Application of an improved WRF-urban
canopy model to urbanization impact study
over northern Taiwan
Chuan-Yao Lin
Research Center for Environmental Changes, Academia Sinica, Taiwan
04 December, 2015 CWB
World Urbanization Prospects
(United Nations, 2014)
2
The importance of land use change
Background-
4
What is the Urban Heat Island ?
(Sailor D.J. 2011)
(Sailor D.J. 2011)
Urbanization History of Taipei
1895
1904
1959
1988
(provided by Prof. Lai Chun Kuei, NTU)
1925
2000
Heat wave in Taipei
1981-2006 energy consumption v.s population
120,000
Energy consumption
Populations
Oil Equivalent KL)
100,000
2.4
2.3
2.2
80,000
2.1
60,000
2
1.9
40,000
1.8
1.7
20,000
1.6
0
2005
2001
2003
1997
1999
1993
1995
1989
1991
1985
1987
1981
Elevated temperatures in urban environments. (Photo: NASA)
1983
1.5
population 10 millions
Urban heat island effect:
2.5
°C
Taipei average temperature in July and Aug.
a
LST
(Shiu et al.2009)
Cross scale modeling system: From regional to
urban scale (WRF-UCM)
Regional scale
Local scale
Urban scale
Dynamic Effects
Thermal effects
WRF/Urban Canopy Model
• Single layer Urban-Canopy Model (UCM, Kusaka et al., 2004)
• UCM treats man-made surfaces
– urban geometry (orientation, diurnal cycle of solar azimuth), symmetrical street
canyons with infinite length
– Shadowing from buildings and reflection of radiation
– Anthropogenic heating
– Multi-layer roof (HR), wall (HW) and road (HG) models
Shadow and Radiation Trapping
Temperatures and Thermal Transfer
Defined and implemented urban
canopy parameterizations such as
height-to width ratios and sky view
factors into model formulations
Review: Impacts of the Urban Heat Island Effect on boundary
layer development and Precipitation over Northern Taiwan
Simulation UHI effect and UHI intensity
(Lin et al. 2008 a A.E.)
UHI effect impact on precipitation:



Precipitation max over urban: Calm condition
Precipitation enhanced at downwind: weak regional winds (many papers)
Moving regional storm: Cities can split convective storm, and change the
behavior of convective precipitation, and enhance downstream
precipitation. (Bornstein and LeRoy,1990; Bornstein and Lin ,2000; Baik et
al., 2001; Niyogi et al. 2011)
(Niyogi et al. 2011)

Urban reduces precipitation: less evaporation, higher sfc temperature,
sensible heat fluxes, and aerosol. Zhang, Chen, Miao (2009) JGR
UHI effect Impact on precipitation:
• TRMM radar imagers:
• An average increase of about 28% in monthly rainfall
rates within 30-60 km downwind of the city
(from Steiger et al., JGR 2002)
(from Shepherd 2002)
Generation/Enhancement of Rainfall by cities
urban heat island produces a strong
convergence of winds and resulted in a
local effect of sea breeze in the urban area
17.
of Houston, Texas. (X. Tang 2008)
(Rosenfeld, 2002)
How the UHI effect perturb thermal and dynamic process
over northern Taiwan ?
• More complicate landscape in northern Taiwan
Radar reflectivity observed by CWB
1400
1700
1430
1500
1800
1600
WRF-Noah UCM model study the summer
thunderstorm
Original
MODIS
WRF-USGS
Observation
(Lin et al. 2011, JAMC)
12LST
14LST
16LST
17LST
MODIS
(color: water vapor mixing ratio)
A
USGS
B
21.
(Lin et al. 2011, JAMC)
Schematic the mechanism of UHI impact on
precipitation system
積雨雲
午後雷陣雨
海風
lifting
Warm, dry
(Lin et al. 2008 b A.E.)
78 cases statistic analysis (2004~2008)
frequency of the occurrence of the
maximum radar reflectivity in the
column >35 dBZ
Frequency of the occurrence of
Rainfall >0.5 mm
Simulation of summer thunderstorm
Composited atmospheric conditions (78 cases)
Impact of an improved WRF-urban canopy
model on diurnal air temperature
simulation over northern Taiwan
Chuan-Yao Lin1*,
Chiung-Jui Su1,Hiroyuki Kusaka2 , Yuko Akimoto2, Yang Fan Sheng1 ,
Jr-Chuan Huang 3, Huang-Hsiung Hsu1
1. Research Center for Environmental Changes, Academia Sinica, Taiwan
2. Center for Computational Science, University of Tsukuba,Japan
3. Department of Geography, National Taiwan University
MODIS Land use classification in Taiwan
MODIS 500 m
Original UCM (1-D)
urban fraction Def.=0.7
1km resolution, generated from100 m resolution from
National Land Surveying and Mapping Center.
