Evaluation of Various Tree-Based Ensemble Models for Estimating Solar Energy Resource Potential in Different Climatic Zones of China

Solar photovoltaic (PV) electricity generation is growing rapidly in China. Accurate estimation of solar energy resource potential (Rs) is crucial for siting, designing, evaluating and optimizing PV systems. Seven types of tree-based ensemble models, including classification and regression trees (CART), extremely randomized trees (ET), random forest (RF), gradient boosting decision tree (GBDT), extreme gradient boosting (XGBoost), gradient boosting with categorical features support (CatBoost) and light gradient boosting method (LightGBM), as well as the multi-layer perceotron (MLP) and support vector machine (SVM), were applied to estimate Rs using a k-fold cross-validation method. The three newly developed models (CatBoost, LighGBM, XGBoost) and GBDT model generally outperformed the other five models with satisfactory accuracy (R2 ranging from 0.893–0.916, RMSE ranging from 1.943–2.195 MJm−2d−1, and MAE ranging from 1.457–1.646 MJm−2d−1 on average) and provided acceptable model stability (increasing the percentage in testing RMSE over training RMSE from 8.3% to 31.9%) under seven input combinations. In addition, the CatBoost (12.3 s), LightGBM (13.9 s), XGBoost (20.5 s) and GBDT (16.8 s) exhibited satisfactory computational efficiency compared with the MLP (132.1 s) and SVM (256.8 s). Comprehensively considering the model accuracy, stability and computational time, the newly developed tree-based models (CatBoost, LighGBM, XGBoost) and commonly used GBDT model were recommended for modeling Rs in contrasting climates of China and possibly similar climatic zones elsewhere around the world. This study evaluated three newly developed tree-based ensemble models of estimating Rs in various climates of China, from model accuracy, model stability and computational efficiency, which provides a new look at indicators of evaluating machine learning methods.

Merging High-Resolution Satellite Surface Radiation Data with Meteorological Sunshine Duration Observations over China from 1983 to 2017

Surface solar radiation (Rs) is essential to climate studies. Thanks to long-term records from the Advanced Very High-Resolution Radiometers (AVHRR), the recent release of International Satellite Cloud Climatology Project (ISCCP) HXG cloud products provide a promising opportunity for building long-term Rs data with high resolutions (3 h and 10 km). In this study, we compare three satellite Rs products based on AVHRR cloud products over China from 1983 to 2017 with direct observations of Rs and sunshine duration (SunDu)-derived Rs. The results show that SunDu-derived Rs have higher accuracy than the direct observed Rs at time scales of a month or longer by comparing with the satellite Rs products. SunDu-derived Rs is available from the 1960s at more than 2000 stations over China, which provides reliable decadal estimations of Rs. However, the three AVHRR-based satellite Rs products have significant biases in quantifying the trend of Rs from 1983 to 2016 (−4.28 W/m2/decade to 2.56 W/m2/decade) due to inhomogeneity in satellite cloud products and the lack of information on atmospheric aerosol optical depth. To adjust the inhomogeneity of the satellite Rs products, we propose a geographically weighted regression fusion method (HGWR) to merge ISCCP-HXG Rs with SunDu-derived Rs. The merged Rs product over China from 1983 to 2017 with a spatial resolution of 10 km produces nearly the same trend as that of the SunDu-derived Rs. This study makes a first attempt to adjust the inhomogeneity of satellite Rs products and provides the merged high-resolution Rs product from 1983 to 2017 over China, which can be downloaded freely.

Comparison of Artificial Intelligence and Physical Models for Forecasting Photosynthetically-Active Radiation

Different kinds of radiative transfer models, including a relative sunshine-based model (BBM), a physical-based model for tropical environment (PBM), an efficient physical-based model (EPP), a look-up-table-based model (LUT), and six artificial intelligence models (AI) were introduced for modeling the daily photosynthetically-active radiation (PAR, solar radiation at 400–700 nm), using ground observations at twenty-nine stations, in different climatic zones and terrain features, over mainland China. The climate and terrain effects on the PAR estimates from the different PAR models have been quantitatively analyzed. The results showed that the Genetic model had overwhelmingly higher accuracy than the other models, with the lowest root mean square error (RMSE = 0.5 MJ m−2day−1), lowest mean absolute bias error (MAE = 0.326 MJ m−2day−1), and highest correlation coefficient (R = 0.972), respectively. The spatial–temporal variations of the annual mean PAR (APAR), in the different climate zones and terrains over mainland China, were further investigated, using the Genetic model; the PAR values in China were generally higher in summer than those in the other seasons. The Qinghai Tibetan Plateau had always been the area with the highest APAR (8.668 MJ m−2day−1), and the Sichuan Basin had always been the area with lowest APAR (4.733 MJ m−2day−1). The PAR datasets generated by the Genetic model, in this study, could be used in numerous PAR applications, with high accuracy.

Improvement in Surface Solar Irradiance Estimation Using HRV/MSG Data

The Advanced Model for the Estimation of Surface Solar Irradiance (AMESIS) was developed at the Institute of Methodologies for Environmental Analysis of the National Research Council of Italy (IMAA-CNR) to derive surface solar irradiance from SEVIRI radiometer on board the MSG geostationary satellite. The operational version of AMESIS has been running continuously at IMAA-CNR over all of Italy since 2017 in support to the monitoring of photovoltaic plants. The AMESIS operative model provides two different estimations of the surface solar irradiance: one is obtained considering only the low-resolution channels (SSI_VIS), while the other also takes into account the high-resolution HRV channel (SSI_HRV). This paper shows the difference between these two products against simultaneous ground-based observations from a network of 63 pyranometers for different sky conditions (clear, overcast and partially cloudy). Comparable statistical scores have been obtained for both AMESIS products in clear and cloud situation. In terms of bias and correlation coefficient over partially cloudy sky, better performances are found for SSI_HRV (0.34 W/m2 and 0.995, respectively) than SSI_VIS (−33.69 W/m2 and 0.862) at the expense of the greater run-time necessary to process HRV data channel.