A 16-year dataset (2000–2015) of high-resolution (3 h, 10 km) global surface solar radiation

The recent release of the International Satellite Cloud Climatology Project (ISCCP) HXG cloud products and new ERA5 reanalysis data enabled us to produce a global surface solar radiation (SSR) dataset: a 16year (2000–2015) high-resolution (3 h, 10 km) global SSR dataset using an improved physical parameterization scheme. The main inputs were cloud optical depth from ISCCP-HXG cloud products; the water vapor, surface pressure and ozone from ERA5 reanalysis data; and albedo and aerosol from Moderate Resolution Imaging Spectroradiometer (MODIS) products. The estimated SSR data were evaluated against surface observations measured at 42 stations of the Baseline Surface Radiation Network (BSRN) and 90 radiation stations of the China Meteorological Administration (CMA). Validation against the BSRN data indicated that the mean bias error (MBE), root mean square error (RMSE) and correlation coefficient (R) for the instantaneous SSR estimates at 10 km scale were − 11.5 W m−2, 113.5 W m−2 and 0.92, respectively. When the estimated instantaneous SSR data were upscaled to 90 km, its error was clearly reduced, with RMSE decreasing to 93.4 W m−2 and R increasing to 0.95. For daily SSR estimates at 90 km scale, the MBE, RMSE and R at the BSRN were −5.8 W m−2, 33.1 W m−2 and 0.95, respectively. These error metrics at the CMA radiation stations were 2.1 W m−2, 26.9 W m−2 and 0.95, respectively. Comparisons with other global satellite radiation products indicated that our SSR estimates were generally better than those of the ISCCP flux dataset (ISCCP-FD), the global energy and water cycle experiment surface radiation budget (GEWEX-SRB), and the Earth’s Radiant Energy System (CERES). Our SSR dataset will contribute to the land-surface process simulations and the photovoltaic applications in the future. The dataset is available at https://doi.org/10.11888/Meteoro.tpdc.270112 (Tang, 2019).

The main inputs were cloud optical depth from ISCCP-HXG cloud products, the 23 water vapor, surface pressure and ozone from ERA5 reanalysis data, and albedo and 24 aerosol from Moderate Resolution Imaging Spectroradiometer (MODIS) products. 25 The estimated SSR data was evaluated against surface observations measured at 42

103
Although each of these have been widely used in various fields, the spatial resolutions 104 (>=100 km) of these SSR products is too coarse to meet the requirements of 105 high-resolution SSR data. A high-resolution (5 km, 3 hours) global SSR product of 106 the Global Land Surface Satellite (GLASS) were recently released, but it contains 107 data spanning only three years (Zhang et al., 2014). The GLASS SSR products were 108 retrieved by a look-up table method with the visible band top-of-atmosphere (TOA) 109 radiance from multi-source geostationary and polar-orbiting satellite data. Tang et al.

110
(2016) also produced a high-resolution SSR product (5 km, 1 hour) by combining data 111 from polar-orbit and geostationary satellites, but the product covers only China and 112 the dataset spans only eight years.

123
Section 3 describes the input data we used for SSR estimation and the observations 124 data used for SSR validation. In Section 4, we presented our evaluation of the SSR 125 product and compared it with other satellites products. Data availability is given in 126 Section 5, and Section 6 presents some conclusions and explores future work to 127 further improve SSR products.  The second data type we used was the new ERA5 reanalysis data. Three 168 variables of the ERA5 reanalysis data were used: surface pressure, total column water 169 9 vapor and total column ozone. The resolutions of the ERA5 reanalysis data are 1 h 170 and 25 km. To derive the same spatial resolution as the ISCCP-HXG cloud product, 171 we re-sampled the three variables of ERA5 reanalysis data to a spatial resolution of 10 172 km.

173
The third data type comprised aerosol and albedo data. The MODIS aerosol     upscaling of spatial resolution would tend to decrease these time mismatches.

236
To further illustrate this issue, the performances of our instantaneous SSR with 237 different spatial resolutions at the 42 BSRN stations were given in Table 1, which 238 suggests that the accuracy was clearly improved when the data were upscaled to 30 239 km, with a further slight improvement at 70 km, but that accuracy started to decrease 240 at 90 km. The performance of the ISCCP-FD was also presented in Table 1. 241 Apparently, the accuracy of our estimated instantaneous SSR is significantly higher 242 than that of the ISCCP-FD. A further advantage of our dataset is that its spatial 243 resolution is far higher than that of the ISCCP-FD products.  Table 2 also lists the performances of our daily SSR estimate with 267 different spatial resolutions and the performance of the ISCCP-FD daily SSR product.

268
Our estimates of daily SSR at all spatial resolutions were clearly more accurate than 269 13 that of ISCCP-FD, and they obviously improved when upscaled to more than 30 km.

270
The spatial distribution of RMSE for our estimated daily SSR at spatial 271 resolution of 90 km (Fig. 5)  and daily SSR at 10 km spatial resolution were 108.1 and 36.5 W m -2 , respectively, 285 both of which are greater than those of GWEX-SRB. However, when we upscaled our 286 estimated SSR to 90 km scale, RMSEs for our instantaneous and daily SSR were 287 lower, 82.4 and 30.6 W m -2 , respectively, indicating that our estimates of SSR were 288 more accurate than those of GEWEX-SRB at the same spatial resolution. We also 289 compared the performance of our estimates of SSR with that of CERES 290 (SYN1deg_Ed4A, Fig. 6). The accuracies of CERES were generally higher than those 291 of ISCCP-FD at both instantaneous and daily scales, but obviously lower than those 292 of our estimates at all spatial resolutions from 10 to 110 km ( Fig. 6 and Table 2).

293
Thus, our estimated SSR based on ISCCP-HXG cloud products provided a more  The MBE, RMSE and R for our estimated daily SSR at 10 km spatial resolution were . However, the RMSE for the GEWEX-SRB daily SSR is clearly higher 309 than that of our estimate of daily SSR at 90 km spatial resolution, thus indicating that 310 the accuracy of our daily SSR estimates is superior to that of the GEWEX-SRB daily 311 SSR product at the same spatial resolution.
312 Table 3 shows that the accuracy of our estimates of daily SSR clearly improved 313 when upscaled to 30 km spatial resolution and were most accurate at 90 km spatial  (Table 4). 326 The performances for CERES daily and monthly SSR were evaluated against 327 observations at the 90 CMA radiation stations (Fig. 10) and also compared with 328 those of our estimates from ISCCP-HXG (Table 4). The MBEs for CERES daily and 329 monthly SSR were greater than those of our estimates at all scales, and the RMSE 330 for CERES daily SSR was slightly smaller than that of our estimates at 10 km spatial 331 resolution, but obviously greater that our estimates at spatial resolutions from 30 to 332 110 km. The RMSE for CERRES monthly SSR was greater than those of our 333 estimates at all scales. Thus, the accuracy of our estimates is generally higher than 334 that of CERES. it can be seen that the global distribution for our SSR estimate based on the ISCCP-340 HXG cloud products is almost the same as that of the ISCCP-FD SSR product, but the 341 spatial resolution of our estimate is far higher than that of ISCCP-FD. There 342 is no doubt that we can get more details that the coarse resolution product ISCCP-FD 343 can not capture. For example, the region of high SSR clearly identified over the Tibetan Plateau by our estimate (Fig. 11a) is barely discernible in the 345 ISCCP-FD-derived data (Fig. 11b)