Content uploaded by Joon-Bum Jee

Author content

All content in this area was uploaded by Joon-Bum Jee on Apr 12, 2016

Content may be subject to copyright.

Asia-Pac. J. Atmos. Sci. pISSN 1976-7633 / eISSN 1976-7951

DOI:10.1007/s13143-014-0047-0

Development of GWNU (Gangneung-Wonju National University) One-Layer

Transfer Model for Calculation of Solar Radiation Distribution of the Korean

Peninsula

Il-Sung Zo1, Joon-Bum Jee2, and Kyu-Tae Lee1

1Department of Atmospheric and Environmental Sciences, Gangneung-Wonju National University, Gangneung, Korea

2Weather Information Service Engine, Center for Atmospheric sciences and Earthquake Research, Seoul, Korea

(Manuscript received 24 October 2013; accepted 28 July 2014)

© The Korean Meteorological Society and Springer 2014

Abstract: Gangneung-Wonju National University (GWNU) one-

layer solar radiation model is developed in order to resolve the lack

of the vertical structure of atmospheric components and fast

calculation with high horizontal spatial resolution. GWNU model is

based on IQBAL and NREL methods and corrected by precise multi-

layer Line-By-Line (LBL) model. Further, the amount of solar

radiation reaching the surface by using 42 types of vertical atmos-

pheric data as input data was compared with detailed models and

one-layer models. One-layer solar radiation models were corrected

depending on sensitivity of each input data (i.e., total precipitable

water, ozone, mixed gas, and solar zenith angle). Global solar radiation

was calculated by corrected GWNU solar model with satellites

(MODIS, OMI and MTSAT-2), KLAPS model prediction data in

Korea peninsula in 2010, and the results were compared to surface

solar radiation observed by 22 KMA solar radiation sites. Calculated

solar radiation annually accumulated showed highest solar radiation

distribution in Andong, Daegu, and Jinju regions, meanwhile the

observation data showed lower solar radiation in Daegu region

compared to model result values.

Key words: One-layer solar radiation model, line-by-line model,

correction, global solar radiation, vertical atmospheric data

1. Introduction

Solar radiation reaching the surface of the earth not only

serves as a crucial role in climate change of global atmosphere

but also is extensively used for industrial activity of human

beings. In particular, as solar energy, one of clean energy

resources, is attenuated by 65% when radiant energy (65,700

TW/year) released from the sun passes atmosphere, the

amount reaching the surface of the earth is about 23,000 TW/

year, a more amount compared to other types of energy

sources (Perez and Perez, 2008). Thus, many countries in the

world show their great interest and have heavily invested in

energy power projects using solar energy. Germany Trade &

Invest(GTAI), a leader in photovoltaic generation, reported that

a growing trend of Germany's annual solar radiation capacity

reached 7.6 GW in 2012, and approximately 180,000 photo-

voltaic systems were newly operated for a year, which proves

photovoltaic generation sharply increases.

The U. S. Department of Energy's National Renewable En-

ergy Laboratory (NREL) developed the Climatological Solar

Radiation (CSR) model in 1995 to calculate daily accumulated

solar radiation (George and Maxwell, 1999). A solar energy

map was produced at 10 km ×10 km spatial resolution by

using the State University of New York (SUNY)'s Satellite

Radiation Model (Perez et al., 2002) developed by NREL and

Perez, and was analyzed by using the US Climate Reference

Network (USCRN) data and the Typical Meteorological Year

(TMY) Ver.3 model. The German Weather Service (Deutscher

Wetterdienst, DWD) completed a solar energy map for the first

time in 1999 by using observation data on monthly and annual

average solar radiation daily accumulated from 1981 to 1998,

and a solar energy map using a solar radiation model was

developed by using relevant meteorological and meteorological

satellite observation data based on Kerschgens et al. (1978)'s

two-stream method. Further, a solar energy map was produced

by using the European Center for Midrange Weather Forecasts

(ECMWF) data and the Model Output Statics (MOS) method,

and additional studies that calculate photovoltaic power gener-

ation are being conducted. Recently, new technic, research

actively conducted using the neural network method (Takenaka

et al., 2011).

