Post on 15-Nov-2015
description
1 23
Journal of The Institution ofEngineers (India): Series ACivil, Architectural, Environmental andAgricultural Engineering ISSN 2250-2149Volume 95Number 2 J. Inst. Eng. India Ser. A (2014) 95:83-90DOI 10.1007/s40030-014-0074-y
Development of Pavement TemperatureContours for India
M.R.Nivitha & J.M.Krishnan
1 23
Your article is protected by copyright and allrights are held exclusively by The Institutionof Engineers (India). This e-offprint is forpersonal use only and shall not be self-archived in electronic repositories. If you wishto self-archive your article, please use theaccepted manuscript version for posting onyour own website. You may further depositthe accepted manuscript version in anyrepository, provided it is only made publiclyavailable 12 months after official publicationor later and provided acknowledgement isgiven to the original source of publicationand a link is inserted to the published articleon Springer's website. The link must beaccompanied by the following text: "The finalpublication is available at link.springer.com.
ORIGINAL CONTRIBUTION
Development of Pavement Temperature Contours for India
M. R. Nivitha J. M. Krishnan
Received: 22 February 2013 / Accepted: 5 May 2014 / Published online: 3 June 2014
The Institution of Engineers (India) 2014
Abstract The stress-strain response of the bituminous
pavements is highly sensitive to temperature. To system-
atically analyze the pavement performance, it is necessary
that one understands the variation of pavement temperature
spatially and temporally during the life time of a pavement.
In this investigation, historic air temperature data for 37
locations across India was collected. Using this database,
pavement temperature data was predicted by an appropriate
air temperature-pavement temperature model. High and
low temperature pavement temperature contours were
generated for the first time for India. It was seen that the
locations spanning from Srinagar to Madhya Pradesh and
Rajasthan to Orissa were extremely critical. The minimum
temperature in these locations was 10 C and the maxi-mum temperature was around 68 C. Clearly such infor-mation is necessary when making choice of binder grade
and bituminous layer thickness.
Keywords Pavement temperature Artificial neural networks Air temperature forecasting Design air temperature
Introduction
Bitumen is a complex construction material used for
pavement construction. Of the total pavements constructed
across the world, 90 % of them use bitumen as the binder.
Though the proportion of binder is only around 46 % in
the bituminous layers, it has a large influence on the dis-
tresses in bituminous pavements.
The three common distresses observed in a bituminous
pavement are rutting, fatigue cracking and low temperature
cracking. The properties of binder chosen is considered to
have a significant influence on these three distresses [1].
Rutting, fatigue cracking and low temperature cracking
observed in a pavement are influenced by the binder
approximately to an extent of 80, 60 and 40 % respectively
[2]. Hence the type of binder chosen for any given location
is a significant factor affecting the performance of the
pavement.
The critical factors influencing the choice of binder for
any pavement are pavement temperature and traffic. Binder
response ranges from elastic solid to viscoelastic solid to
viscoelastic fluid to non-Newtonian and Newtonian
depending on the temperature it is subjected to. In the
working temperature ranges of a pavement (1065 C),bitumen exhibits predominantly viscoelastic behaviour and
hence the mechanical response is dependent on speed of
traffic (rate of loading). Owing to this effect of temperature
and traffic, the required performance of a binder can be
obtained only in a certain temperature and traffic range.
Hence it is essential to use the binder at locations which
have identical temperature and traffic factors. To enable
this, temperature and traffic details are required to be
known for any location before selecting the appropriate
type of binder.
