Data Paper Zone II Versions EN1 Vol 2 (3) : 1-12 2017
Reconstructed data of photosynthetically active radiation in China (1961 – 2014)
: 2017 - 04 - 13
: 2017 - 05 - 04
: 2017 - 09 - 19
2013 5 0
Abstract & Keywords
Abstract: Photosynthetically active radiation (PAR) is a key factor in the disciplinary research of ecology, agriculture, climatology and so on. It is a fundamental physiological variable that reveals the exchange process of matter and energy. As an important climatic resource, PAR plays an indispensable role in the simulation of photosynthetic potential, yield potential, crop growth and soil carbon sequestration. To create this data set, we first obtain the historical data of surface solar radiation simulated from the conventional observation data of China Meteorological Administration. Then, the historical data are combined with the PAR estimation module, which includes clearness index, solar altitude and sunshine duration, to reconstruct the daily PAR data of 724 observation stations from 1961 to 2014.
Keywords: China; photosynthetically active radiation; hybrid model; reconstructed data; clear sky index
Dataset Profile
Chinese title1961~2014年中国光合有效辐射重构数据集
English titleReconstructed data of photosynthetically active radiation in China (1961 – 2014)
Corresponding authorHu Bo (
Data producersLiu Hui, Tang Liqin, Hu Bo, Xiao Tiangui, Liu Guangren, Wang Yuesi, Shi Yingying
Time range1961 –2014
Geographical scopeChinese mainland
Data format*.txtData amount69.5 MB
Data service system<>
Sources of fundingScience and Technology Service Network Initiative of Chinese Academy of Sciences (STS Plan, KFJ-SW-STS-168);
National Key R&D Plan (Quantitative Relationship and Regulation Principle Between Regional Oxidation Capacity of Atmospheric and Air Quality, 2017YFC0210003)
Dataset compositionThe dataset includes 724 txt files and one excel file. Each txt file contains the daily average PAR data reconstructed for a single station, of which the file name is the site number. The excel file consists of information on the 724 stations.
1.   Introduction
Photosynthetically active radiation (PAR) refers to the part of solar radiation with the spectral range from 400 nm to 700 nm that can be absorbed by plant chlorophyll in the process of photosynthesis. PAR is an energy source for the formation of dry matter in plants and an indicator of the energy absorption ability of vegetation to photosynthetically active radiation. It is also a vital factor to drive the exchange process of matter and energy in ecosystems. As an important climatic resource, PAR constitutes an indispensable data type in researches of potential photosynthesis, potential yield, crop growth and soil carbon sequestration. Also, PAR is one of the key parameters of the terrestrial and marine ecosystems carbon cycle. Precise PAR measurement is significant for correcting NPP and CO2 source/sink models, which makes them closer to actual changes of the earth-atmosphere system. The spatial and temporal distribution of PAR provides a basis for accurate estimation of the ecosystem carbon cycle using models. Therefore, accurate PAR data are fundamental for the research of ecosystem carbon flux and the modeling study of ecology, agronomy and climate change.
In 1984, difference method started to be applied in PAR measurement in China, but a routine network for PAR measurements was not yet established. In 1997, satellite observations and simulations were used to estimate accurate PAR values assisted by radioactive transfer models. The variation characteristics of PAR in Tibetan Plateau, Haibei Alpine Meadow Area, northeast China, and central China have been investigated by means of directly observed PAR data.3–6 Zhu et al.7 developed a semi-empirical model for estimating PAR in China, and reported its spatio-temporal variation characteristics in recent 50 years. Meanwhile, many empirical models for estimating PAR have been established based on the ratio of PAR to broadband solar radiation in different climate regions.8–12 To date, studies of PAR are largely of regional coverage, so are the empirical models for estimating PAR. Long-term nation-scale PAR data still remain scarce, and the simultaneous observation of broadband solar radiation and PAR data is required for extended application. Large-scale PAR observation and estimation need to be carried out.