Solar energy that is released from the sun and reaches the

surface of the earth changes with astronomical conditions (i.e.,

location change of the sun and the earth), distribution of the

earth's atmospheric components (e.g., water vapor, mixed gas,

aerosol, cloud, etc., Yeom et al., 2012; Lee et al., 2013), and

the earth's surface conditions (i.e., geography and topography).

It is important to consider multiple scattering between sub-

strates by dividing atmosphere into multiple layers and input-

ting observation data by altitude. However, as stations ob-

serving the atmosphere's vertical structure are limited, and

observation for calculating radiation characteristics does not

exist except for specific observation, calculation for high-

resolution solar radiation requires a one-layer solar radiation

model that assumes atmosphere as a one-layer to calculate.

Therefore, this study herein corrects the existing Gangneung-

Wonju National University (GWNU; hereafter GWNU) one-

Corresponding Author: Joon-Bum Jee, Weather Information Service

Engine, SAIT Tower 12 Fl., 434 Worldcupbukro, Mapo-gu, Seoul

121-835, Korea

E-mail: rokmcjjb717@gmail.com

ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES

layer solar radiation model's calculation process on transmit-

tance and solar radiation by comparing results derived from a

multiple-layer solar radiation model. The input data for solar

radiation model operation calculated a correction equation by

using the atmosphere's 42 types of vertical data. The correction

equation calculated solar energy distribution of the Korean

peninsula by applying the GWNU one-layer solar radiation

model, and was analyzed by comparing observation data. The

research results will resolve deficiency of input data to calculate

solar radiation models and will be used for super resolution

and accurate solar energy calculation in an efficient manner.

2. Solar radiation model

Solar radiation released from the sun and reaching the surface

are classified as direct and diffuse solar radiation by physical

process, and these elements can be observed in a separate

manner (i.e., direct, diffuse and global solar radiation) on the

surface. In other words, the solar radiations reaching the

surface are observed and attenuated by clouds, aerosol, and gas

components in the atmosphere, which are varied by wavelength

(see Fig. 1). The ozone layer in the stratosphere absorbs some

parts of visible and ultraviolet ray areas at 0.4 µm or less,

while water vapor and carbon dioxide absorb infrared wave-

length regions at 0.75 µm or more.

a. Multi-layer line-by-line solar radiation model

Solar radiation energy entering at top of atmosphere is

absorbed, diffused, and reflected by gas components, clouds,

and aerosol in the atmosphere, and the surface. Under the

assumption of a plane-parallel atmosphere, provided that the

definition of solar zenith angle (θ), azimuth (φ), and µ≡cosθ

on radiation process directions is used, a general solar

radiation transfer equation is as follows (Dave, 1974).

.(1)

In this equation,τλ stands for optical thickness by wavelength

and Jλ stands for a source function.

In atmosphere, although solar radiation transfer equation can

be simplified by neglecting the Planck function, it is not easy

to find accurate values with the equation if multiple scattering

is included, resulting from many layers and the surface of the

earth. However, provided that a phase function (Pv) does not

change depending on azimuth (φ), downward flux of radiation

in a τ layer can be specified as follows.

F↓(τ)= . (2)

Herein, F0 stands for solar radiation flux entered in atmos-

phere, and as solar radiation calculated in this equation

drastically changes by wavelength, the Line-by-Line (LBL)

method that calculates by densely dividing wavelength spacing

was used to accurately calculate transfer equation values for

solar radiation, meanwhile Stamnes et al. (1988)'s discrete

ordinate approximation was used for transmission and diffusion

calculation by atmospheric components.

This method is based on Chandrasekhar (1960), and with it,

a multiple scattering process of layer was included to calcu-

lation by Stamnes et al. For absorption line data, as intervals

by wavelength are not constant and are arranged in a random

manner, band models for the equation (2)'s transmission

function calculation produce errors in some parts. For a range

of solar radiation wavelength, coefficient by wavelength was

calculated at 0.002 cm−1 intervals from HITRAN2k absorption

line data (Rothman et al., 2003).

b. One-layer solar radiation model

The intensity (Iλ) of solar energy reaching the surface of the

earth can be written as follows based on Beer's law (Siegel and

Howell, 1981).