2-D Urban Canopy Model (UCM) model
-Urban Fraction
 Original UCM :
Urban Fraction is fixed, for instance 0.7, or given in three types of urban
 2-D UCM :
• 2-D urban fraction : generated from 100 m resolution from
National Land Surveying and Mapping Center.
Original UCM (1-D)
urban fraction Def.=0.7
2-D UCM: urban fraction
(1km resolution)
2-D Urban Canopy Model (UCM) model
-Anthropogenic heat
 Original UCM :
Anthropogenic heat(AH) is fixed, for example, 50W/m2
 2-D UCM :
– Anthropogenic Heat
• 2-D anthropogenic heat is generated from 100 m resolution of
building density (2006),
• The maximum AH value is 50 W/m2.
Original 1-D UCM AH=50 (W/m2)
2-D AH map
(1km resolution)
Heat Wave case study
• 2012/07/10
• Taipei: 38.3℃
Obs. data
Model Evaluation (WRF-UCM2D)
Dom1=3km
Dom2=1km
Case Study: 10 July, 2012
Taipei: 38.3℃
Model configuration:
Initial & boundary : NCEP-GFS
Boundary layer: MYJ scheme
Cloud physics: WSM6
Radiation: RRTMG
study period: 09-11 July
Model Evaluation
OBS.
11 LST
WRF-UCM
WRF-UCM2D
Model Evaluation
12 LST
WRF-UCM
WRF-UCM2D
Model Evaluation
13 LST
WRF-UCM
WRF-UCM2D
Model evaluation for 19 urban stations (red dots)
Model evaluation for 21 non-urban stations (yellow dots)
Non-urban WRF-UCM
Non-urban WRF-UCM2D
Model evaluation for 21 non-urban stations
WRF-UCM
Non-urban WRF-UCM
WRF-UCM2D
Non-urban WRF-UCM2D
daytime
Non-urban WRF-UCM
nighttime
Non-urban WRF-UCM2D
19 LST
00 LST
05 LST
Urban fraction=0.313
Urban fraction=0.04
Urban fraction=0.127
Temperature difference between modeling and observation
at 21 non-urban stations
Non-urban (FRC=0.0~0.4) 21 stations
Factors influencing model performance in non-urban
areas during nighttime
𝐹𝑠ℎ − σ𝑇 4 = ρ𝑠 𝐶𝑝 𝐶ℎ (𝑇𝑠𝑘 − 𝑇2𝑚 )
Where 𝐹𝑠ℎ is grid-averaged sensible heat flux,
σ𝑇 4 is the upward long-wave radiation,
𝜌𝑠 is the density of surface air,
𝐶𝑝 is the specific heat capacity of air at constant pressure,
𝐶ℎ is the surface exchange coefficient for heat from the surface-layer scheme,
𝑇𝑠𝑘 is skin temperature and
𝑇2𝑚 is diagnostic 2 m height air temperature.
Non-urban area, nighttime: 𝐹𝑠ℎ − σ𝑇 4 = ρ𝑠 𝐶𝑝 𝐶ℎ (𝑇𝑠𝑘 − 𝑇2𝑚 )
T2m
𝝆T𝒔sk
𝑪F𝒑sh𝑪𝒉
Urban
Fraction
T2m
(oK)
UCM2D UCM
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
297.4
298.8
299.5
299.9
300.3
300.3
300.9
301.6
299.3
299.7
299.9
299.9
300
300
300.4
301
Fsh
(W/m2)
Tsk
(oK)
2o
(W/m K)
Diff. UCM2D UCM
Diff. UCM2D UCM
Diff. UCM2D UCM
-1.8
-0.9
-0.3
0
0.2
0.3
0.4
0.6
2.5
5.6
8.6
11.8
14.6
17.5
17.2
19
6.9
6
5.6
5.9
6.7
7.7
10.2
12.1
-10.5
-10.2
-8.9
-6.5
-3.5
0.7
3.7
9.7
-13.1
-15.8
-17.4
-18.3
-18.1
-16.8
-13.5
-9.3
Diff=UCM2D- UCM
16.5
20
22.8
25
24.7
24.4
21.9
20.6
9.6
14
17.1
19.1
18
16.7
11.6
8.5
296.9
298.4
299.3
299.8
300.2
300.5
301.1
302.1
296.5
297.7
298.2
298.4
298.5
298.5
298.6
299.2
Tsk-T2m
(oK)
Diff. UCM2D UCM
0.4
0.8
1.1
1.3
1.7
2
2.6
2.9
-0.52
-0.37
-0.25
-0.14
-0.02
0.15
0.28
0.5
-2.78
-2.03
-1.66
-1.44
-1.5
-1.53
-1.88
-1.81
Urban (red dots)
Non-urban (yellow dots)
Model evaluation ( July, 2010)
Model data was excluded from analysis for all times where simulated rainfall was found.