Chinese Ecosystem Research Network (CERN) is a national observation system, whereby China's ecological environment and major resource problems are observed and analyzed on a long-term basis. CERN provides data and theoretical support to the study of major ecological problems in China through its research on water, soil, atmosphere and biology. The meteorological radiation observation system of CERN is the first ground observation network of PAR on a national scale. It is composed of 44 observation stations distributed across China’s eight typical ecological types (farmland, forest, city, grass, lake, desert, wetlands, and gulf). These stations have begun working since 2004. This study uses long-time span observation data of 39 stations to develop a PAR estimation method and evaluate PAR data, leaving aside short-time span observation data of the other five stations. Previous research outcomes have indicated that the data of CERN can reflect the temporal and spatial variation of PAR across China.13–14 For more precise spatio-temporal resolution, the "hybrid model"15–16 and a reconstruction module were used to calculate PAR data based on the routine observation data of CMA, as well as CERN data on PAR and total radiation, aerosol optical depth, total ozone volume data, and astronomical solar radiation. According to climatic characteristics, administrative divisions and other findings reported by Shen et al.,15 China was divided into eight regions: northeast China (NEC), north China plain (NCP), east China (EC), southeast China (SEC), north central China (NC), southwest China (SWC), northwest China (NWC) and the Tibetan Plateau (TP). Figure 1 shows the distribution of CMA and CERN sites as well as the regional partition of the PAR reconstruction module, with black solid line designating regional boundaries.

Figure 1   Distribution of CMA and CERN sites and regional partition of the PAR reconstruction module
2.   Data collection and processing
After the upgrade of its meteorological radiation observation systems in 2004, CERN had all its sensors meet the WMO standards, whereby high-precision and stable long-term data could be obtained. In order to gain long-term, high spatial resolution PAR data, the "hybrid model" was used to calculate historical broadband solar radiation based on CMA routine observations (such as sunshine duration, pressure, temperature and relative humidity), MODIS AOD data and NASA/GSFC O3 data. Data observed from the 39 CERN sites were used to develop the model for calculating PAR under all-sky conditions. Then the estimation model was combined with clearness index, solar elevation angle and sunshine duration to obtain the historical daily PAR data of the 724 routine stations of CMA from 1961 to 2014. Figure 2 shows the data acquisition process.

Figure 2   Data acquisition process
The "hybrid model" developed by Yang Kun et al.16–17 is a semi-physical and semi-empirical model, which retains the simplicity of the Ångström model while containing the physical process of radiation transmission. The process can be described as bellow. Due to Rayleigh scattering, ozone absorption, aerosol absorption and scattering, water vapor absorption and permanent gas absorption, solar radiation would be attenuated when it goes through the atmosphere in clear sky. The five transmittance functions are expressed as \({\tau }_{r}, {\tau }_{oz}, {\tau }_{a}, {\tau }_{w}, {\tau }_{g}\), respectively. Solar beam radiative transmittance (\({\tau }_{b}\)) and solar diffuse radiative transmittance (\({\tau }_{d}\)) can be calculated by Equations (1) and (2):
\[{\tau }_{b}={\tau }_{r}{\tau }_{oz}{\tau }_{a}{\tau }_{w}{\tau }_{g}\]
\[{\tau }_{d}=0.5\text{ }\left[{\tau }_{oz}{\tau }_{g}{\tau }_{w}\left(1-{\tau }_{a}{\tau }_{r}\right)\right]\]
The five types of transmittance can be calculated by ground pressure, atmospheric precipitation, concentration of ozone volume and atmospheric turbidity β. Detailed calculation process can be found in Yang et al.16–17 Daily surface radiation (\({R}_{clear}\)) under clear sky can be obtained by Equation (3):
\[{R}_{clear}={\int }_{{t}_{1}}^{{t}_{2}}\left({\tau }_{b}+{\tau }_{d}\right){R}_{0}dt\]
where \({R}_{0}\) is the solar radiation at the top of the atmosphere, \({t}_{1}\) is sunrise and \({t}_{2}\) is sunset.
The cloud transmittance (\({\tau }_{c}\)) can be gained by the sunshine duration, and the parameterization scheme is shown in Equation (4), where n is actual sunshine duration, \({N }_{s}\) is the maximum possible sunshine duration, the length of time for which solar direct normal irradiance exceeds a threshold value of 120 W·m-2 in clear-sky conditions. Under cloudy skies, daily surface radiation \({R}_{s}\) can be calculated by Equation (5).
\[{\tau }_{c}=0.2505+1.1468n/{N}_{s}-0.3974{\left(n/{N}_{s}\right)}^{2}\]
\[{R}_{s}={\tau }_{c}\cdot {R}_{clear}\]
Because PAR variation is consistent with the change of solar radiation, some scholars have established a PAR parameter estimation model with clear sky index (\({K}_{s}\)) and sine of solar elevation angle (\(\mu\)). Clear sky index can be obtained by Equation (6).