Igλ=Idλcosθ+Isλ.(3)

Herein, λ stands for wavelength, global solar radiation (Igλ)

can be divided into direct components (Idλ) and diffuse

components (ISλ), and provided that atmosphere is a one layer,

direct solar energy reaching the surface is calculated as the

following equation (4).

Idλ=I0λexp(−τλ). (4)

Herein, I0λ and τλ, respectively, stand for extraterrestrial radi-

ation and optical thickness resulting from absorption gas, and

atmospheric transmittance can be specified as tλ=exp(−τλ).

For absorption gas and particles, the following empirical

equations were used to obtain each transmittance ratio of

optical thickness on air molecule, aerosol, ozone, water vapor,

and mixed gas.

toλ= exp(−koλ lmo), (5)

µdIλτλµφ,,()

dτ

--------------------------- Iλτλµφ,,()Jλτλµφ,,()–=

vd

∫2πIλ

0

1–

∫τvµ,()µdµµ

0

+πF0eτµ

0

⁄–

Fig. 1. A comparison of the extraterrestrial spectrum with the diffuse,

direct, and global radiation (W m−2 µm−1) on June 15.

Il-Sung Zo et al.

twaλ= exp[−0.2385kwaλωmr/ (1 + 20.07kwaλωmr)0.45], (6)

trλ= exp(−0.008735maλ−4.08), (7)

taλ= exp[−β(λ/0.55)

−αmr], (8)

tgλ= exp[−1.41kgλma/ (1 + 118.93kgλma)0.45]. (9)

Herein, koλ, kgλ and kwaλ stand for classified wavelength ab-

sorption coefficient on ozone, mixed gas, and water vapor.

Provided that aerosol is less than 0.5 µm, α is 1.027, mean-

while 1.206 is used for others, and l and ω, respectively, stand

for vertical concentration and total precipitable water. Further,

mr, ma, and mo, respectively, stand for relative optical mass,

pressure-corrected value, and relative optical mass on ozone.

Diffuse solar radiation (Isλ) on the surface is changed by air

molecule, aerosol, and multiple scattering at the surface, which

can be calculated as follows.

Isλ= Irλ+Iaλ+Igλ. (10)

From the above equation, solar radiation diffused by Rayleigh

(Irλ), aerosol (Iaλ), and atmosphere (Igλ) is calculated by

empirical equations as follows.

Irλ=I0λE0cosθτ

oλτgλτwaλ [0.5(1 −τrλ)τaλ], (11)

Iaλ=I0λE0cosθτ

oλτgλτwaλ [Fcω0(1 −τ

aλ)τrλ], (12)

Igλ=Qλ[(ρgλ)/(1−ρgλ)], (13)

Qλ=(Irλ+Iaλ)+Idλ, (14)

=τoλτgλτwaλ[0.5(1 −τrλ)τaλ+(1−Fc)ω0(1 −τ

aλ)τrλ]. (15)

Herein, E0 stands for eccentricity, and Fc stands for a ratio of

forward scattering on the total energy scattering, meanwhile

1−Fc stands for backward scattering. ω0 stands for a single

scattering albedo. and ρgλ, respectively, stand for multiple

scattering by all the absorption gases in atmosphere and

surface albedo.

The GWNU model used for this study was based on the

IQBAL model, which is a solar radiation model corrected by

using a multiple solar radiation model after adapting and

applying the NREL's aerosol process method. The IQBAL

model was produced on a basis of Iqbal's theory (Iqbal, 1983)

and the NREL model is a one-layer model Bird and Riordan

(1986) produced by the National Renewable Energy Labora-

tory (NREL) for development and to compare observed values,

and the two above-stated models are a spectrum model calcu-

lating solar radiation by wavelength. Further, BIRD (Bird and

Hulstrom, 1981) stands for a one-layer solar radiation model to

develop wavelength ranges of solar radiation as a one band.

