InfraRed Investigation in Singapore IRIS Observatory Urban heat island contributors and mitigators analysis using neighborhood-scale thermal imaging

2025-05-05 0 0 4.85MB 18 页 10玖币

侵权投诉

InfraRed Investigation in Singapore (IRIS) Observatory: Urban heat island contributors and

mitigators analysis using neighborhood-scale thermal imaging

Miguel Martin1, Vasantha Ramani1, Clayton Miller2,∗

1Berkeley Education Alliance for Research in Singapore, Singapore

2College of Design and Engineering, National University of Singapore (NUS), Singapore

∗Corresponding Author: clayton@nus.edus.sg, +65 81602452

Abstract

This paper studies heat ﬂuxes from contributors and mitigators of urban heat islands using thermal images and weather data. Thermal

images were collected by a rooftop observatory between November 2021 and April 2022. Over the same period, weather stations

were operating at several locations on a university campus in Singapore. From collected data, a method was deﬁned to estimate

sensible and latent heat ﬂuxes from building fa

c¸

ades, vegetation, and traﬃc. Before analyzing heat ﬂuxes using the method, thermal

images were calibrated against measurements made with contact surface sensors. Results show that the method can be used to study

heat ﬂuxes with a higher temporal resolution at the neighbourhood scale than any other technique using thermal images collected by

a satellite. Heat ﬂuxes can also be analyzed over a longer period while considering urban morphology with a higher ﬁdelity than

these estimated using most urban climate models. However, as heat ﬂuxes are directly calculated from measurements, the method is

not able to forecast the impact of certain elements in the built environment on urban heat islands. In the future, this limitation can be

overcome if heat ﬂuxes assessed by the method are used to train and test data driven models.

Keywords: Urban heat islands, Infrared thermography, Automatic weather stations, Heat balance, Building fac¸ades, Vegetation,

Traﬃc

Nomenclature

εThermal emissivity 0-1

ωWater vapour content kgwater /kgair

ϕRelative humidity 0-1 or %

ρDensity kg/m3

σStefan-boltzmann constant (5.67 ·10−8) W/m2-K4

τSolar transmissivity 0-1

AArea m2

cpSpeciﬁc heat J/kg-K

cveg Speciﬁc heat of vegetation J/m2-K

Ef uel Fuel consumption J/m

eWater vapour pressure hPa

hConvective heat transfer coeﬃcient W/m2-K

KShortwave radiation W/m2

kThermal conductivity W/m-K

LLongwave radiation W/m2

LAI Leaf Area Index m2/m2

lLength m

NNumber of -

QHeat ﬂux W/m2

raAerodynamic resistance s/m

rsStomatal resistance s/m

TTemperature oC or K

tTime s

UCamera output voltage V

VVolume m3

WsWind speed m/s

∆xThickness m

List of abbreviations

CFD Computational Fluid Dynamics

HVAC Heat, Ventilation, and Air-Conditioning

LST Land Surface Temperature

MBE Mean Bias Error

RMSE Root Mean Square Error

UCM Urban Canopy Model

UHI Urban Heat Island

1. Introduction

Nowadays, more than half of the world’s population lives in

cities [

]. To accommodate the urban population, buildings and

streets have been constructed in large quantities. The expansion

of built-up surfaces is a major cause of Urban Heat Island (UHIs),

together with the absence of vegetation and the augmentation of

human activity. UHIs are responsible for heat stress in various

cities around the world, and thus, it is perceived as a threat to

public health in particular during heat wave episodes [

]. Due

to the importance of this climatic hazard, various methods have

been used to study UHIs [

]. While some methods primarily rely

on measurements, others have been used to understand causes

and eﬀects of UHIs using an urban climate model.

One of the ﬁrst methods has used thermal images collected

by satellites to observe UHIs [

]. From the thermal images,

it is possible to evaluate the Land Surface Temperature (LST)

and its diﬀerence between urban and rural areas. They can thus

provide indications on how hot or cold is the surface of the

Preprint submitted to Energy and Buildings February 12, 2024

arXiv:2210.11663v2 [physics.ao-ph] 9 Feb 2024

built environment of a city in comparison to this of its rural

surroundings. However, thermal images alone can hardly be

used to assess the air temperature diﬀerence between urban and

rural areas, which is the most common indicator of UHIs. They

can also be taken at large time intervals and at a limited scale.