July 2012
Daytime
Nighttime
Non-urban
WRF-UCM WRF-UCM2D WRF-UCM WRF-UCM2D WRF-UCM WRF-UCM2D
BIAS (C)
0.44 (0.06)
0.01 (-0.1)
0.29 (-0.15)
0.27 (-0.02)
0.57 (0.41)
-0.22 (-0.22)
RMSE (C)
1.55 (1.53)
1.29 (1.38)
1.53 (1.58)
1.43 (1.49)
1.56 (1.46)
1.14 (1.18)
R2
0.78 (0.78)
0.84 (0.82)
0.83 (0.76)
0.84 (0.78)
0.53 (0.57)
0.76 (0.73)
Summary
•
With the improved WRF-UCM2D model, the oversimplified results can be avoided
with the percentage of urbanization in the model grid nets more accurately
identified according to the actual land use and building density for AH, not only in
the city center but also in rural small towns.
•
The performance of WRF-UCM2D is much better than WRF-UCM at non-urban
stations with low urban fraction during nighttime. It is attributed to energy
exchange that enables efficient turbulence mixing in areas with low urban
fraction (in particular with urban fraction < 0.2). Energy exchange contributes to
reduce air temperatures simulated by WRF-UCM2D, followed by decrease in
ground surface temperatures
•
simulation results show that the critical urban fraction is around 0.2, at which the
difference in T2m obtained by WRF-UCM2D and WRF-UCM is zero. Finally, the
proposed WRF-UCM2D successfully improved the simulation of diurnal variation
of air temperature in urban and non-urban areas. The results of this study can be
applicable to assessing the impacts of urbanization on air quality and regional
climate.
Ongoing study
Improve anthropogenic heat (AH) estimation
AH from building density
population density
AH sources: industry, buildings, vehicles, metabolism
AHF Estimation from Air pollutants
• The new method suggested by the Lee et al. (2014)
for gridded AHF estimation
𝐶𝑂
0.64
𝑌𝐴𝐻𝐹
= 2.55 × 𝑥𝐶𝑂
(1)
𝑁𝑂𝑥
𝑌𝐴𝐻𝐹
(2)
0.69
= 8.32 × 𝑥𝑁𝑂
𝑥
𝑵𝑶
𝒙
𝒀𝑨𝑯𝑭 = 𝜶𝒀𝑪𝑶
+
𝟏
−
𝜶
𝒀
𝑨𝑯𝑭
𝑨𝑯𝑭
(𝟑)
Anthropogenic heat Estimation from Air pollutants
Consider urban height in UCM2D
e.g.
(Wu. et al. 2011)
Future Heat wave projection
ECHAM5-WRF, A1B scenario
Model evaluation and simulation results:
(2075-2099)-(1979-2003)
1979-2003
Taipei Summer (Jul. and Aug.)
air Temperature in 2012 and 2075
hour
2012 (NCEP-WRF/2D UCM)
Temp. (C)
Heat wave and WBGT heat index
WBGT = 0.7Tw + 0.2Tg + 0.1Td
Where,
Tw = Natural wet-bulb temperature;
Tg = Globe thermometer temperature
Td = Dry-bulb temperature
ECHAM5-WRF dynamic downscaling
5km resolution
Future urbanization scenarios and urban planning
S2-2026
S2-2046
S2-2066
(Provided by Prof. 詹士樑&陳 亮全)
Urban pollution : Heat wave and Air quality
WRF-chem/UCM(1-D):PM10 (μg/m3)
10am
中午
2pm
Urban pollution : Heat wave and Air quality
AH impact on the CO vertical distribution
AH=0
AH=150 W/m2
hpa
PM10 = 130 ug/m3; O3 = 111 ppb
ppm
PM10 = 23
ug/m3;
O3 = 42 ppb
ppm
Taipei night Photo from [email protected]
Thank you !!!