Clear-sky Irradiance Research Articles

Assessment of clear-sky irradiance from 6S affected by local climatology of India

Accurate estimation of solar power generation is crucial for the effective planning and operation of solar energy sector. Solar potential can be estimated at a location from long-term historical irradiance data processed from atmospheric reanalysis datasets. The current study employs a clear-sky radiative transfer model (6S) to simulate the net surface irradiance at four distinct locations in India, comparing our model's output with mean monthly ground-based measurements of surface irradiance. Our findings reveal that the 6S model marginally underestimates solar irradiance, with potential deviations varying depending on the accuracy of input data. This research evaluates the influence of particulate matter concentration and relative humidity on the scattering and absorption of solar radiation. We also state the variability in surface irradiance across the country with the rainfall trends that enhance assessment accuracy. The study reports an annual deviation of 10–15 % in surface irradiance across the country, where rural settlements show lower deviations in simulations. The radiative transfer calculations mentioned in the study are simplistic yet beneficial for a priori evaluation of solar potential. Transmittance estimated from clear-sky models is further implemented in all-sky (with clouds), and thus, model accuracy is an important parameter. Additionally, we explore the trends in aerosol concentrations and the impact of local climatological factors on surface irradiance, providing insights critical for optimizing solar energy utilization.

Mapping of 10-km daily diffuse solar radiation across China from reanalysis data and a Machine-Learning method

Diffuse solar radiation (DSR) plays a critical role in renewable energy utilization and efficient agricultural production. However, there is a scarcity of high-precision, long-term, and spatially continuous datasets for DSR in the world, and particularly in China. To address this gap, a 41-year (1982–2022) daily diffuse solar radiation dataset (CHDSR) is constructed with a spatial resolution of 10 km, based on a new ensemble model that combines the clear-sky irradiance estimated by the REST2 model and a machine-learning technique using precise cloud information derived from reanalysis data. Validation against ground-based measurements indicates strong performance of the new hybrid model, with a correlation coefficient, root mean square error and mean bias error (MBE) of 0.94, 13.9 W m−2 and −0.49 W m−2, respectively. The CHDSR dataset shows good spatial and temporal continuity over the time horizon from 1982 to 2022, with a multi-year mean value of 74.51 W m−2. This dataset is now freely available on figshare to the potential benefit of any analytical work in solar energy, agriculture, climate change, etc (https://doi.org/10.6084/m9.figshare.21763223.v3).

Uncertainties in Clear-Sky Solar Irradiance Modeling Induced by the Limited Availability of the Atmospheric Parameters

Abstract In the recent years, great progress has been made in developing terrestrial and satellite-based networks for the measurement of atmospheric parameters, which serve as inputs in clear sky parametric models. However, some parameters may still be missing from the model input due to small temporal sampling or lack of in situ data. This paper presents a case study on the accuracy of the clear-sky solar irradiance estimation by a parametric model, when the availability of inputs is limited. The study was conducted with high-quality radiometric and atmospheric data recorded by the INOE-Magurele station, Romania. Seven different scenarios were studied. In each scenario the clear-sky solar irradiance model was run assuming that certain parameters are missing from the input. The results quantitatively confirm the dominant role of aerosols in establishing the accuracy of solar irradiance estimation under clear sky conditions. The unavailability of the measured Ångström turbidity coefficient may lead to a twofold/threefold increase in the uncertainty of direct-normal/diffuse solar irradiance estimation.