The relationship between hourly mean PAR and \(\mu\) is expressed by the power law equation (7), where \({PAR}_{m}\) is the maximum value of PAR per \({K}_{s}\), and e determines how PAR varies with \(\mu\). The dependence of \({PAR}_{m}\) on \({K}_{s}\) is described by Equation (8), where a, b, c, d are the fitting parameters.
\[PAR=PA{R}_{m}×{\mu }^{e}\]
The daily value of PAR is calculated by Equation (9). \(\overline{{K}_{s}}\)is the ratio of the daily surface radiation to the daily extraterrestrial solar irradiance; \(\overline{\mu}\) is the daily average value of the sine of the solar elevation angle from sunrise to sunset; \({t}_{d}\) is the daily sunshine duration; A, B, C, D, E are parameters relying on climatic zones.
\[PA{R}_{daily}=\left(A+B×\overline{{K}_{s}}+C×{\overline{{K}_{s}}}^{2}+D×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{E}×{t}_{d}\]
The parameterization schemes for PAR data reconstruction in China's eight regions are shown in Table 1.
Table 1   PAR estimation models for the eight regions
Typical stationRegionEstimation equations
HailunNEC\(\left(0.28+9.01×\overline{{K}_{s}}+2.03×{\overline{{K}_{s}}}^{2}-1.89×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.19}×{t}_{d}\)
Beijing Forrest StationNCP\(\left(0.03+10.57×\overline{{K}_{s}}-4.44×{\overline{{K}_{s}}}^{2}+3.37×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.06}×{t}_{d}\)
ShapotouNC\(\left(0.24+10.18×\overline{{K}_{s}}+1.43×{\overline{{K}_{s}}}^{2}-1.78×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.24}×{t}_{d}\)
FukangNWC\(\left(0.44+7.97×\overline{{K}_{s}}+5.84×{\overline{{K}_{s}}}^{2}-5.42×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.12}×{t}_{d}\)
LhsasaTP\(\left(2.67-5.83×\overline{{K}_{s}}+30.42×{\overline{{K}_{s}}}^{2}-19.37×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.14}×{t}_{d}\)
YantingSWC\(\left(0.20+9.22×\overline{{K}_{s}}+1.34×{\overline{{K}_{s}}}^{2}-1.43×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.25}×{t}_{d}\)
Dinghu MountainSEC\(\left(0.07+9.47×\overline{{K}_{s}}-2.10×{\overline{{K}_{s}}}^{2}+2.26×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.06}×{t}_{d}\)
Lake DongEC\(\left(0.18+9.26×\overline{{K}_{s}}+0.91×{\overline{{K}_{s}}}^{2}-1.01×{\overline{{K}_{s}}}^{3}\right)×{\overline{\mu }}^{1.18}×{t}_{d}\)
3.   Sample description
3.1   Description of data content
Presentation of PAR data abides by the following rules:
"Reconstructed data of photosynthetically active radiation in China (1961 – 2014). rar" is composed of 724 txt files of reconstructed PAR data from 724 sites and one excel file of site information. The text file was named "*.txt", where * represents station codes. The four columns in the text file represented year, month, day and reconstructed daily PAR values (mol/(m2·d)), respectively. The three columns in the excel file represented the codes, longitudes and latitudes of stations, respectively.
3.2   Description of data sample
Taking the Beijing site of CMA (54511) as an example, Table 2 presents the reconstructed PAR values for the station in June 2014 with the four columns representing year, month, day and reconstructed daily PAR values, respectively. Figure 3 shows the annual variation of reconstructed PAR values in 2014 at Beijing station (54511). The nationwide spatial distribution of reconstructed PAR values in 2014 is presented in Figure 4.
Table 2   Reconstructed PAR values for Beijing station (54511) in June 2014
YearMonthDayPAR (mol/( m2·d))

Figure 3   Annual variation of reconstructed PAR values in 2014 at Beijing station (54511)

Figure 4   Spatial distribution of reconstructed PAR data at CMA stations in 2014, China (Unit: mol/(m2·d))
4.   Quality assessment
The reconstructed PAR data are validated against the PAR data measured in situ from 2005 to 2014 at 39 adjacent CERN stations. X axis represents the reconstructed data of CMA site, and Y axis represents the observation data of CERN station. The fitting results are shown in Table 3. Slope (S), intercept (I), significance test (P) and mean bias error (MBE) are used as benchmarks for the radiation products. The mean bias error is defined in Equation (10), where \({E}_{i}\) and \({M}_{i}\) are the calculated and observed values of the i-th sample respectively, \({M}_{ave}\) is the average of the observed values, and \(N\) is the number of observations.