The importance of temperature and traffic on the per-
formance of the pavement is understood and the pavement
design specifications proposed by various countries
account for these factors while selecting an appropriate
grade of binder. In India, the design guidelines for
M. R. Nivitha J. M. Krishnan (&)Department of Civil Engineering, Indian Institute of Technology
Madras, Chennai 600036, India
e-mail: jmk@iitm.ac.in
123
J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390
DOI 10.1007/s40030-014-0074-y
Author's personal copy
bituminous pavements [3] gives a very general approach
regarding the choice of binder. A broad classification is
provided for temperature and traffic and the binder grade is
specified depending upon the combination of these two
factors. The climatic regions of the country are classified as
hot, cold and moderate and the traffic is considered in two
categories namely high and low volume. The Australian
specifications [4] use the same baseline except that quan-
titative values are given for temperature and traffic.
The Superpave method of binder selection [2] currently
being followed in North America encompasses detailed
analysis on the influence of temperature and traffic on the
performance of pavements. The Superpave method of
binder specification selects the grade of binder based on
temperature of the location where it is to be used and tests
the binder for its performance properties with respect to the
temperature range prevalent at the selected location. Seven
day average maximum and one day minimum pavement
temperatures are considered as the design pavement tem-
peratures. Traffic is included in the specification by means
of testing the binder properties at a specified frequency.
The binder is tested for a constant traffic volume of 10
Million Standard Axles (MSA) and traffic speed of
80 kmph. For other traffic conditions, suitable modification
procedures are specified [2]. Thus the final grade of binder
can be obtained for a given location having a specified
traffic speed and volume.
To develop a robust binder selection methodology cus-
tomized to the climatic and traffic conditions of any region,
one needs meticulous data collection. This is the major
limitation as far as India is considered. The basic infor-
mation required for binder selection and stress-strain ana-
lysis is the data pertaining to the pavement temperature. To
the best of the knowledge of the authors, such data was
collected by the Highways Research Station (HRS) in
Chennai as early as 1969 [5]. In this paper, an attempt is
made to use the available air temperature data and statis-
tical models to generate pavement temperature data for
India.
Data Collection
Thirty seven locations were selected such that they cater to
all the geographical areas of India. Daily maximum and
minimum air temperature data were collected for these
locations, for a period of 30 years (19702000), from the
archived database available at Indian Meteorological
Department, Pune. A design period of 20 years was chosen
during which air temperature and pavement temperature
has to be predicted using the 30 years historical weather
data. Hence the window here is a total period of 50 years
and the yearly design air temperature pattern is expected to
vary in the 50 year period. This difference is expected to be
amplified to a significant level when it is converted into
pavement temperature. It is well known that the air tem-
perature follows periodic variations with varying cycle
times. Some predominant air temperature patterns are the
Bruckners cycle (around 35 years) [6] and Hales cycle
(around 8 years) [7]. Forecasting air temperature using
pattern recognition tools can capture the pattern in the air
temperature history data and reflect the same for the design
period. This to an extent will provide the realistic daily air
temperature for the design period.
For the thirty seven chosen locations and for a design
period of thirty years, a total number of 8,10,300 data
points were collected. Of the total 14,600 data points
available for each city, at least 50 data points were missing.
The various methods available to fill in missing entries may
be categorized under three heads: within-station, regres-
sion-based and inter-station [8]. Using the methodology for
within-station category, the average of the past five days
and the next 5 days was taken as the current days tem-
perature for filling in missing and anomalous data. This can
be expressed as follows:
xi P5
n1 xin P5
n1 xin
2
1
where, xi = Air temperature of the missing entry
As a check to validate this approach, 1,000 air temper-
ature data points were selected from the input data for
Chennai and 10 temperature values starting from 1st Jan-
uary, 1970 were removed at an interval of 50 data points.
These missing values were calculated using Eq. 1. The
average mean absolute error(MAE) of the selected points
was 1.032 C and MAE up to 2 C is considered acceptablefor filling in missing data in the literature [9]. Hence this
method was considered acceptable and it was used to fill in
all the missing entries. There were also cases when the data
was missing for more than two consecutive days. In these
cases, recursive iteration was carried out on the basis that,
the difference in values between two successive iterations
was less than 0.001.
