1. High resolution model simulation of
Department of Climatology and Landscape Ecology
University of Szeged, Hungary
High resolution model simulation of
Urban Heat Island for a medium sized Hungarian city
Case studies and verification
András Zénó Gyöngyösi, Gergely Molnár, and Tamás Gál

2. Motivation Small area (280 km2), 170k inhabitants,
Surface of (low or high intensity) urban landuse is even much less
Frequent hot, summer nights (tropical nights/“beer-garden” days: 50/80d per annum, respectively).
Heat waves are relatively common
Only few grid cells are involved
High model resolution is required

3. Szeged

4. Szeged

5. Szeged

6. UC monitoring sys Stations installed Represent LCZs
Installation: light poles
Electricity + protection

7. UC monitoring sys 22 (urban) + 2 (rural) stations
Minute average observations at 4 mAGL
T, Rh
From end of March, 2014 w/ gaps

8. Typical diurnal variation
Representative stations
BAR urban
HMS rural
Other stations
Nocturnal UHI
ΔT defined as:
ΔT=TBAR – THMS
Daily variations
ΔTn: 06 – 12Z
ΔTx: 00 – 06Z

9. UC monitoring sys
HMS
BAR

10. Seasonal variation Clear summer nights: ΔT=[-2,+6] (ΔT)x ≠Δ(Tx)
Modeling 26 days in 2015 summer – reproduction

11. Test period in 2015 Selection criteria: ∆T Observed larger than 4.5°C
Clear conditions: Radiation larger than 750Wm-2
No precip

12. Test period in 2015 Periods (26 days) Cases (7) 31st May – 7th June
1st – 7th July
5th – 14th August
Cases (7)
3rd June
03rd, 6th and 7th July
6th, 7th and14th August

13. Model domain
∆x = 10km

14. Model domain

15. Model domain
∆x = 2km

16. Model domain

17. Model domain
∆x = 0.4km

18. Model domain
97 x 97

19. Model domain

20. Model domain

21. Model domain
HMS
BAR

22. Model setup More info: Poster H60 Molnar et al.MP RA
SFC
LSM
PBL CU
e_vert
WSM5
RRTMG
revMM5
Noah
BouLac KF 44
WSM6
MYJ
nil
500hPa
test0
test1
test2
test3
nil
UCM
BEP
BEM
Modified URB PARM table
Modified geog database
Landuse
USGS → CORINE → LandSat
Soiltype
FAO → DKSIS
More info:
Poster H60
Molnar et al.

23. Modified URB PARM data
More info:
Poster H60
Molnar et al.

24. Geog database More info: Poster H60 Molnar et al. LU ST DKSIS geog0
USGS
CORINE
CORINE+LandSat
modified geog2 ST
FAO
DKSIS
More info:
Poster H60
Molnar et al.

25. Modeling strategy 48-hour runs on GFS .25 3-hourly input
24-hour overlap for spin-up
Urban heat effects appear in the model results only on the second evening
Output in urban (BAR) and rural (HMS) grid cells at every time steps (60, 12 and 2.4s resp).
1-minute average 2m Temperature
Comparison to observations

26. Results –
Output from different model setups have been compared to observed urban effects
Sensitivities for
geographic database
resolution
parameterizations
urban parameters
have been tested.

27. Results –

28. Results –
Temporal and spatial variation patterns have been captured properly
Tx and ΔT values are acceptable
However, both URB and RUR Tn values are too high compared to measurements
Best performance:
3 domains, geog3 modified database, UCM urban physics and modified urban parameter table
Even for 2 domains results were acceptable
Further tests were performed on two domains with uniform setup

29. More results...
Temporal and spatial variation patterns have been captured properly
Tx and ΔT values are acceptable
However, both URB and RUR Tn values are too high compared to measurements

30. Conclusions
An appropriate modeling system for the representation of urban heat effects for a medium size city in Hungary have been established
Appropriate static data and evaluation method have been constructed
Model results have been verified on extensive and representative measurements
Good agreement is found but further development is necessary
Inclusion of anthropogenic heat
Definition of more urban categories based on LCZs
Testing further parameterization schemes (RAD?) and domain setups

31. Thank you very much for your kind attention
Urban Climate Research Group at the
University of Szeged, Hungary
Thank you very much for your kind attention
Hungarian Scientific Research Fund (OTKA K )