3. Correction of one-layer solar radiation model

The τλ of the equation (4) in section 2.b stands for optical

thickness, which is a function of absorption coefficients (e.g.,

koλ, kgλ, kwaλ, etc.) of pressure components, and basically, a

function of altitude or pressure, but the GWNU one-layer solar

radiation model does not include change in altitude or pressure

of absorption coefficient because atmosphere was assumed as

a single layer to save calculation time and calculation re-

solution. Therefore, for this study, methods correcting one-

layer solar radiation models of the equations from (3) to (15),

were used by comparing detailed solar radiation models

including altitude or pressure effects by dividing atmosphere

into multiple layer.

To operate accurate solar radiation models, vertical atmos-

pheric distribution data require input data, and reference atmos-

phere data (i.e., standard atmosphere data by latitude) are

mostly used. To compare calculation results of radiation

models, Garand et al. (2001)'s 42 types of vertical atmospheric

data were used for this study. The 42 types of vertical

atmospheric data include not only 6 standard atmospheres (i.e.,

standard, tropical, middle latitude summer, middle latitude

winter, sub-arctic summer, and sub-arctic winter) but also

distribution by altitude observed at many stations of the earth.

ρaλ

′ρaλ

′

ρaλ

′

ρaλ

′

Fig. 2. Global solar radiation of LBL solar radiation model (black point) and error between GWNU and LBL solar radiation

model for transmittance (red triangle) with total precipitable water (a) and total ozone amount (b).

ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES

For total precipitable water, total ozone amount and mixed

gas, the results and the error of transmittance derived from the

LBL model were shown by using 42 types of vertical

atmospheric data. Total precipitation water (a) and total ozone

amount (b) of Fig. 2 showed that solar radiation and the error

of transmittance decreased depending on change in the amount

of input data, indicating that the one-layer model calculates

transmittance more depending on the amount of absorber

compared to detailed models. In other words, for the amount

of gas absorbed in the same atmosphere, the one-layer model

calculates less transmittance compared to detailed models.

Therefore, transmittance calculated in the one-layer model

should be corrected by the amount of gas absorption so that

transmittance derived from total precipitable water and total

ozone amount can be decreased to the same amount of detailed

models. Transmittance was corrected by using a third-degree

polynomial to obtain the amount of total precipitable water as

the fluctuation of its error is large compared to total ozone

amount, meanwhile a second-degree polynomial was used to

obtain the amount of total ozone amount.

Further, the mixed gas of Fig. 3 was calculated on the

assumption that the mixture ratio in atmosphere is constant.

And as the amount of solar radiation change resulting from the

amount of mixed gas is not large, the relationship of global

solar radiation and temperature of the surface was examined.

As a result, correlation coefficient showed a relatively high

correlation as 0.97. Therefore, for correction derived from the

amount of mixed gas, transmittance was corrected by surface

temperature by applying a second-degree polynomial, which is

the same method applied to total precipitable water and total

ozone amount. When it comes to correction of global solar

radiation depending on solar zenith angle, as shown in Table 1

below, surface albedo was each corrected at 0.1 interval of

cosine the solar zenith angle by using the rate of global solar

radiation. The interaction between atmosphere and the ground

surface is sensitive to change in solar zenith angle, and the

LBL model and the error increase as solar zenith angle

increases. This is caused by while the LBL model considers a

mixture ratio by altitude and accumulates calculated values to

calculate, the one-layer solar radiation model considers multiple

scattering and path length of absorption gas by considering a

one-layer to calculate the amount of solar radiation. At a part

with small solar zenith angle, although solar radiation corrected

Fig. 3. Same as Fig. 2 except for surface air temperature (K).

Tabl e 1 . The error correction equation between GWNU and LBL

solar radiation model for global solar radiation with cosine of solar

zenith angle.