Air Temperature Forecast
Among the various models available to forecast time series,
ANN was chosen to predict the air temperature as it is
considered to be effective in pattern recognition and long
term forecasting [10]. The in-built Neural Networks tool
box available in MATLAB [11] was used to predict the air
temperature and for that purpose the available data was
separated into training, testing and validation sets. All the
initial trials were carried out for Chennai data by varying
84 J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390
123
Author's personal copy
the number of layers, number of neurons in each layer and
the proportion of training and testing data to obtain the best
fit parameters for predicting the air temperature. As an
initial trial, 15 years (19701985) data was taken as
training data, 10 years (19861995) for testing and five
years (19962000) for validation. The number of layers
were varied from 2 to 5 and the neurons in each layer were
varied from 1 to 20. The best fit was obtained when 2
layers were used having 20 and 2 neurons respectively and
a proportion of 20:7:3 for training:testing:validation data
set were used to forecast the air temperature. The same best
fit parameters were used for all the 37 cities and for each
city ANN was trained separately with daily maximum and
minimum air temperatures to predict the same for the
design period.
To validate the ANN model, daily air temperature data
for Chennai, available at an internet domain [12] was
collected for the year 2009 and compared with the pre-
dicted air temperature (Fig. 1). The MAE for daily maxi-
mum air temperature is 1.88 C and it is slightly higherthan the daily minimum air temperature having a MAE of
1.35 C. An MAE up to 2 C is commonly reported inliterature [13, 14] for air temperature forecasting using
ANN and hence the MAE obtained in this study is con-
sidered to be acceptable. The yearly maximum and mini-
mum air temperature for the 50 year period comprising the
data collection period (19702000) and the design period
(20012020) for Chennai is shown in Fig. 2. It can be
observed from Fig. 2 that the variation of design air tem-
perature for the 50 years is within a band of 5 C with thevariation of air temperature more pronounced in the actual
data compared to the forecast.
The next step after obtaining the daily air temperatures
for the design period would be to choose a particular air
temperature or a norm of air temperature values to
represent the temperature variation in a year for the
selected location. The temperature variation of a location is
commonly indicated in terms of the average or maximum
and minimum air temperatures experienced at the location.
In India, the design guidelines for bituminous pavements
[3] considers the annual average pavement temperature
directly for determining the material properties for stress-
strain analysis. This measure of temperature does not
highlight the extremes of temperature a pavement is
experiencing in its service life. It is expected that locations
having the same annual average pavement temperature can
have different maximum and minimum temperatures and
hence different susceptibility to rutting and cracking. It is
thus necessary to consider the yearly maximum and mini-
mum temperatures in a binder selection methodology. In
this paper, the seven day average maximum and one day
minimum air temperatures as per the Superpave specifi-
cation [2] is followed to calculate the design air tempera-
ture. These approaches also have their own limitations
[15].
Pavement Temperature Model
The design pavement temperatures can be calculated from
design air temperatures using analytical models and
regression equations. In the analytical models, the heat
equation is solved and the pavement temperature is cal-
culated knowing the weather parameters such as solar
radiation, absorptivity and emissivity of the surface and air
temperature [16]. Regression equation developed from the
measured pavement temperature database relating air
temperature and latitude can also be used and the pavement
temperature for the required time period can be calculated
[17]. Due to the complexity in measuring the material
Air
tem
pera
ture
, C
Fig. 1 Actual and predicted (using ANN) air temperature for Chennaifor the year 2009
Air
tem
pera
ture
, C
Fig. 2 Variation of yearly maximum and minimum air temperaturesfor Chennai from 1970 to 2020
J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390 85
123
Author's personal copy
parameters required for use in the analytical models, most
of the attempts related to pavement temperature have
focused on regression models. The various factors influ-
encing the pavement temperature are air temperature, lat-
itude, solar radiation, wind speed and rainfall. Among these
factors, air temperature and latitude are considered as the
sole factors influencing the pavement temperature [17].