As reported in a recent review published by [

], satellites usually

collect thermal images on a daily basis and show the LST at the

city scale.

To quantify the air temperature diﬀerence between urban

and rural areas at a higher temporal resolution and lower scale,

other methods have used data collected by a network of weather

stations [

]. In urban areas, weather stations

are commonly installed at the rooftop of buildings or on lamp

posts at the street level. Although UHIs can be observed with

a higher temporal resolution at the neighbourhood scale from a

network of weather stations, they can be studied within a limited

number of positions in the city or rural surroundings. To enhance

the spatial resolution of observations made from weather stations,

some methods have used interpolation techniques [

]. These methods fail in accurately estimating

the air temperature between two positions located far from one

and the other in the city. The reason is that the air temperature in

a city depends on many factors which can hardly be taken into

account by an interpolation technique in space.

Instead of interpolating measurements obtained by weather

stations, studies have used Computational Fluid Dynamics

(CFD), a highly sophisticated modelling technique of urban mi-

croclimates, to improve the spatial resolution with which UHIs

can be investigated at the neighbourhood scale [

]. Based

on spatial and temporal discretization methods of Navier-Stokes

equations, CFD aims at estimating air motion and temperature at

each cell of a three dimensional mesh. Meshes usually consists

of a large numbers of cells, and therefore, CFD requires signiﬁ-

cant computational eﬀorts to assess air motion and temperature

at the neighbourhood scale. A consequence of high computa-

tional eﬀorts is that CFD can provide information about UHIs

and their countermeasures at the neighbourhood scale within a

limited time period.

Whether the air temperature is assessed from weather stations

or estimated using CFD, none of these methods can be used to

determine how signiﬁcantly certain elements of the built environ-

ment contribute to UHIs at the neighbourhood scale. To evaluate

the importance of some contributors or mitigators of UHIs in

the built environment, studies have mainly used Urban Canopy

Models (UCMs) [

]. UCMs are essentially meant to

estimate the air temperature and humidity within a street canyon

from sensible and latent heat balances. Heat balances are de-

ﬁned from ﬂuxes emitted by building fa

c¸

ades, the street surface,

vegetation, Heating, Ventilation, and Air-Conditioning (HVAC)

systems, and traﬃc. The heat ﬂuxes are estimated directly from

an empirical formula or indirectly from heat balances. They

can be used to explain the reasons why certain elements of the

built environment contribute more to UHIs than others. How-

ever, when analysing contributors and mitigators of UHIs using

UCMs, it is important to remember that this type of urban cli-

mate model have a simpliﬁed consideration of urban morphology

in which buildings are assumed to have similar dimensions and

streets equal width.

In the literature, studies have shown that heat ﬂuxes from

contributors and mitigators of UHIs can be analyzed considering

the urban morphology with a high ﬁdelity to the one observed in

reality. One method is to assess heat ﬂuxes from remote sensed

data obtained by a satellite together with measurements collected

by weather stations. Satellite data are particularly useful for

estimating the net-all wave radiation ﬂux [

]. If combined with weather data, they can be used

to evaluate convective and/or latent heat ﬂuxes [

The net-all wave radiation ﬂux, the convective heat ﬂux, and the

latent heat ﬂux can be balanced to estimate the anthropogenic

heat ﬂux and/or the net-heat storage [

Despite the complexity of the analysis that can be made on

contributors and mitigators of UHIs using methods combining

satellite and weather data, it remains that heat ﬂuxes are assessed

with a low temporal resolution at a high scale. In addition to this

limitation, there is also the fact that only horizontal surfaces can

be observed from a satellite.

To assess heat ﬂuxes from both vertical and horizontal sur-

faces with a higher temporal resolution at the neighborhood

scale, it is now possible to collect thermal images from observa-

tories. An observatory consists of an infrared thermal camera

installed at the rooftop level. It can also be composed of a pan/tilt

unit to collect thermal images at diﬀerent positions over time.

So far, a few studies have used this modern technology to

analyze contributors and mitigators of UHIs at the neighborhood

scale. For example, [

] and Morrison et al. [

] assessed

the longwave radiation emitted by several urban facets. [

]

estimated the sensible heat transferred by building fa

c¸

ades to

the outdoor air. By sensible heat, it is here referred to the total

heat transferred by convection and radiation over a period. In

other studies, like [

], only considered the heat transferred by

convection. Instead of the heat transferred by convection and/or

radiation from building fa

c¸

ades, [

] tried to detect sources of

anthropogenic heat where HVAC systems are installed at the

rooftop of buildings.