Solar thermal reactor model for pyrolysis of waste plastic

ABSTRACT The ability to convert waste plastic into combustible liquids and gases using solar energy could help transform the problem of disposal of non-recyclable plastic into a valuable and environmentally responsible source of fuel. The purpose of this study is to propose a practical model for a compound parabolic trough solar thermal reactor for pyrolysis of waste plastic. The model integrates predictions of energy available from solar radiation (at a given location, time of day and time of year) with parabolic trough collector orientation and efficiency, a transient energy balance for an evacuated tube reactor and pyrolysis kinetics of waste plastic. The experimental setup used to test the model includes a pyranometer, a commercial solar collector consisting of a 60 cm long evacuated tube with a compound parabolic reflector and multi-channel data loggers to collect temperature, humidity and radiation data. The solar radiation sub-model was found to be in excellent agreement with clear-sky irradiance data collected using the pyranometer. Predictions of reactor temperature and reaction rate were found to be sensitive to the concentrator aperture area, solar irradiance, type of plastic (Arrhenius kinetics) and radiation properties of the evacuated tube reactor but relatively insensitive to humidity, wind velocity and terrestrial irradiance. The model shows that even on a small scale, favourable conditions for pyrolysis of waste plastic can be achieved within a solar reactor.

SPARTA: Solar parameterization for the radiative transfer of the cloudless atmosphere

The high-performance SPARTA broadband clear-sky solar irradiance model is presented. It evaluates global (GHI), diffuse (DIF) and direct normal (DNI) irradiances from atmospheric transmittance functions developed using an advanced 2-band hybrid spectral integration scheme, calculates aerosol extinction using a universal and highly accurate aerosol transmittance scheme, and incorporates a versatile parameterization of the aerosol circumsolar solar irradiance. The model's performance is assessed against 1-min quality-assured measurements at three research-class radiometric ground stations spanning 15 years of data combined, and it is benchmarked against three high-performance radiative transfer models. The performance assessment proved that SPARTA was superior to the reference models for GHI, DIF and DNI at the three observational sites combined. SPARTA produced unbiased estimates of the three solar irradiance components (remarkably, it was the only model producing unbiased estimates of DIF) and yielded the smallest standard deviations of the model−observation differences (≈2 %, ≈10 % and ≈2 % for GHI, DIF and DNI, respectively). SPARTA was the only model capable of improving the estimation of DIF when using observed inputs of the aerosol scattering optical properties instead of modelled values. The results of the performance assessment proved that, provided accurate inputs to the model, SPARTA produces predictions with similar uncertainty than ground observations, especially if the ground sensors are not optimaly maintained or they are not A or B ISO-9060 class.

Mixed Multi-Pattern Regression for DNI Prediction in Arid Desert Areas

As a crucial issue in renewable energy, accurate prediction of direct normal solar irradiance (DNI) is essential for the stable operation of concentrated solar power (CSP) stations, especially for those in arid desert areas. In this study, in order to fully explore the laws of climate change and assess the solar resources in arid desert areas, we have proposed a mixed multi-pattern regression model (MMP) for short-term DNI prediction using prior knowledge provided by the clear-sky solar irradiance (CSI) model and time series patterns of key meteorological factors mined using PR-DTW on different time scales. The contrastive experimental results demonstrated that MMP can outperform existing DNI prediction models in terms of three recognized statistical metrics. To address the challenge of limited data in arid desert areas, we presented the T-MMP model involving combined transfer learning and MMP. The experimental results demonstrated that T-MMP outperformed MMP in DNI prediction by exploiting the significant correlation between meteorological time series patterns in similar areas for data augmentation. Our study provided a valuable prediction model for accurate DNI prediction in arid desert areas, facilitating the economical and stable operation of CSP plants.

Data-Driven Diagnosis of PV-Connected Batteries: Analysis of Two Years of Observed Irradiance

The diagnosis and prognosis of PV-connected batteries are complicated because cells might never experience controlled conditions during operation as both the charge and discharge duty cycles are sporadic. This work presents the application of a new methodology that enables diagnosis without the need for any maintenance cycle. It uses a 1-dimensional convolutional neural network trained on the output from a clear sky irradiance model and validated on the observed irradiances for 720 days of synthetic battery data generated from pyranometer irradiance observations. The analysis was performed from three angles: the impact of sky conditions, degradation composition, and degradation extent. Our results indicate that for days with over 50% clear sky or with an average irradiance over 650 W/m2, diagnosis with an average RMSE of 1.75% is obtainable independent of the composition of the degradation and of its extent.

