same climatic profile. These last two methods gave very similar results but rather different from the computation with the first method and the function of Aladin. More tests to compare these methods will be performed in a future work to find the best fit.
In parallel daily computation have been made starting from spectrophotometer measured data: a daily profile (triplet (a,b,c)) have been obtained. The difference reported to the climatic profile was variable especially due to dynamical process, in the range of a 10-40 UD magnitude of difference.
An example is shown for the 05-th of July 2000, a case which have been studied due to its special feature: the last 100 years maxima in Bucharest, 42 C has been attaint. Satellite measured data showed that day (Fig. 2) a gradient of more than 50 UD difference over the domain latitudes band, which was also about the deviation in that day from the standard profile used in the model.
The consequences of using a local climatic vertical ozone profile in the initial conditions for the short term forecast was analyzed.
First, an ensemble of 5 experiments was performed to find out the relative importance of the parameters a, b, c, on the total amount and profile shape for this region. A closer attention was given to the simulation of the 5-th July 2000.
The domain of the numerical simulations is the operational one in Romania, at 10.3 km. horizontal resolution, cycle 11 of development of the Aladin model.
A serial of 3 experiments has been daily performed for few days in february, april, may and july. The operational run was compared, together with a daily adjusted vertical ozone profile and a monthly climatic adjusted profile integration, against observations.
In parallel, the sensitivity to small variations in the a, b, and c parameters of the function (2) was analyzed for the climatic profile, keeping two of the parameters constants the third being slightly modified. The parameter c, controlling the maxima position on altitude showed the bigger impact on the integration result. The a parameter, with a major control on the total column amount, showed less sensitivity if the profile shape was unchanged at a same percentage of integrated amount variation as for the experiment controlling c parameter.
The mean skills over the tested situations in may (for which sensitivity experiments have also been done) are shown, for the 2m. temperature in the table 2. The global skills show small improvement, averaged over the domain, but differences reported to the operational solution lied in the interval (-2C, 2C), with even larger interval in july ( (-4C,4C) for the 5-th of july). Climatic run showed in general better forecast, maybe due to the main assumption made: the generalization over the whole domain of the profile fitted at Bucharest, hypothesis better fulfilled at climatic range.
Mobsv |
Mforv |
Miner |
Maxer |
Mer |
Maer |
Std | |
Oper |
21.98 |
21.55 |
-4.49 |
4.70 |
-0.43 |
1.42 |
1.75 |
Run |
21.98 |
21.56 |
-4.44 |
4.73 |
-0.42 |
1.41 |
1.74 |
climatic |
21.98 |
21.62 |
-4.41 |
4.78 |
-0.33 |
1.40 |
1.73 |
A_mod |
21.98 |
21.58 |
-4.46 |
4.82 |
-0.40 |
1.41 |
1.76 |
B_mod |
21.98 |
21.54 |
-4.50 |
4.75 |
-0.41 |
1.41 |
1.74 |
C_mod |
21.98 |
21.60 |
-4.45 |
4.77 |
-0.38 |
1.40 |
1.74 |
The 05-th of July 2000 simulation was simulated by the two usual runs: the daily and the climatic fitted and, due to it's special feature (last 100 years maximum at Bucharest of 42 C) two more simulations have been performed: a mixed vertical profile (COMB) in order to better approximate the real measured data (Fig. 1), and the experiment M1. The climatic run overestimated (cf. Table 1 data) the Aladin total integrated ozone amount, the daily fitted profile (at Bucharest) was close to Aladin profile (slightly underestimating Aladin profile), which is that day a mean of the distribution over the domain (Fig. 1).
Fig. 3 : a) climatic data run against observed values (-1, 1) belt; b) climatic data run against Aladin in observation points for 2m temperature.It means that the climatic data run should improve the northern part at short term, Aladin run should show better average skills while the daily fitted run should improve the southern part. All these features are indeed shown by the three runs: Fig. 3 and Fig. 4 show the differences between climatic data run and observation, respectively differences between climatic and Aladin runs.