\[MBE\left(\%\right)=\frac{100}{{M}_{ave}}\left( \frac{\sum _{i=1}^{i=N}\left({E}_{i}-{M}_{i}\right)}{N} \right)\]
Table 3   Comparison of reconstructed PAR data and adjacent station observations
51628AKA1.1292.860P < 0.01-18.32
56856ALF0.8954.641P < 0.01-6.31
53845ASA0.9471.811P < 0.01-0.08
54518BJC0.8914.686P < 0.01-6.30
56959BNF0.7937.076P < 0.01-5.59
54285CBF0.9922.026P < 0.01-5.83
51828CLD1.0175.581P < 0.01-18.18
58259CSA1.0634.456P < 0.01-20.10
53929CWA0.9883.979P < 0.01-10.93
59278DHF1.1181.726P < 0.01-16.56
57494DHL1.0423.926P < 0.01-17.12
59493DYB0.8996.229P < 0.01-12.96
53545ESD0.9543.874P < 0.01-7.03
51076FKD0.8948.149P < 0.01-11.63
57091FQA1.1843.287P < 0.01-25.02
56374GGS0.9563.844P < 0.01-9.54
52765HBG0.7887.836P < 0.01-2.38
59023HJA0.8943.002P < 0.01-3.02
50756HLA0.8955.588P < 0.01-7.61
59478HSF0.8993.658P < 0.01-5.35
57745HTF0.9983.069P < 0.01-11.65
54857JZB1.0822.342P < 0.01-15.22
53698LCA1.0321.744P < 0.01-8.83
55591LSA0.8946.968P < 0.01-8.78
52546LZD1.1152.839P < 0.01-17.19
56188MXF1.2691.878P < 0.01-26.58
54226NMD1.0223.744P < 0.01-12.74
54102NMG0.8967.566P < 0.01-12.26
57799QYA1.0312.918P < 0.01-12.91
50788SJM0.9555.412P < 0.01-12.79
57355SNF1.2790.435P < 0.01-22.90
53704SPD1.0136.072P < 0.01-18.85
54342SYA1.0332.662P < 0.01-11.52
59948SYB0.9298.004P < 0.01-18.45
58358THL1.4772.803P < 0.01-36.88
57662TYA1.0981.678P < 0.01-14.29
54715YCA1.0043.339P < 0.01-11.81
57306YGA1.0792.246P < 0.01-16.22
58626YTA1.0243.778P < 0.01-14.47
As shown in Table 3, the slope values of the linear fitting for 39 sites center around 1, with only BNF and HBG stations lower than 0.8 and MXF, SNF and THL stations higher than 1.2. The intercept is less than 8 except SYB. For all sites, the values of P are less than 0.01, indicating that all fitting results have passed 99% of the significance test. And the MBE values are negative at all stations, which indicates that the reconstructed PAR values are underestimated slightly compared with the observation data, but only THL and MXF exceed 25%. Site location and underlying surface have an impact on the observation of surface radiation and PAR data, which leads to the high observation values at the THL station of CERN located on Taihu Lake, and the MXF site located halfway up the mountain at an elevation of 1,826 m. This accounts for the high MBE between THL and site 58358, between MXF and site 56188. All in all, the above statistical results show that the reconstructed PAR values are reliable.
5.   Usage notes
PAR plays a very important role in the quantitative estimation of photosynthesis and the exploration of bioavailability of solar energy. It can also help improve the accuracy of global ecosystems carbon estimation and provide theoretical basis for the scientific development of agriculture. This dataset reflects long-term temporal and spatial changes of PAR, which can be used in climate resource assessment nationwide. PAR can be used to improve physical parameters of the radiation process in the climate prediction model. Users should pay attention to the data unit, which is mol / (m2·d).
We thank CERN for providing the radiation observation data, and CMA for supplying global solar radiation data and meteorological elements. We also gratefully acknowledge the MODIS Science Team for providing the AOD dataset and the NASA/GSFC Ozone Processing Team for supplying the ozone data.
Dong T & Meng J. Overview on the estimation of photosynthetically active radiation. Progress in Geography 30 (2011): 1125 – 1134.
Miskolczi F, Aro T, Iziomon M et al. Surface radiative fluxes in Sub-Sahel Africa. Journal of Applied Meteorology 36 (2010): 521 – 530.
Zhang X & Zhou Y. Climatological estimation of photosynthetically active quantum values in the Tibetan Plateau from April to October. Acta Geographica Sinica 52 (1997): 361 – 365.
Li Y & Zhou H. Features of photosynthetic active radiation (PAR) in Haibei alpine meadow area of Qilian Mountain during plant growing period. Plateau Meteorology 21 (2002): 90 – 95.
Liu X & Fan L. The calculation methods and distributive character of solar radiation in Liaoning Province. Resources Science 24 (2002): 82 – 87.
Zhang Y & Qin B. The basic characteristic and climatological calculation of the photosynthetically available radiation in Taihu Region. Acta Energiae Solaris Sinica 23 (2002): 118 – 123.
Zhu X & He H. Spatio-temporal variation characteristics of photosynthetically active radiation in China in recent 50 years. Acta Geographica Sinica 65 (2010): 270 – 280.
Li R & Ji G. The observation on PAR coefficient over northern part of Tibetan Plateau. Acta Energiae Solaris Sinica 28 (2007): 241 – 247.
Zhang X, Zhang Y & Zhou Y. Measuring and modeling photosynthetically active radiation in Tibet Plateau during April-October. Agricultural & Forest Meteorology 102 (2000): 207 – 212.
Yu Y & Chen H. Observations and analyses on photosynthetically active radiation (PAR) in Xianghe Region. Plateau Meteorology 27 (2008): 1094 – 1101.
Hu B & Wang Y. Observational study of variation of photosynthetically active radiation in the Sanjiang Plain. Climatic and Environmental Research x (2014): 343 – 350.
Wang L & Wei G. Measurement and estimation of photosynthetically active radiation from 1961 to 2011 in Central China. Applied Energy 111 (2013): 1010 – 1017.
Hu B, Liu H & Wang Y. Investigation of the variability of photosynthetically active radiation in the Tibetan Plateau, China. Renewable & Sustainable Energy Reviews 55 (2016): 240 – 248.
Hu B. Establishing a solar radiation measurement network in Chinese ecosystem research network and study on the distribution characteristics of ultraviolet and photosynthetically radiate in China. Beijing: Institute of Atmospheric Physics, Chinese Academy of Sciences, 2005.
Shen X, Liu B, Li G et al. Spatiotemporal change of diurnal temperature range and its relationship with sunshine duration and precipitation in China. Journal of Geophysical Research: Atmospheres 119 (2014): 13,163 – 13,179.
Yang K, Huang G & Tamai N. A hybrid model for estimating global solar radiation. Solar Energy 70 (2001): 13 – 22.
Yang K, Koike T & Ye B. Improving estimation of hourly, daily, and monthly solar radiation by importing global data sets. Agricultural & Forest Meteorology 137 (2006): 43 – 55.
Data citation
1. Liu H, Tang L, Hu B et al. Reconstructed data of photosynthetically active radiation in China (1961 – 2014). Science Data Bank. DOI: 10.11922/sciencedb.400
Article and author information
How to cite this article
Tang L, Liu H, Hu B et al. Reconstructed data of photosynthetically active radiation in China (1961 – 2014). China Scientific Data 2 (2017), DOI: 10.11922/csdata.170.2017.0135
Tang Liqin
paper writing.
MSc candidate; research area: atmospheric radiation.
Liu Hui
data processing and paper writing.
PhD candidate; research area: atmospheric radiation and remote sensing.
Hu Bo
data quality control.
PhD, Professor; research area: atmospheric radiation.
Xiao Tiangui
data quality control.
PhD, Professor; research area: climate change and simulation.
Liu Guangren
data quality control.
BS; research area: communications engineering.
Wang Yuesi
observation network design.
PhD, Professor; research area: atmospheric chemistry.
Shi Yingying
data reduction.
MSc; research area: atmospheric aerosol optical properties.
Science and Technology Service Network Initiative of Chinese Academy of Sciences (STS Plan, KFJ-SW-STS-168);National Key R&D Plan (Quantitative Relationship and Regulation Principle Between Regional Oxidation Capacity of Atmospheric and Air Quality, 2017YFC0210003)
Publication records
Published: Sept. 19, 2017 ( VersionsEN1
Released: May 4, 2017 ( VersionsZH1
Published: Sept. 19, 2017 ( VersionsZH2