Ten years of 1 Hz solar irradiance observations at Cabauw, the Netherlands, with cloud observations, variability classifications, and statistics

Abstract. Surface solar irradiance varies on scales down to seconds, and detailed long-term observational datasets of this variable are rare but in high demand. Here, we present an observational dataset of global, direct, and diffuse solar irradiance sampled at 1 Hz as well as fully resolved variability until at least 0.1 Hz over a period of 10 years from the Baseline Surface Radiation Network (BSRN) station at Cabauw, the Netherlands. The dataset is complemented with irradiance variability classifications, clear-sky irradiance and aerosol reanalysis, information about the solar position, observations of clouds and sky type, and wind measurements up to 200 m above ground level. Statistics of variability derived from all time series include approximately 185 000 detected events of both cloud enhancement and cloud shadows. Additional observations from the Cabauw measurement site are freely available from the open-data platform of the Royal Netherlands Meteorological Institute. This paper describes the observational site, quality control, classification algorithm with validation, and the processing method of complementary products. Additionally, we discuss and showcase (potential) applications, including limitations due to sensor response time. These observations and derived statistics provide detailed information to aid research into how clouds and atmospheric composition influence solar irradiance variability as well as information to help validate models that are starting to resolve variability at higher fidelity. The main datasets are available at https://doi.org/10.5281/zenodo.7093164 (Knap and Mol, 2022) and https://doi.org/10.5281/zenodo.7462362 (Mol et al., 2022); the reader is referred to the “Code and data availability” section of this paper for the complete list.

Determination of broadband atmospheric turbidity from global irradiance or photovoltaic power data using deep neural nets

Photovoltaic technologies provide significant capacity to electric grids, however, resource variability and production uncertainty complicate power balancing and reserve management. A crucial step in predicting solar generation is determining clear-sky irradiance. Clear-sky attenuation can be modeled using broadband atmospheric turbidity factors, but model accuracy is dependent on the measurements used to determine the current and future state of aerosol loading and water vapor content, which requires close proximity measurements, in time and space, to account for turbidity variability. Such measurements, though, are only available in near real-time at a limited, and decreasing, number of sites. This paper proposes a new method for estimating time-varying local turbidity conditions from more readily available pyranometer or PV output data. The method employs a long short-term memory recurrent neural network to distill the turbidity-driven signal from global irradiance (or global irradiance driven) observations, despite an inherent dampening issue. The method is developed to operate in near real-time for solar forecasting applications. Validation examines the ability of the method to (1) reproduce turbidity estimates derived from historical measurements of beam irradiance under clear-sky conditions; and (2) provide input for clear-sky models in the form of persistence forecasts generated from daily mean values.

A transferable turbidity estimation method for estimating clear-sky solar irradiance

A transferable turbidity estimation method is proposed for estimating the turbidity and clear-sky solar irradiance. Instead of using on-site irradiance measurements (i.e., the local model), a transferable model is developed involving stations with sufficient information, and then applied at locations with limited data availability. Compared with the local method, the transferable model yields results with slightly higher discrepancies regrading normalized root mean squared error (nRMSE, 2.80% vs 2.75%). When compared with the Ineichen–Perez (PVLIB) model, the nRMSE of clear-sky global horizontal irradiance (GHIcs) estimation is reduced from 4.99% to 2.44%, and the normalized mean bias error (nMBE) is improved from -3.37% to 0.57%. The GHIcs estimation is comparable with physical models (i.e., McClear and REST2), where the McClear produces a nRMSE of 3.32% and the nMBE is 2.10%, while the REST2 generates results with an nRMSE of 2.55% and an nMBE of 1.30%. We further compare aforementioned models for day-ahead GHIcs forecasts using a day persistent way. GHIcs forecast from the transferable method has slightly lower discrepancies of nRMSE and nMBE than the physical models. Considering the complexity of physical models, the transferable turbidity estimation method with comparable performance demonstrates valuable potential for solar resourcing and forecasting applications.