These studies show there are three major gaps in the analysis

of contributors and mitigators of UHIs using thermal images

collected by an observatory:

•

Firstly, thermal images have focused on building fa

c¸

ades

and HVAC systems, which are contributors of UHIs. Conse-

quently, no information is available on mitigators of UHIs,

like vegetation, as seen from an observatory.

•

Secondly, only the sensible heat ﬂux has been assessed

from thermal images, and thus, no observation has been

made on the net-all wave radiation ﬂux, the latent heat ﬂux,

and the net-heat storage.

•

Thirdly, there is no thermal image collected by an observa-

tory that shows the impact of traﬃc on the outdoor environ-

ment.

To ﬁll these gaps, this study aims at accomplishing the follow-

ing research objectives using thermal images collected by an

observatory and measurements obtained by weather stations: 1)

Install an observatory from which building fa

c¸

ades, vegetation,

and traﬃc can be observed with a high temporal resolution at

the neighborhood scale, 2) Assess all possible heat ﬂuxes from

building fa

c¸

ades and vegetation using thermal images collected

by the observatory, and 3) Detect traﬃc from thermal images to

estimate their heat releases.

Results of this study should be of interest to the scientiﬁc

community and urban planners. The scientiﬁc community would

be able to see the contribution of certain elements of the built

environment to UHIs from a diﬀerent perspective and temporal

resolution at the neighborhood scale as normally reported in the

literature. These observations can also motivate investigations

on data driven modelling to predict the evolution of UHIs and the

eﬃcacy of mitigation strategies. Data driven models, together

with thermal images collected from an observatory, could easily

be integrated in a digital twin of a city to help urban planners in

determining how to make cities more sustainable and resilient

towards UHIs.

The method used to analyze heat ﬂuxes using thermal im-

ages collected by an observatory and data obtained by weather

stations is described in Section 2. It comprises the mathemati-

cal formulation of heat ﬂuxes, the method used to collect data

from an observatory and weather stations, the procedure used

to study the sensitivity of the surface temperature assessed from

thermal images with respect to some parameters, and the calibra-

tion of the surface temperature obtained from thermal images

against measurements of contact surface sensors. In Section

3, the results of this study are detailed and discussed. Finally,

conclusions are explained in Section 4.

2. Materials and Methods

2.1. Assessment of heat ﬂuxes

This section describes the method to analyze contributors

and mitigators of UHIs using thermal images collected by an

observatory. Contributors here refer to building fa

c¸

ades and

traﬃc, while mitigators corresponds to vegetation. The inﬂuence

of contributors and mitigators on the outdoor air temperature

and humidity within the urban canopy depends on the magnitude

their sensible and latent heat ﬂuxes, respectively. To assess the

magnitude of heat ﬂuxes from contributors and mitigators of

UHIs, it is not suﬃcient to use only thermal images collected

by an observatory. Temperature, relative humidity, wind speed,

and solar radiation must be measured by weather stations at

the same time. Using thermal images and weather data, it is

possible to estimate sensible and latent heat ﬂuxes from building

c¸

ades and vegetation directly from their surface temperature or

indirectly from an heat balance. Sensible and latent heat ﬂuxes

from traﬃc are calculated using a method to detect cars from

thermal images.

Any built-up surface in an urban area absorbs and emits heat

in accordance with the following heat balance:

Q∗=QH+QG+ ∆QS(1)

where

Q∗

is the net-all wave radiation ﬂux; that is, the heat

gained or lost by radiation in the short- and longwave range,

the heat transferred from the surface to the outdoor air by

convection, that is the convective heat ﬂux,

the heat trans-

fer from the outer to the inner layer of the built-up surface by

conduction, and ∆

the heat stored by the built-up surface (see

Appendix A). Due to the diﬃculty in directly evaluating

Q∗

from

Equation 3 using thermal images, it is usually recommended

to indirectly assess it from

, and ∆

and ∆

however, can directly be estimated from the surface temperature

assessed from thermal images (

) and measurements of one or

several weather stations. Given the surface temperature recorded

at Position i j by a thermal image (Ti j), TSis expressed as:

TS=1

SX

i j∈S

Ti j (2)

where

is the set of all Positions

i j

in the thermal image cor-

responding to the surface

. To evaluate

, a contact surface

sensor needs to be installed inside a building being seen by the

observatory.