Solar zenith angle Equation (quadratic polynomial)

0.0 y = 1.0083 + (−0.0062/x) + (−2.8119e −5/x2)

0.1 y = 1.0080 + (−0.0063/x) + (−2.6584e −5/x2)

0.2 y = 1.0077 + (−0.0063/x) + (−2.4944e −5/x2)

0.3 y = 1.0074 + (−0.0064/x) + (−2.3077e −5/x2)

0.4 y = 1.0071 + (−0.0065/x) + (−2.1002e −5/x2)

0.5 y = 1.0067 + (−0.0066/x) + (−1.8675e −5/x2)

0.6 y = 1.0063 + (−0.0067/x) + (−1.6053e −5/x2)

0.7 y = 1.0059 + (−0.0068/x) + (−1.3253e −5/x2)

0.8 y = 1.0054 + (−0.0069/x) + (−9.8984e −6/x2)

0.9 y = 1.0049 + (−0.0070/x) + (−6.1025e −6/x2)

Table 2. Surface solar radiation (W m−2) by solar radiation models with surface albedo and mid-latitude summer atmosphere. And differences (%)

between LBL and solar radiation models.

Surface albedo

Multi-layer

LBL model

(W m−2)

Difference(%)

Multi-layer One-layer

Band model Spectral model Band model

NASA IQBAL NREL GWNU BIRD

0.0 1060.1 −0.121 −2.274 −1.654 −1.654 −0.812

0.1 1067.7 −0.078 −2.308 −1.613 −0.211 0.512

0.2 1075.5 −0.031 −2.344 −1.547 −0.217 1.031

0.3 1083.7 0.019 −2.383 −1.453 −0.215 1.822

0.4 1091.4 0.074 −2.425 −1.329 −0.216 2.585

0.5 1101.0 0.133 −2.470 −1.174 −0.210 3.412

0.6 1110.3 0.196 −2.519 −0.986 −0.208 4.051

0.7 1119.9 0.265 −2.573 −0.761 −0.211 4.717

0.8 1130.1 0.339 −2.632 −0.498 −0.209 5.581

0.9 1140.8 0.419 −2.697 −0.132 −0.213 6.833

Il-Sung Zo et al.

Fig. 4. Global solar radiation (W m−2) by LBL model and box-plots of differences (%) between solar radiation models and

LBL model with atmospheres from Garand et al. (2001).

ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES

Fig. 5. Monthly accumulated surface solar radiations (unit: MJ m−2) calculated by GWNU model with 1 km ×1km

resolution (Jan. to Dec., 2010).

Il-Sung Zo et al.

is small, a second-degree polynomial regression equation was

applied as errors radically occur when the solar zenith angle is

big. As shown in Fig. 2 and 3, and Table 1, the GWNU model

was established for this study as a conclusive solar radiation

model, by including corrected regression equation to solar

radiation models of the equations (3)-(9) depending on ab-

sorption gas and surface albedo.

Out of calculation results derived from the multi-layer LBL

radiation model, the multi-layer band model (Chou and Suarez,

1999; hereafter NASA), and 4 types of one-layer solar radi-

ation model (e.g., IQBAL, NREL, and BIRD) including the

GWNU model, results of mid-latitude summer atmosphere

were shown in Table 2. Surface temperature of mid-latitude

summer atmosphere is 294.2 K, total precipitable water is 2.91

cm, total ozone amount is 330.5 DU, and solar zenith angle is

0o, and the flux of extraterrestrial solar radiation used 1366.05

Wm

−2. Results showed that the amount of solar radiation on

the surface increased by surface albedo, and the GWNU was

the most similar model out of one-layer models, and the Bird

model, one-layer band model, showed the biggest difference.

When surface albedo is 0.0, the value of solar radiation

calculated on the surface and the difference between NASA,

IQBAL, NREL, and GWNU were shown in Fig. 4 as box-plot

after changing the cosine value of solar zenith angle into 1.0,

0.8, 0.6, and 0.4. It is analyzed that the values calculated in the

LBL model were calculated dependently on amounts of ozone

and total precipitable water that are input data, and the smallest

difference showed in NASA, followed by GWNU, NREL and

IQBAL. Comparing with calculation results derived from the

LBL model showed that while the error increased in the

GWNU as albedo increased, and the error was below 0.50%

on average. The error of IQBAL and NREL model, similar

kinds, was 2% or more. Further, the error of one-layer solar

radiation model increased as solar zenith angle increased

(cosine values decreased). This occurs when the increasing

ratio on length of optical path is bigger in the one-layer model,

and the amount of absorber in layer equally increases.