The development of regression models require consid-
erable amount of data and since such data was not available
for India, the air temperature and the corresponding
pavement temperatures data collected as part of the Long
Term Pavement Performance Program (LTPP) of USA was
used to develop the regression model [18]. The LTPP
database [18] monitors and collects data for 890 sites under
various studies such as General Pavement Section (GPS),
Specific Pavement Study (SPS) and Seasonal Monitoring
Program (SMP). The SMP, which collects the various data
including the air and pavement temperature for a location,
has been implemented only in 66 sites. Eleven sites (Table
1) out of the sixty six sites which had similar latitude and
altitude as that of India were selected on the basis that
locations having similar latitude and altitude are expected
to have identical climatic conditions. The southern part of
the USA and the northern part of India have similar latitude
(36 to 25N) (Table 1) and hence are expected to haveidentical climatic conditions for specific altitude ranges. To
verify the similarity in air temperature pattern, the monthly
average air temperatures of a location in India (Patiala,
having a latitude of 30.33N and an altitude of 249 m) wascompared with another in USA (Atlanta (SHRP ID:1031),
having a latitude of 32.61N and an altitude of 138 m). Itcan be observed from Fig. 3 that the air temperature pattern
is similar for both the locations and hence details from
LTPP can be used for the regression model. The altitude of
the selected locations in India and USA was limited to less
than 2000 m. A total of 172 data points were collected and
the data collection period was from 1994 to 2001. The
database had data points from all the months throughout
the year. To formulate a regression equation using this
database, a linear regression equation with two variables
was assumed considering the effect of latitude and air
temperature on the pavement temperature [17]. The coef-
ficients of the equation were determined by the in-built
linear regression function in MATLAB and given as
follows:
Pt 0:7147 1:3023At 0:1103L; 2where, Pt = pavement temperature (
C); At = air temper-ature (C); L = latitude of the selected location. Thisregression model developed is used to calculate pavement
temperature for the entire country.
As there was no pavement temperature information
available to validate the regression model, a comparison
Air
tem
pera
ture
, C
Fig. 3 Comparison of air temperature for Atlanta and Patiala for theyear 2000
Table 1 Details of locations selected for collection of pavement temperature from LTPP database [18]
United States of America India
SHRP ID Latitude Altitude, m Location Latitude Altitude, m
1024 35.27 1663 Guwahati 26.11 47
1053 38.69 1567 Gwalior 26.14 205
1005 32.61 138 Jodhpur 26.17 217
1031 34.40 37 Gorakhpur 26.45 76
1112 34.30 1146 Jaipur 26.53 385
1060 28.51 24 Lucknow 26.55 122
1068 33.50 136 Dibrugarh 27.29 110
1077 34.54 559 Delhi 28.37 233
1122 29.23 143 Patiala 30.2 249
3739 26.98 11 Dharamshala 32.16 1457
1001 37.28 1336 Srinagar 34.08 1585
86 J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390
123
Author's personal copy
was performed with the data points extracted from the HRS
data [5]. This study was carried out in the year 19691971.
The highways research station at Chennai constructed an
experimental bituminous concrete section and collected
pavement temperature at the surface of the pavement as
well as at different depths. From this report, 48 data points
were extracted to validate the regression model developed
as a part of this study. Among these, 24 data points were
hourly pavement temperature measurements on a single
day, 8th June, 1969 and the other 24 points were monthly
average pavement temperatures for two years, 1970 and
1971. Figure 4 shows the predictions for the full 24 h. Due
to brevity, the monthly average predictions are not shown
here. It is seen that the present model can predict the
pavement temperature to a reasonable accuracy.