Equation 3 is valid for opaque surfaces; that is surfaces only

absorbing and reﬂecting incoming solar radiation (

K↓

). Trans-

parent surfaces also transmit a portion

K↓

into buildings. It

means that their energy balance is expressed as:

Q∗−τK↓=QH+QG+ ∆QS(3)

where

τK↓

is the transmitted incoming solar radiation through

the transparent surface.

In their vegetated urban canopy model, [

] expressed the

heat absorbed and emitted by vegetation as:

Q∗=QH+QE+ ∆QS(4)

where

is the latent heat ﬂux produced by evapotranspiration

(see Appendix A). Similarly to

, it can be assessed from

thermal images and weather data. The net-heat stored by the

vegetation (∆QS) is calculated as:

∆QS=cveg

∆TS

∆t(5)

where cveg is the heat capacitance of the vegetation.

The heat releases from traﬃc (

Qtra f f ic

), including both sensi-

ble and latent heat ﬂuxes, can be expressed as a function of the

number of vehicles (

) crossing a portion of a road using the

formula of [53], that is:

Qtra f f ic =1

3600 ·Aroad Nv·lroad ·Ef uel(6)

The number of cars crossing a portion of the road at an instant

(

) can be detected from thermal images taken at two subse-

quent times. It means there is a function fsuch that:

f(|U′n+1

R−U′n

R|)=Nn

v(7)

where

U′n

R−<Un

is the thermal image within the

region

taken at time

n·

∆

and centered over the average

<Un

. Given

, the number of vehicles crossing the portion

of the road

over one hour from time

(

) can be calculated

as:

Nv=

(t+3600−t0)/∆t

n=(t−t0)/∆t

(t+3600−t0)/∆t

n=(t−t0)/∆t

f(|U′n+1

R−U′n

R|) (8)

2.2. Rooftop observatory

From November 2021 to March 2022, an observatory was op-

erating at the rooftop level to collect thermal images of diﬀerent

buildings on a university campus in Singapore. Singapore is a

city-state in South-East Asia located near the equator. At this

location, a hot and humid climate is experienced over the year.

The air temperature varies between 26 and 30 degrees Celsius

on average every month. The monthly average humidity is also

relatively constant, with variations between 80 and 90 percent.

When not obstructed by buildings, the wind most frequently

blows at a speed between 1 and 4 meters per second from the

South-East direction.

The observatory was installed on the rooftop of a 42-meter-

tall building located in a residential area, as illustrated in Figure

1. The residential area is located in front of a university campus

consisting of oﬃce and educational buildings. Among the build-

ings, four can be observed from the observatory with a proper

resolution. Building A is one of the tallest on the university

campus. It is about 68-meter-tall with an important portion of its

c¸

ade covered by curtain walls. Closer to the observatory are

Buildings B and C, which are both about 27-meter-tall. Their

c¸

ade consists of concrete walls and single-pane windows. In

addition to concrete walls, the fa

c¸

ade of Building D consists of

metal grids installed on a concrete frame. Building D was de-

signed to be net-zero, and its height is around 24 meters. Around

buildings A, B, C, and D, it is possible to observe several tropical

trees from the observatory. In front of buildings B, C, and D,

there is a road with heavy traﬃc.

On the top of the observatory, there was a housing containing

a FLIR A300 (9Hz) thermal camera as described in Table 1.

The housing enables the thermal camera to be protected against

heavy rains with IP67 protection. It was ﬁxed on a pan/tilt unit

in order to record thermal images at diﬀerent positions. To avoid

any obstacles while recording thermal images, the pan/tilt unit,

together with the housing, including the thermal camera, was

placed on a 2-meter-high truss tower. This structure is stabilized

by concrete blocks and protected against lightning by an air

terminal. On the truss tower, two sockets were installed to power

up the thermal camera and the direct current motor of the pan/tilt

unit from a backup battery located in a water tank room. The

backup battery is continuously recharged from the electrical

source of the building so as to keep the thermal camera and

the pan/tilt unit operating for up to 2 hours in case of power

shutdown. The thermal camera and the pan/tilt unit were also

connected to a laptop for conﬁguring and checking the collection

of thermal images.