4. Results

The GWNU model, a corrected one-layer solar radiation

model, calculated surface solar radiation by inputting real

atmospheric conditions obtained from model forecasting and

satellite observations data of the Korean peninsula. To calculate

the amount of solar radiation reaching the surface of the earth,

required are amounts of gas absorbing solar radiation such as

water vapor (or total precipitable water), ozone, etc., aerosol,

and cloud cover data, and also surface pressure and altitude at

calculation spots and surface albedo data. Out of these data, for

data on pressure and the amount of water vapor, etc., the Korea

Local Analysis and Prediction System (KLAPS), a regional

forecast model of the Korea Meteorological Administration

(KMA; hereafter KMA), was used, for the amount of ozone,

Fig. 6. Monthly accumulated cloud amounts (unit: okta) observation by 22 KMA solar sites (Jun. to Aug. 2010).

Fig. 7. KMA 22 solar radiation sites on Korea peninsula.

ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES

the Ozone Monitoring Instrument (OMI) sensor data (daily

average; when missing use to month average data) at 1o×1o

resolution was used, and for aerosol, showing very strong

characteristics in size, shape, and region, MODIS satellite data

(daily average; when missing use to month average data) at

1o×1o resolution were used. Further, for surface albedo data,

used were MODIS's high resolution (0.05o×0.05o resolution)

data, and for the digital elevation model, used were 3-second

data (about 90 m resolution) of NASA's Shuttle Radar Topo-

graphy Mission (SRTM) from United State Geological Survey

(USGS). For cloud data, one of the most important factors

attenuating solar radiation energy reaching the surface, MTSAT-

2 satellite data were processed by a method used in Kawamura

et al. (1998) and Communication, Ocean and Meteorological

Satellite Data Processing System (Korea Meteorological Ad-

ministration, 2009). Cloud data (hereafter cloud factor) using

visible and infrared spin scan radiometer data supplied by the

MTSAT-2 satellite and solar zenith angle. Cloud factor is the

observation and calculation (in clear sky) on cloudy pixel. And

look up table of cloud factor is calculated by Gauss-Jordan

elimination (Gilbert, 2003) and multiple regression (Cohen et

al., 2003) methods with 1o of solar zenith angle and visible

albedo, respectively. The above-stated data were used in a way

of interpolation scheme in accordance with 1 km ×1km

resolution regarding the Korean peninsula, research area.

Analysis was conducted from January 2010 to December

2010, and spatial distribution of monthly accumulated solar

radiation calculated by the GWNU model, which was cor-

Fig. 8. Monthly accumulated surface solar radiations (unit: MJ m−2) observation by 22 KMA solar radiation sites in 2010.

Il-Sung Zo et al.

rected, was shown in Fig. 5. For monthly solar radiation

distribution of the Korean peninsula, seasonal and regional

distribution characteristics were distinctively shown by the

effect of cloud, solar zenith angle, aerosol, and the amount of

water vapor. In particular, as the solar zenith angle is small

during summer season, the solar radiation of surface is expected

to be strong. However, due to frequent clouds, monthly maxi-

mum accumulated solar radiation was shown in June, early

summer. That is, the reason that surface solar radiation is

strong compared to mid-summer (July to August) is that

although the sun's altitude was lower in June compared to mid-

summer season, the amount of clouds, the most important

factor in the attenuation effect, was less in June compared to

July and August, mid-summer season (5.75 okta in June, 7.34

okta in July, 6.88 okta in August, See Fig. 6). Further, the

lowest month of the year was December in the amount of

average monthly solar radiation, about 27% compared to June,

the highest month, analyzed that it is due to the solar zenith

angle is large and much cloud amount in Dec.. Fig. 7 indicates

that 22 solar radiation sites operated by the KMA, and out of

monthly accumulated data observed in the sites, main months

of season was shown in Fig. 8. Although the distribution of

solar radiation was similar with model calculation values in

Fig. 5, it partially shows a difference, which is resulting from

that as model calculation results are calculated at one-hour

intervals while observation data are hourly accumulated. How-

ever, this will be resolved, provided that collection intervals

are shortened on input data of solar radiation models including

cloud.