Using Eq. 2, the design maximum and minimum
pavement temperatures were calculated for all the selec-
ted 37 locations. The design pavement temperature was
calculated using the maximum of seven day average
maximum and minimum of one day minimum air tem-
perature obtained for the design period [2]. The reliability
considered in this case is of the order of 99.6 % as the
maximum of seven day average maximum air temperature
was used in calculating the pavement temperature and
similarly the one day minimum air temperature for the
design minimum case. However, in circumstances where
agencies specify other reliability levels, the design air
temperature can be considered accordingly. The mean and
standard deviation values were calculated for design
maximum and minimum air temperatures for all the cities,
as represented in Table 2. One can now use these values
to calculate the design air temperatures for any required
level of reliability and convert them into pavement tem-
peratures using Eq. 2. The maximum and minimum
pavement temperatures calculated for reliability levels of
99 and 75% are shown in Table 2.
Pavement Temperature Contours
Quantum GIS software was used to interpolate and
determine the spatial distribution of pavement tempera-
tures. Pavement temperature values were input for the
selected 37 locations and temperature values for the
intermediate locations were interpolated using triangular
interpolation. The contour maps were then generated for
the entire country based on the interpolated values for all
the locations. The entire map area was divided into 1,000
grids vertically and horizontally and the values were
interpolated for each cell. As a check to validate the
interpolation method inbuilt in the software, three loca-
tions namely Bhubaneswar, Coimbatore and Kurnool
were removed from the data set and the pavement tem-
peratures were determined based on interpolation. The
average MAE was 0.59 C for design maximum tem-perature and 1.35 C for design minimum temperaturewere obtained for the interpolated values. The maximum
deviation obtained for these cities was less than 2 C formaximum and minimum pavement temperatures. The
maximum pavement temperature contour for India is
shown in Fig. 5 for every two degree increase in pave-
ment temperature starting from 54 to 68 C. For anylocation lying in between the contours, one can interpo-
late between the two temperature contours.
It can be deduced that the pavement temperatures in
India are highest in regions of central Rajasthan, portions
of Haryana, Uttar Pradesh, Madhya Pradesh, Bihar,
Jharkhand and Chhattisgarh. In these locations, the pave-
ment temperature crosses 65 C. The design minimumpavement temperature in some of these same locations
goes up to 10 C making them critical both in terms ofrutting and low temperature cracking. Hence the binder to
be used in these locations is required to have excellent
temperature susceptibility to withstand both rutting and
low temperature cracking. The core of the southern part of
India experiences high temperature during summer over 65C but the low temperature does not fall below 18 Ccompared to the northern part where the low temperature is
observed to be 10 C. In the coastal regions of thesouthern part of India, the high temperature does not rise
beyond 60 C and the low temperature does not fall below10 C. In these locations, a low temperature susceptiblebinder is considered to be sufficient provided the binder
exhibits the required performance at high temperatures.
Pavem
ent t
empe
ratu
re,
C
Time, hr
Fig. 4 Comparison of pavement temperatures measured by HRS [5]and predicted by regression equation
J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390 87
123
Author's personal copy
Conclusion
The choice of a correct quality of the binder in a pavement
construction is paramount in ensuring that it exhibits the
required performance for the complete design life. This
paper focussed on the providing the information related to
expected temperature ranges during the design life. ANN
was used to predict air temperature for the design period
using 30 years of history data for 37 locations across India.
The design air temperatures were estimated and converted
to pavement temperature using regression model developed
as a part of this study. From the pavement temperatures
calculated, pavement temperature (low and high) contours
were drawn for the first time for India.