Performance assessment of clear-sky solar irradiance predictions using state-of-the-art radiation models and input atmospheric data from reanalysis or ground measurements

In this work, the performance of clear-sky direct normal irradiance (DNI) and global horizontal irradiance (GHI) predictions generated with three state-of-the-art solar radiation models with different degrees of complexity is assessed by comparison with high-quality measured irradiance data at Évora, Portugal. The libRadtran, SMARTS, and REST2 radiation models are alternatively operated using input data from three different data sources: co-located AERONET ground-based measurements, and CAMS and MERRA-2 gridded reanalysis data. For these nine combinations (three models and three data sources), the results are assessed using five statistical indicators, namely mean bias error (MBE), root mean square error (RMSE), fractional bias (FB), fractional gross error (FGE), and coefficient of determination (R2). Overall, it is found that AERONET is the data source that provides the best DNI estimates. In general, libRadtran and SMARTS produced closer estimates to the ground-based DNI observations. For GHI, however, no firm conclusion can be drawn regarding the best data source. MERRA-2 produces better estimates in combination with libRadtran and SMARTS according to all statistical indicators except R2, whereas AERONET is to be preferred according to FB, FGE, and R2 when using REST2. Curiously, the latter generates better GHI estimates despite being the simplest model. Overall, it is concluded that the best combinations of model/data source to estimate DNI are either libRadtran/MERRA-2 (according to MBE and FB) and SMARTS/AERONET (according to RMSE, FGE, and R2). In the case of GHI, the best combinations are REST2/AERONET (according to FB, FGE, and R2) and REST2/MERRA-2 (according to MBE and RMSE).

Global Corrections to Reference Irradiance Spectra for Non-Clear-Sky Conditions.

Photochemical reactions in surface waters play important roles in element cycling and in the removal of organic contaminants, among other processes. A central environmental variable affecting photochemical processes in surface waters is the incoming solar irradiance, as this initiates these processes. However, clear-sky incident irradiance spectra are often used when evaluating the fate of aquatic contaminants, leading to an overestimation of contaminant decay rates due to photochemical transformation. In this work, incident irradiance satellite data were used to develop global-scale non-clear-sky correction factors for commonly used reference irradiance spectra. Non-clear-sky conditions can decrease incident irradiance by over 90% depending on the geographic location and time of the year, with latitudes above 40°N being most heavily affected by seasons. The impact of non-clear-sky conditions on contaminant half-lives was illustrated in a case study of triclosan in lake Greifensee, which showed a 39% increase in the triclosan half-life over the course of a year under non-clear-sky conditions. A global annual average correction factor of 0.76 was determined as an approximate way to account for non-clear-sky conditions. The correction factors are developed at monthly and seasonal resolutions for every location on the globe between 70°N and 60°S at a 4 km spatial resolution and can be used by researchers, practitioners, and regulators who need improved estimates of incident irradiance.

Probabilistic solar forecasting: Benchmarks, post-processing, verification

Probabilistic solar forecasts may take the form of predictive probability distributions, ensembles, quantiles, or interval forecasts. State-of-the-art approaches build on input from numerical weather prediction (NWP) models and post-processing with statistical and machine learning methods. We propose a probabilistic benchmark based on a deterministic forecast of clear-sky irradiance, introduce new methods for post-processing that merge statistical techniques with modern neural networks, discuss methods for spatio-temporal scenario forecasts, and illustrate the assessment of predictive ability via proper scoring rules and calibration checks. We expect future solar forecasting efforts to be increasingly probabilistic, and encourage continuing close interaction with operational weather prediction, where innovations based on sophisticated neural networks supplement and challenge traditional approaches.

Impacts of the Aerosol Representation in WRF-Solar Clear-Sky Irradiance Forecasts over CONUS