Further, Fig. 9 stands for calculation of the GWNU model

on the amount of solar radiation annually accumulated and

solar radiation observation data provided by the KMA's 22

sites. The amount of average solar radiation on the Korean

peninsula's calculated solar radiation was 4,800 MJ m−2, and

the average value of 22 observatories' data was 4,932 MJ m−2.

Results of model calculation showed that solar radiation annu-

ally accumulated was relatively less in the Korean peninsula's

western coast area as the area has more amount of cloud

compared to other regions, while the intensity of solar energy

was strong as Andong, Daegu, and Jinju show little amount of

cloud related to downwind location of Sobaek mountainous

areas. During the same period, although the KMA's solar

radiation shows a similar tendency, Daegu showed relatively

lower solar radiation compared to the model. This difference

can be analyzed and described through environment investiga-

tion of solar radiation observatories and comparative obser-

vation of pyranometer.

5. Summary

The one-layer solar radiation model was developed to resolve

deficiency of vertical atmospheric data and improve high-

resolution computing speed. The GWNU solar radiation model

was developed on a basis of the IQBAL and NREL theories, a

Fig. 9. Annual accumulated global solar radiations (unit: MJ m−2) of calculation by GWNU model (a) and observation by 22 KMA

solar radiation sites (b) in 2010.

ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES

one-layer solar radiation model, and the LBL was selected as a

reference model to improve accuracy. Further, the amount of

solar radiation reaching the surface of the earth by using 42

types of vertical atmospheric data as input data was compared

with detailed models and one-layer models. One-layer solar

radiation models were corrected depending on sensitivity of

each input data (i.e., total precipitable water, ozone, mixed gas,

and solar zenith angle). Analysis results derived from the

GWNU solar radiation model showed that a difference showed

0.5% or less compared to the LBL model, which is a similar

value with the NASA model, a multi-layer model, and the

error increased by solar zenith angle, which was lower com-

pared to other one-layer solar radiation models.

By using satellite and numerical model data as input data,

calculated was solar radiation reaching the surface in the

Korean peninsula for one year in 2010, which were compared

with surface observation data. The results showed a similar

distribution with observation data, partially showing a differ-

ence, which was caused by a time difference between model

and observation data. This is analyzed as an error occurring,

resulting from that the observation data are accumulated by

time meanwhile the model is calculated at hourly intervals. As

a factor affecting the most is cloud, June least affected by

cloud during summer showed high solar radiation compared to

July and August of mid-summer, due to change in cloud.

Calculated solar radiation annually accumulated showed hig-

hest solar radiation distribution in Andong, Daegu, and Jinju

regions, meanwhile the observation data showed lower solar

radiation in Daegu region compared to model result values.

This difference can be analyzed and described through com-

parative observation conducted by solar radiation observation

stations on environment investigation and pyranometer.

In conclusion, the one-layer solar radiation model developed

herein in this study can partially resolve problems occurring in

input data of solar radiation models, and can be applied to

high-resolution calculation requiring much computation. Fur-

ther, it is considered that the one-layer solar radiation model

can be basic research for further renewable energy and

photovoltaic generation studies.

Acknowledgments. This work is funded by the Korea Meteor-

ological Administration Research and Development Program

under the Weather Information Service Engine (WISE) project

(Grant No. 153-3100-3133-302-350).

Edited by: Tadahiro Hayasaka

REFERENCES

Bird, R. E., and R. L. Hulstrom, 1981: Simplified clear sky model for

direct and diffuse insolation on horizontal surfaces, Technical Report

No. SERI/TR-642-761, Golden, CO: Solar Energy Research Institute,

46 pp.