Maximum improvement of the climatic data run is to be found superposing absolute minimas on the first chart with maximas on the second. We find indeed the bigger positive impact in the northern (NE) part of the domain.
It is to be precised that the highest improvement was realised in strong convective activity areas: for example in the north-eastern observation point in Romania were measured 64 mm precipitation in 24 minutes, and it is the point were the bigger correction was attained: 1.5 C (Fig. 4d shows the convective potential at 12 hours computed by a method described in Olessen et al. , 1992) . This result confirms the results obtained also by other authors on limited area (Sapporo, 2000 Proceedings of the Quadrennial ozone symposium,T. Halenka).
Fig. 4 Differences of temperature (a- 2m, b-50Hpa level) and 2m relative humidity (c) between 2 runs using climatological and operational ozone profile after 12 hours of integration.Maximum differences reported to the operational model are localised towards surface, because of the short range of the integration (48 hours), even if the highest ozone profile differences are in stratosphere ( Fig. 5). In this sense the Fig. 4 shows again higher differences in the northern part at all levels (Fig. 4b) but with increased values towards surface (Fig. 4a). Differences in 2m. temperature are also felt in the relative humidity at 2 m (Fig. 4c).
Fig. 5 Aladin ozon profile (function (1) )together with the climatic and daily fitted profiles with the function (2), with pressure as vertical axis.At this range, the cumulated impact on radiative balance shows (table 3) that the increase of the stratospheric ozone amount (corresponding to the climatic profile against the operational one) at about 30-50 UD leads at short term to decrease of surface absorbed solar radiation solar at about 0.1%, hence to sensible heat mean decrease of 0.65%. In opposite, a stratospheric reduction, as in the case of the experiment M1 leads to a positive, more emphasised effect on tropopause: about 0.4% on solar radiation at surface and 2.4 %for the sensible heat flux).
Table 3 | |||
Cumulated mean flux over domain |
CLIMATIC (JULY) |
OPERATIONAL | |
Solar surface absorbed flux |
1.5241476E+07 |
1.5260151E+07 | |
Earth surface radiation flux |
-4043134. |
-4044040. | |
Latent heat flux |
-5890797. |
-5894787. | |
Evaporation flux |
-2.419137 |
-2.420792 | |
Sensible heat flux |
-892700.4 |
-898550.0 | |
Solar surface absorbed flux (M1) |
1.5319988E+07 |
1.5260151E+07 | |
Sensible heat flux (M1) |
-920239.1 |
-898550.0 |
The combined run (COMB) has closer skills to the Aladin model (OPER - table 4, with same notations as in table 3). In fact, the mean integrated real amount over the domain that day is close to the Aladin profile, so we can take, only for averaged values, Aladin integration as reference. The best skills are obtained by the simulation M1.
Table 4 | |||||||
run |
Mobsv |
Mforv |
Miner |
Maxer |
Mer |
Maer |
Std |
OPER |
32.95 |
30.73 |
-8.67 |
9.55 |
-2.22 |
2.91 |
2.79 |
COMB |
32.95 |
30.86 |
-8.89 |
9.65 |
-2.09 |
2.82 |
2.78 |
M1 |
32.95 |
30.85 |
-8.76 |
9.66 |
-2.10 |
2.82 |
2.77 |
In opposite, both climatic and daily runs, imposing a half-true value to the whole domain increased error skills.
Ozone profile adjustment to local climatology have not a strong impact on short range forecast skills, on a limited domain model. But the amount and the vertical distribution modification, leading mainly to stratospheric amount modification leads to significant impact on tropospheric solar heating.
The experiments done using a monthly climatic profile showed that the difference can lead, in some special cases (as in the 5-th of july, 2000 case) to some local amplified impact. This can lead, through increased instability conditions to convective potential increase and to important direct effects connected with this triggering.
Home |