From the pavement temperature contours, it was seen
that the design maximum pavement temperature varied
Table 2 Design temperatures for selected locations
Loc ID Location Maximum air
temperature, CMinimum air
temperature, CDesign pavement temperature for
75 % reliability (C)Design pavement temperature for
99 % reliability (C)
Mean SD Mean SD Max Min Max Min
1 Anantpur 33.9 3.24 22.66 2.65 56.31 17.77 63.31 14.60
2 Aurangabad 32.13 3.4 18.39 4.12 54.71 12.26 62.06 7.34
3 Bangalore 29.2 2.5 19.07 1.74 49.34 16.32 54.75 14.24
4 Belgaum 30.14 3.41 18.24 2.36 51.68 14.76 59.05 11.94
5 Bhopal 31.35 4.48 18.45 5.14 55.04 10.40 64.72 4.26
6 Bhubaneshwar 32.53 2.53 21.96 3.8 54.54 15.13 60.01 10.59
7 Chennai 32.52 2.54 24.53 1.95 53.75 20.00 59.24 17.67
8 Coimbatore 32.21 2.17 21.37 1.36 52.80 18.47 57.49 16.85
9 Cudappah 34.8 3.55 22.85 3.09 57.76 17.40 65.43 13.71
10 Delhi 30.85 5.72 18.73 7.21 56.04 6.80 68.41 -1.81
11 Dharamshala 23.83 4.88 14.51 5.72 46.58 4.82 57.13 -2.02
12 Dibrugarh 27.67 2.18 18.7 5.15 48.69 9.73 53.40 3.58
13 Gorakhpur 31.19 4.64 19.18 6.1 55.33 9.02 65.36 1.73
14 Guwahati 29.11 2.89 19.62 5.13 51.06 10.67 57.30 4.54
15 Gwalior 31.9 5.11 18.05 6.9 56.63 7.21 67.68 -1.03
16 Hyderabad 32.1 3.61 20.76 3.22 54.60 15.40 62.40 11.55
17 Imphal 26.7 2.66 15.42 5.83 47.53 7.08 53.28 0.11
18 Jagdalpur 31.24 3.4 19.13 4.34 53.50 12.60 60.85 7.41
19 Jaipur 31.41 5.46 18.91 6.25 56.34 8.61 68.14 1.14
20 Jharsuguda 32.7 3.98 21.01 4.73 56.18 13.06 64.78 7.41
21 Jodhpur 33.29 4.63 19.25 6.36 58.02 8.79 68.03 1.19
22 Kakinada 32.52 2.35 24.24 2.14 54.01 19.13 59.09 16.57
23 Kanyakumari 30.6 0.55 24.34 0.69 49.00 21.34 50.19 20.52
24 Kolkatta 31.29 2.73 22.07 4.51 53.34 13.96 59.24 8.57
25 Kottayam 31.77 1.36 22.3 0.56 51.34 19.89 54.28 19.22
26 Kurnool 34.55 3.63 22.97 2.9 57.65 17.54 65.50 14.07
27 Lucknow 31.58 5.17 18.59 6.65 56.31 7.85 67.49 -0.10
28 Mumbai 31.86 1.23 22.66 3.24 52.36 16.55 55.01 12.68
29 Nagpur 33.29 4.43 20.7 4.75 57.29 12.88 66.87 7.21
30 Patiala 29.32 5.75 17.64 7.05 54.28 5.80 66.71 -2.62
31 Patna 31.16 4.54 19.61 6.15 55.08 9.49 64.90 2.14
32 Raipur 32.97 4.09 20.5 4.79 56.63 12.61 65.47 6.89
33 Rajkot 33.82 3.52 20.88 4.34 57.31 13.34 64.92 8.15
34 Ramagundam 34 3.81 21.84 4.43 57.38 14.52 65.62 9.23
35 Satna 32.04 4.86 18.93 6.51 56.40 8.73 66.90 0.95
36 Srinagar 19.77 5.27 7.83 7.11 41.85 -2.34 53.24 -10.84
37 Trichy 33.91 2.77 23.88 2.09 55.48 19.63 61.47 17.13
88 J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390
123
Author's personal copy
from 54 to 68 C with highest temperature in the centralpart of India. The design minimum pavement temperatures
went sub-zero in some locations in the northern part of
India while up to 20 C was observed in the southern mostpart of India. On similar lines, detailed analysis of traffic is
also necessary before one could make a judicious choice on
the type of binder required to be used for any location. The
regression equation developed in this study has to be val-
idated with more data points as data becomes available.