______, and C. Riordan, 1986: Simple solar spectral model for direct and

diffuse irradiance on horizontal and tilted planes at the earth's surface

for cloudless atmospheres. J. Climate Appl. Meteor., 25, 87-97.

Chandrasekhar, S., 1960: Radiative Transfer. Dover Publications Inc., New

York, 393 pp.

Chou, M. D., and M. J. Suarez, 1999: A solar radiation parameterization

for atmospheric studies. NASA/TM-19990104606, 15.

Cohen, J., P. Cohen, S. G. West, and L. S. Aiken, 2003: Applied multiple

regression/correlation analysis for the behavioral sciences (3rd ed.).

Lawrence Erlbaum Associates, Inc. Mahwah, New Jersey, 702 pp.

Dave, J. V., 1974: A direct solution of the spherical harmonics approxi-

mation to the radiative transfer equation for an arbitrary solar elevation.

Part I: Theory. J. Atmos. Sci., 32, 790-798.

Garand, L., and Coauthors, 2001: Radiance and jacobian intercomparison

of radiative transfer model applied to HIRS and AMSU channels. J.

Geophys. Res., 24, 17-31.

George, R., and E. Maxwell, 1999: High-resolution maps of solar collector

performance using a climatological solar radiation model. Proceedings

of the 1999 Annual Conference, American Solar Energy Society,

Portland, Maine, 243-248.

Gilbert, S., 2003: Introduction to Linear Algebra (3rd ed.). Wellesley-

Cambridge Press, Wellesley, Massachusetts, 571 pp.

Iqbal, M, 1983: An introduction to solar radiation. Academic Press, New

York, 390 pp.

Kamamura, H., S. Tanahashi, and T. Takahashi, 1998, Estimation of

insolation over the Pacific Ocean off the Sanriku coast. J. Oceanogr.,

54, 457-464.

Kerschgens, M., U. Pilz, and E. Raschke, 1978: A modified two-stream

approximation for computations of the solar radiation budget in a

cloudy atmosphere. Te l l us, 30, 429-435.

Korea Meteorological Administration, 2009: Development of Meteor-

ological Data Processing System of Communication, Ocean and

Meteorological Satellite (V). Final Report, 513-538.

Lee, J.-S, W.-J. Choi, D.-R. Kim, S.-Y. Kim, C.-K. Song, J.-S. Hong, Y.-D.

Hong, and S.-J. Lee, 2013: The effect of ozone and aerosols on the

surface erythemal UV radiation estimated from OMI measurements.

Asia-Pac. J. Atmos. Sci., 49, 271-278.

Perez, R., and M. Perez, 2008: Making the case for solar energy. Daylight

& architecture magazine by velux. Autum 2008, 9, 8-17.

______, P. Ineichen, K. Moore, M. Kmiecik, C. Chain, R. George, and F.

Vignola 2002: A new operational model for satellite-derived

irradiances: Description and validation. Solar Energy, 73, 307-317.

Rothman, L. S., and Coauthors, 2003: The HITRAN molecular spec-

troscopic database: edition of 2000 including updates through 2001. J.

Quant Spectrosc RA, 82, 5-44.

Siegel R., and J. R. Hewell, 1981: Thermal radiation heat transfer.

Hemisphere Pub. Corp., Washington, 524 pp.

Stamnes, K., S. C. Tsay, W. Wiscombe, and K. Jayaweera, 1998:

Numerically stable algorithm for discrete-ordinate-method radiative

transfer in multiple scattering and emitting layered media. Appl. Opt.,

27, 2502-2509.

Takenaka, H., T. Y. Nakajima, A. Higurashi, A. Higuchi, T. Takamura, R.

T. Pinker, and T. Nakajima, 2011: Estimation of solar radiation using a

neural network based on radiative transfer. J. Geophys. Res., 116 , doi:

10.1029/2009JD013337.

Yeom, J.-M., K.-S. Han, and J.-J. Kim, 2012: Evaluation on penetration

rate of cloud for incoming solar radiation using geostationary satellite

data. Asia-Pac. J. Atmos. Sci., 48, 115-123.