The collection of thermal images was conﬁgured from two

separate software. One software was installed on a video encoder

to command the pan/tilt unit. From its graphical interface, it is

possible to deﬁne the positions where the pan/tilt unit must stop

to take a thermal image. The moment when a thermal image is

taken is controlled by another software installed on the laptop.

Thermal images can be saved either in JPEG or FFF ﬁle format

inside a folder to be speciﬁed in the software.

The thermal camera and the pan/tilt unit were conﬁgured so

that images can be taken at four positions, as shown in Figure

2. Position I is centered on Building A. From this position, it is

also possible to observe vegetation consisting of tropical trees

mostly. After staying at Position I for a while, the observatory

moves to Position II. This position primarily focuses on Building

B and its surrounding vegetation. A similar thermal image is

taken at Position III but centered on Building D. Finally, the

observatory stops at Position IV where various elements can be

observed, including Building D, vegetation, and a road. For each

position, thermal images are recorded at a rate of one minute

approximately. They are stored on a Google Drive repository

through a 4G Internet connection installed on the laptop.

2.3. Network of automatic weather stations

Figure 3 shows the network of weather stations that was used

to estimate heat ﬂuxes of built up surfaces and vegetation. The

network was deployed by [

] in February 2019. It consists of

12 weather stations measuring the air temperature and relative

humidity. All stations, except 12, measure the wind speed and

direction. Solar radiation is measured by all stations apart from

11 and 12. Instruments to measure temperature, relative humid-

ity, wind speed/direction, and solar radiation were connected

to a data logger to make measurements every 1-minute interval.

Their speciﬁcation is summarized in Table 2.

As mentioned in Section 2.1, various parameters measured

by weather stations need to be used for estimating sensible

and latent heat ﬂuxes. For instance, sensible and latent heat

ﬂuxes emitted by buildings and vegetation observed at Position I

was assessed from the temperature, relative humidity, and wind

speed as measured by the Weather Station 12. The latent heat

ﬂux emitted by vegetation also depends on the solar radiation,

which was deﬁned from measurements of the Weather Station 2.

Measurements obtained from the Weather Station 2 were used

to evaluate sensible and latent heat ﬂuxes observed at Position

II. Heat ﬂuxes observed at Position III and IV were estimated

from weather conditions measured by Weather Stations 3 and 5,

respectively.

2.4. Sensitivity analysis

Before assessing

Ti j

from the observatory, a sensitivity analy-

sis was conducted on parameters that might aﬀect its variance

(see Appendix B). A parameters can be a constant or a variable.

The contribution of a parameter to the variance of

Ti j

was es-

timated from the ﬁrst-order Sobol index [

]. The higher the

ﬁrst-order index associated to a parameter is, the more

Ti j

sensitive to that parameter. However, the ﬁrst order index does

not consider interactions that one parameter might have with

others. For this reason, the total Sobol index was also calculated

during the sensitivity analysis of Ti j.

Table 3 illustrate the parameters considered during the sensi-

tivity analysis of

Ti j

and their respective boundaries. According

to [

], the thermal emissivity of target object varies between 0.8

and 1.0 in the built environment. In case the emissivity is slightly

below 0.8, it was decided to calculate its sensitivity in a range

between 0.7 and 1.0. The sky temperature was measured by

[

] in Singapore, and varies between 11 and 33 degrees Celsius.

Based on weather data recorded by the Meteorological Service

of Singapore [

], the outdoor air temperature can change be-

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

InfraRedInvestigationinSingapore(IRIS)Observatory:Urbanheatislandcontributorsandmitigatorsanalysisusingneighborhood-scalethermalimagingMiguelMartin1,VasanthaRamani1,ClaytonMiller2,∗1BerkeleyEducationAllianceforResearchinSingapore,Singapore2CollegeofDesignandEngineering,NationalUniversityofSingapore(...

展开>> 收起<<

InfraRed Investigation in Singapore IRIS Observatory Urban heat island contributors and mitigators analysis using neighborhood-scale thermal imaging.pdf

共18页,预览4页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

InfraRed Investigation in Singapore IRIS Observatory Urban heat island contributors and mitigators analysis using neighborhood-scale thermal imaging

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: