International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 4 April 2012

http://www.ijsat.com
132

Comparison of Estimated Daily Global Solar
Radiation Using Different Empirical Models
K. Namrata
Asst. Professor Department of Electrical Engineering
National Institute of Technology,
Jamshedpur, India.
E-mail: nirmala_bimal@yahoo.com
Dr. S.P.Sharma
Professor Department of Mechanical Engineering,
National Institute of Technology,
Jamshedpur, India.
E-mail: sps.nitjsr2000@yahoo.co.in
Dr. S.B.L. Saksena
Professor Department of Electrical Engineering
National Institute of Technology,
Jamshedpur, India.
E-mail: subodhseksena_03@yahoo.co.in


AbstractWith an increase in power demand all over the
globe it is very important to study the renewable energy
sources and solar energy is quite a good option. To study
solar energy and design the conversion devices monthly
average daily global radiation data are quite essential. I n
this regard, different empirical models based on Angstrom-
Prescott model were selected to estimate the monthly
average daily global solar radiation, Hg, on a horizontal
surface for Ranchi using only the sunshine duration. The
hourly solar radiation data measured at Ranchi during the
period 2007-2008 were used to calculate the monthly mean
values of Hg using selected models. The selected models
were compared on the basis of the statistical error tests as
the mean bias error, the mean percentage error, the root
mean square error, NashSutcliffe equation, correlation
coefficient and the t-test. I n the present work the Angstrom
Model is modified and the new regression coefficient for
Ranchi is proposed i.e. H
g
/H
o
=0.279 +0.399(n/N) using
the statistical results. This will help to advance the state of
knowledge of global solar radiation to the point where it
has applications in the estimation of monthly average daily
global solar radiation.

KeywordsEmpirical Models, Daily Global Solar
Radiation, Sunshine Hours, Ranchi, I ndia, regression
constant.
I. INTRODUCTION
Knowledge of global solar radiation is essential in the
prediction, study and design of the economic viability
of systems which use solar energy. Information on
global solar radiation received at any site (preferably
gained over a long period) should be useful not only
to the locality where the radiation data is collected but
also for the wider world community. A global study
of the world distribution of global solar radiation
requires knowledge of the radiation data in various
countries and for the purpose of world wide
marketing, the designers and manufacturers of solar
equipment will need to know the average global solar
radiation available in different and specific regions.
Obviously, measured data is the best form of this
knowledge. Unfortunately, there are very few
meteorological stations that measure global solar
radiation, especially in developing countries. In
places where no measured values are available, a
common application has been to determine this
parameter by appropriate correlations which are
empirically established using the measured data.
Several empirical models have been used to calculate
solar radiation, utilizing available meteorological,
geographical and climatologically parameters such as
sunshine hours [1], [2], [3] air temperature [4],
latitude [5], precipitation[6], relative humidity [7], [8],
and cloudiness[9]. The most commonly used
parameter for estimating global solar radiation is
sunshine duration. Among various correlations, the
modified version of Angstrom equation who proposed
a linear relationship between the ratio of average daily
global radiation to the corresponding value on a
completely clear day and the ratio of average daily
sunshine duration to the maximum possible sunshine
duration and its derivations have been widely used.
Furthermore, these correlations can be used to
estimate the values of the monthly average daily
global radiation for other locations on the basis of
their similarities. Consequently, the bright sunshine
data at these places can be used to estimate the values
of the global radiation with the help of models
detailed in literatures.
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 4 April 2012
http://www.ijsat.com
133


Nomenclature
H
o
extra terrestrial radiation
H
g
global solar radiation
H
d
diffuse solar radiation
I
sc
the solar constant
| the latitude
the solar declination angle

ss
the sunset hour angles
n the monthly mean daily number of hour of
observed sunshine hours
N the monthly mean value of day length
MBE Mean Bias Error
MPE Mean Percentage Error
RMSE Root Mean Square Error
NSE Nash-Sutcliffe Equation

II. METHODS OF PREDICTION
A. Radiation Calculation
a. Prediction of Monthly Average Daily Global
Solar Radiation
Among the above mentioned empirical models, the
most popular is the regression equation of the
Angstrom type [10].
|
|
.
|

\
|
+ =
N
n
b a
H
H
o
g
(1)
where H
g
is the monthly average daily global solar
radiation falling on a horizontal surface at a particular
location. H
o
is the monthly mean daily radiation on a
horizontal surface in the absence of atmosphere. n is
the monthly mean daily number of hour of observed
sunshine hours , N is the monthly mean value of day
length at a particular location. a and b are
climatologically determined regression constant. n/N
is often called the percentage of possible sunshine
hour.
b. Prediction of Monthly Mean Daily Radiation
on Horizontal Surface (H
o
)
For (1) the value of H
o
is determined using (1.8.3) of
Duffi and Beckman [11].
| | sin sin cos cos sin n
365
360
cos 0.033 1 I

24
H ss ss sc o | o | +
|
|
.
|

\
|
+ =
(
(
(

(2)
Where I
sc
is the solar constant,
|
is the latitude, is
the solar declination,
ss
is the sunset hour angles
where
( )
(

+ = n 284
365
360
sin 45 . 23 o
(3)
o | e tan tan cos = ss (4)
ss N e
15
2
= (5)
B. Different Regression Models
The regression models proposed in the literature based
on Angstrom and Prescott [10] and other parameters
are listed:
TABLE I. REGRESSION EQUATION FOR DIFFERENT MODEL
TYPE
Model
No.
Regression equation Model type Source

1

H/Ho=a + b(n/N)

Linear
Angstrom,
(1924)[10] and
Prescott
(1940)[12]

2


H/Ho=a+ b(nN)+c(n/N)
2


Quadratic
Akinoglu and
Ecevit
(1990)[3]
3 H/Ho=a + b(n/N)+c
log(n/N)
Linear
logarithmic
Newland
(1988)[13]

4

H/Ho=a + b log(n/N)

Logarithmic
Ampratwum
and Dorvlo
(1999)[14]
5 H/Ho=a + b(n/N)+c
exp(n/N)
Linear
exponential
Kadir Bakirci
(2009)
6 H/Ho=0.18+0.62(n/N) Linear, known
constants
Rietveld
(1978)[6]
7 H/Ho= 0.29cos() +
0.52 (n/N)
Linear,
latitude
related
Glover and
McCulloch
(1958)[15]
III. COMPARISON OF MODELS
There are many parameters which deal with the
assessment and comparison of monthly mean daily
solar radiation estimation models. Here the statistical
parameters like the mean bias error (MBE) and the
root mean square error (RMSE) helps to calculate the
error or the deviation of the calculated value from the
measured value. Mean percentage error (MPE) and
coefficient of correlation (r) tests the linear
relationship between predicted and measured values.
The best result is when these statistics are closer to
zero, but coefficient of correlation, r, should approach
to 1 as closely as possible for better modeling. To
improve the results and better comparison the Nash
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 4 April 2012
http://www.ijsat.com
134

Sutcliffe equation (NSE) is also selected as an
evaluation criterion. A model is more efficient when
NSE is closer to 1. The errors that have been
estimated help to compare the models but they do not
make the model statistically significant. The t-statistic
allows models to be compared and at the same time it
is carried out to determine statistical significance of
the predicted values by the models.
a. The Mean Bias Error
( )

=
n
meas i calc i
H H
n
MBE
1
, ,
1

This test provides information on long-term
performance. A low MBE value is desired. A negative
value gives the average amount of underestimation in
the calculated value. So, one drawback of these two
mentioned tests is that overestimation of an individual
observation will cancel underestimation in a separate
observation.
b. The Mean Percentage Error
( )
100 *
1
(%)
1 ,
, ,

|
|
.
|

\
|
=
n
meas i
meas i calc i
H
H H
n
MPE


c. The Root Mean Square Error
( )
2
1
2
1
, ,
1
(
(

=

n
meas i calc i
H H
n
RMSE

The value of RMSE is always positive, representing
zero in the ideal case. The normalized root mean
square error gives information on the short term
performance of the correlations by allowing a term by
term comparison of the actual deviation between the
predicted and measured values. The smaller the value,
the better is the models performance.
d. The NashSutcliffe Equation
( )
( )
2
1
,
2
1
, ,
1

=
n
meas i
meas
n
meas i calc i
H H
H H
NSE

where meas H is the mean measured global radiation. A
model is more efficient when NSE is closer to 1 [16].



e. The Coefficient of Correlation
The coefficient of correlation, r can be used to
determine the linear relationship between the
measured and estimated values.
f. t-Statistic Test
As defined by Student [17] in one of the tests for
mean values, the random variable t with n1 degrees
of freedom may be written here as follows:
2
1
2 2
2
) ( ) (
) )( 1 (
(
(

=
MBE RMSE
MBE n
t

The smaller the value of t the better is the
performance. To determine whether a models
estimates are statistically significant, one simply has
to determine, from standard statistical tables, the
critical t value. For the models estimates to be judged
statistically significant at the calculated t value must
be less than the critical value.
IV. RESULTS AND DISCUSSIONS
The accuracy of different models was determined
using the data obtained from Indian Metrological
Department[18]. The values of monthly mean daily
global solar radiation intensity estimated using the
above 7 models. Models 1 to 7 were compared with
the corresponding measured values. The values
obtained from the new proposed model were also
estimated. The statistical tests of MBE, MPE, RMSE,
NSE, r and t-test were determined for the entire
period. The results are summarized. The regression
coefficient for the different models were taken for the
tropics as suggested by Udo[19].
TABLE 2: THE REGRESSION COEFFICIENTS FOR RANCHI FOR
DIFFERENT MODELS ARE:
Model
No.
Model Type a b c
1 Linear 0.2207 0.5249
2 Quadratic 0.2299 0.5137 0.01104
3 Linear
logarithmic
0.2239 0.5098 0.01644
4 Logarithmic 0.6013 0.3271
5 Linear
exponential
0.19449 0.4771 0.02994
New
Model
Linear 0.279 0.399





International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 4 April 2012
http://www.ijsat.com
135


TABLE 3: MODELS VALIDATIONS UNDER DIFFERENT STATISTICAL HEADS
Statistical Parameters Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

New Model
MBE 0.51 0.73 0.18 0.14 0.54 0.99 1.92

-0.01
MPE(%) 3.1 4.46 1.13 1.62 3.29 5.74 11.6

0.50
RMSE 0.69 0.88 0.5 1.67 0.72 1.24 2.01

0.90
NSE 0.94 0.9 0.97 0.52 0.93 0.85 0.51

0.89
r 0.97 0.96 0.99 0.99 0.97 0.95 0.9

0.99
t-value 3.64 4.93 1.28 0.28 3.76 4.39 10.70

0.03


Figue 1.Comparision of global radiation using different models(monthwise)
From the above Table 3 it can be seen that the Mean
Bias Error is close to zero as desired. Only model 6
and model 7 has error near 1. Also it can be seen that
model 6 and model 7 has maximum percentage error.
The other statistical parameters Root Mean Square
Error and NashSutcliffe Equation shows there is
permissible error in the measured value using the
different models. The value of r shows the linear
dependency of the measured value on the calculated
value and it should be equal to 1 which is true
throughout in the above used models. The t-value is
also acceptable in all the cases other than that of
model 7. The new values of regression constant as
proposed in this paper has the best results with
minimum error when compared to others and
maximum linear dependency (r-value). The results
confirm that the regression constant as shown in Table
2 are well and good in calculating the global radiation
of Ranchi. And can be used to get very accurate
results in this part of India.
The measured and calculated values of the monthly
average daily global radiation using models 1 to 7 for
the Ranchi is illustrated in Figure 1. As can be seen
from Figure 1, agreement between the values obtained
from models (1 to 6) and the measured data are good
for all the months of the year. It is clear that the
deviation between the measured and calculated values
using model 7 in Figure 1 is very larger than that of
others. Therefore, it has been concluded that models 1
to 6 can be recommended for use to estimate monthly
mean daily global solar radiation at Ranchi.
0
5
10
15
20
25
30
1 2 3 4 5 6 7 8 9 10 11 12
A
v
e
r
a
g
e
R
a
d
d
i
a
t
i
o
n

i
n


K
J
/
M
2
-
d
a
y

1
2
3
4
5
6
7
meas
new
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 4 April 2012
http://www.ijsat.com
136


Figure 2. Linear Dependency of New Model vs Measured Value.
From Figure 2 it is noted that the proposed model
nearly follows the measured value throughout the
year. But during the rainy season there is a dip in the
graph and the measured and estimated values do not
overlap each other. This is due to the fact that the
diffused radiation is not taken into account while
proposing the new model.
V. CONCLUSIONS
The objective of this study was to evaluate various
models for the estimation of the monthly average
daily global radiation on a horizontal surface from
bright sunshine hours and to propose a new and
appropriate model for Ranchi. All available empirical
models that can be used to estimate monthly average
daily global solar radiation over Ranchi in India that
have been collected from different research papers to
evaluate the applicability of these models. The
collected models were compared on the basis of the
statistical error tests such as mean bias error (MBE),
the mean percentage error (MPE), root mean square
error (RMSE), NashSutcliffe equation (NSE),
correlation coefficient (r) and the t-test. According to
the results, models 1 to 6, which are based on
Angstrom and Prescott [10] model showed the good
estimation of the monthly average daily global solar
radiation on a horizontal surface for Ranchi. But the
best result is obtained using the proposed regression
constants. The present work will help to advance the
state of knowledge of global solar radiation to the
point where it has applications in the estimation of
daily global solar radiation.

REFERENCES
[1] Koussa M., Malek A., Haddadi M. 2009. Statistical
comparison of monthly mean hourly and daily diffuse and
global solar irradiation models and a Simulink program
development for various Algerian climates, Energy Conversion
and Management, 50, 1227-1235
[2] Bulut H., Buyukalaca O. 2007. Simple model for the generation
of daily global solar-radiation data in Turkey, Applied Energy,
84, 477491
[3] Akinoglu B.G., Ecevit A. 1990. Construction of a quadratic
model using modified Angstrom coefficients to estimate global
solar radiation, Solar Energy, 45, 8592.
[4] Fletcher A.L., 2007. Estimating daily solar radiation in New
Zealand using air temperatures, New Zealand Journal of Crop
Horticultural Science, 35, 147-157
[5] Raja I.A. 1994. Insolation sunshine relation with site elevation
and latitude, Solar Energy, 53, 5356
[6] Rietveld M.R., 1978. A new method for estimating the
regression coefficients in the formula relating solar radiation to
sunshine, Agricultural Meteorology, 9, 243252
[7] Trabea A.A., Shaltout M.A.M. 2000. Correlation of global
solar radiation with meteorological parameters over Egypt,
Renewable Energy, 21, 297308
[8] Alnaser W.E., 1993. New model to estimate the solar global
irradiation using astronomical and meteorological parameters,
Renew Energy, 3, 175177
[9] Kumar R., Umanand L. 2005. Estimation of global radiation
using clearness index model for sizing photovoltaic system,
Renewable Energy, 30, 22212233
[10] Angstrom A., 1924. Solar and terrestrial radiation, Quarterly
Journal of Royal Meteorological Society, 50, 121125
[11] Duffie, JA. and Beckman, WA., 1991. Solar Engineering of
Thermal Processes. John-Wiley & Sons, New York.
[12] Prescott J.A. 1940. Evaporation from water surface in relation
to solar radiation, Transactions of the Royal Society of
Australia, 46, 114118
[13] Newland F.J. 1988. A study of solar radiation models for the
coastal region of South China, Solar Energy, 31, 227235.
[14] Ampratwum D.B., Dorvlo A.S.S. 1999. Estimation of solar
radiation from the number of sunshine hours, Applied Energy,
63,161167
[15] Glover J., McCulloch J.S. 1958. The empirical relation between
solar radiation and hours of sunshine, Journal of Royal
Meteorological Society, 84, 172175.
[16] Chen R., Ersi K., Yang, J. Lu S., Zhao W., 2004.Validation of
five global radiation models with measured daily data in China,
Energy Conversion and Management, 45, 17591769.
[17] Bevington P.R. 1969. Data reduction and error analysis for the
physical sciences, first ed. McGraw Hill Book Co., New York.
[18] Solar Radiation Hand Book. 2008. Solar Energy Centre,
MNRE Indian Metrological Department. 13.
[19] Udo, SO. 2002. Contribution to the relationship between Solar
Radiation and sunshine duration in the tropics: A study case of
experimental data at Illorin, Nigeria. Turkish J. Physics. 26:
229.
0
5
10
15
20
25
1 2 3 4 5 6 7 8 9 10 11 12
meas
new
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 4 April 2012
http://www.ijsat.com
137

[20] Togral, IT. and Hasan, Togrul. 2002. Global solar radiation
over Turkey. Comparison of predicted and measured data.
Renewable Energy. 25: 55
[21] F. Kreith and J. F. Kreider, Principles of Solar Engineering,
Chapter 2, McGraw-Hill, New York(1978).
[22] Tiwari, GN. and Suleja, Sangeeta. 1997. Solar Thermal
Engineering System, Narosa Publishing House, New Dehli,
India.
[23] Bakirci, K. 2009. Correlations for estimation of daily global
solar radiation with hours of bright sunshine in Turkey, Energy,
34, 485-501.
[24] Samuel T.D.M.A. 1991. Estimation of global radiation for Sri
Lanka, Solar Energy, 47, 333337
[25] Sopian, K. Othman, M.Y. 1992. Estimation of monthly average
daily global solar radiation in Malaysia, Renewable Energy, 2,
319-325








































[26] Zekai S., 2008. Solar energy fundamentals and modeling
techniques: atmosphere, environment, climate change and
renewable energy, first ed. Springer, London.
[27] Black, JN., Bonython, CW. and Prescett, JA. 1954. Solar
radiation and duration of Sunshine, Q.J.R Metero. Soc. 80:
231-235.
[28] Glover, J. and McCulloch, F. 1958. The empirical relationship
between solar radiation and hours of sunshine. Q.J.R. Met.
Soc. 84(359): 56-60.
[29] Liu, YH. and Jordan, RC. 1960. The inter relationship and
characteristic distribution of direct, diffuse and total solar
radiation from metrological data. Solar Energy. 4: 1-19.
[30] Chandal, SS., Agarwal, RK., Pandey, AN. 2005. New
correlation to estimate Global Solar Radiation on Horizontal
Surface using sunshine Hours and Temperature data for Indian
Cities. Jour. of Solar Energy Engineering. 127: 417-420.