The binder properties have to be ascertained specific to the
location it is serving and they have to be correlated to the
field performance. All these require enormous data col-
lection and this is the need of the hour for India.
Acknowledgments The authors are thankful to Department ofScience and Technology, Govt. of India for research grant DST/TSG/
STS/2011/46 and Indian Meteorological Department, Pune for pro-
viding weather data.
References
1. A.F. Braham, W.G. Buttlar, M.O. Marasteanu, Effect of binder
type, aggregate, and mixture composition on fracture energy of
hot-mix asphalt in cold climates. Transp. Res. Rec. 2001,102109 (2007)
2. T. Kennedy, G.A. Huber, E.T. Harrigan, R.J. Cominsky, C.S.
Hughes, H.L. Von Quintus, J.S. Moulthrop, SHRP-A-410:
Superior Performing Asphalt Pavements (Superpave): The
Product of the SHRP Asphalt Research Program (Strategic
Highway Research Program, National Research Council, Wash-
ington DC , 1994)
3. IRC-37, Guidelines for the design of flexible pavements. Indian
Roads Congress, 2001.
4. AUSTROADS 2008b4F, Guide to Pavement Technology:
Bituminous binders. Association of Australian and New Zealand
Road Transport and Traffic Authorities, 2008.
5. HRS, Temperature and deflection studies on pavements, tech.
rep., (Highway Research Station, Chennai, 1971).
6. A.J. Henry, The Bruckner cycle of climatic oscillations in the
United States. Ann. Assoc. Am. Geogr. 17(2), 6071 (1927)7. R.G. Vines, Rainfall patterns in India. Int. J. Climatol. 6(2),
135148 (1986)
8. W.P. Kemp, D.G. Burnell, D.O. Everson, A.J. Thomson, Esti-
mating missing daily maximum and minimum temperatures.
J. Clim. Appl. Meteorol. 22, 15871593 (1983)9. P. Coulibaly, N. Evora, Comparison of neural network methods
for infilling missing daily weather records. J. Hydrol. 341, 2741(2007)
10. I. Bertini, F. Ceravolo, M. Citterio, M.D. Felice, B.D. Pietra, F.
Margiotta, S. Pizzuti, G. Puglisi, Ambient temperature modelling
Fig. 5 Maximum pavementtemperature contour for India
J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390 89
123
Author's personal copy
with soft computing techniques. Sol. Energy 84(7), 12641272(2010)
11. H. Demuth and M. Beale, Neural network toolbox for use with
Matlab, MATLAB 2007a, (The MathWorks, Natick, 1993).
12. W. Underground, http://www.wunderground.com/, Accessed
30 Nov (2010).
13. B. A. Smith, R. W. Mcclendon, and G. Hoogenboom, Improving
air temperature prediction with artificial neural networks. Int.
J. Comput. Intell. 3(3), 179186 (2007)14. M. Hayati, Z. Mohebi, Application of artificial neural networks
for temperature forecasting. World. Acad. Sci. Eng. Technol. 28,159167 (2007)
15. J.S. Kern, S.H. Carpenter, Performance graded high temperature
selection criterioncan we do it better. Transp. Res. Rec. 1661,122131 (1999)
16. E.S. Barber, Calculation of maximum pavement temperatures
from weather reports. Highw. Res. Board 168, 18 (1957)17. H.A.-A. Wahhab, I. Al-Dubabe, I.M. Asi, M.F. Ali, Performance-
based characterization of Arab asphalt. Constr. Build. Mater. 11,1522 (1998)
18. Datapave online, http://www.ltppproducts.com/datapave/visual
ization/map.aspx, Oct 2010
90 J. Inst. Eng. India Ser. A (AprilJune 2014) 95(2):8390
123
Author's personal copy
Development of Pavement Temperature Contours for IndiaAbstractIntroductionData CollectionAir Temperature ForecastPavement Temperature ModelPavement Temperature Contours
ConclusionAcknowledgmentsReferences