Abstract Aerosol optical depth (AOD) is a primary source of solar irradiance forecast error in clear-sky conditions. Improving the accuracy of AOD in NWP models like WRF will thus reduce error in both direct normal irradiance (DNI) and global horizontal irradiance (GHI), which should improve solar power forecast errors, at least in cloud-free conditions. In this study clear-sky GHI and DNI was analyzed from four configurations of the WRF-Solar model with different aerosol representations: 1) the default Tegen climatology, 2) imposing AOD forecasts from the GEOS-5 model, 3) imposing AOD forecasts from the Copernicus Atmosphere Monitoring Service (CAMS) model, and 4) the Thompson–Eidhammer aerosol-aware water/ice-friendly aerosol climatology. More than 8 months of these 15-min output forecasts are compared with high-quality irradiance observations at NOAA SURFRAD and Solar Radiation (SOLRAD) stations located across CONUS. In general, WRF-Solar with GEOS-5 AOD had the lowest errors in clear-sky DNI, while WRF-Solar with CAMS AOD had the highest errors, higher even than the two aerosol climatologies, which is consistent with validation of the four AOD550 datasets against AERONET stations. For clear-sky GHI, the statistics differed little between the four models, as expected because of the lesser sensitivity of GHI to aerosol loading. Hourly average clear-sky DNI and GHI were also analyzed, and they were additionally compared with CAMS model output directly. CAMS irradiance performed competitively with the best WRF-Solar configuration (with GEOS-5 AOD). The markedly different performance of CAMS versus WRF-Solar with CAMS AOD indicates that CAMS is apparently less sensitive to AOD550 than WRF-Solar is. Significance Statement Particles in the atmosphere called aerosols, which can include dust, smoke, sea salt, sulfates, black carbon, and organic carbon, absorb and scatter incoming sunlight. Improving the representation of aerosols in numerical weather prediction models reduces forecast errors in solar irradiance at ground level, particularly direct normal irradiance, during cloud-free conditions. This in turn should result in improved accuracy of solar power forecasts, especially for concentrated solar power (CSP) plants. CSP plants tend to be built in more arid, less cloudy regions that are also prone to dust loading, so accurate aerosol forecasts are particularly relevant. Comparing four representations of aerosols in the WRF-Solar model over eight months of forecasts across the United States reveals substantial differences in clear-sky irradiance forecast skill.

Interpretable temporal-spatial graph attention network for multi-site PV power forecasting

Accurate forecasting of photovoltaic (PV) and wind production is crucial for the integration of more renewable energy sources into the power grid. To address the limited resolution and costs of methods based on numerical weather predictions (NWP), we take PV production data as main input for forecasting. Since PV power is affected by weather and cloud dynamics, we model spatio-temporal correlations between production data by representing PV systems as nodes of a dynamic graph and embedding production data, geographical information and clear-sky irradiance as signals on that graph. We introduce a new temporal-spatial multi-windows graph attention network (TSM-GAT) for predicting future PV power production. TSM-GAT can adapt to the dynamics of the problem, by learning different graphs over time. It consists of temporal attention with an overlapping-window mechanism that finds the temporal correlations and spatial attention with a multi-window mechanism, which captures different dynamical spatio-temporal correlations for different parts of the forecasting horizon. Thus, it is possible to interpret which PV stations have the most influence when making a prediction for short-, medium- and long-term intra-day forecasts. TSM-GAT outperforms multi-site state-of-the-art models for four to six hours ahead predictions, with average NRMSE 12.4% and 10.5% on a real and synthetic dataset, respectively. Furthermore, it outperforms state-of-the-art models that use NWP as inputs for up to five hours ahead predictions. TSM-GAT yields predicted signals with a closer shape to ground truth than state-of-the-art models, which indicates that it is better at capturing cloud motion and may lead to better generalization capabilities.

Data-Driven Direct Diagnosis of PV Connected Batteries

Photovoltaic systems are providing a growing share of power to the electric grid worldwide. As more and more of these systems are tied with battery energy storage to balance their intermittency, being able to ensure long and safe operation of these batteries will become critical. Because of their sporadic usage, far from the typical constant current or constant power, diagnosability without having to perform lengthy maintenance cycles will be a key factor to accelerate deployment.This work proposes a new methodology to enable the training of machine learning algorithms so they could be applied directly to photovoltaic (PV) battery charging data for opportunistic diagnosis. This is done by training the algorithms on synthetic voltage data under different degradations calculated from clear-sky model irradiance data. Validation was performed on synthetic voltage responses calculated from plane of array irradiance observations for a PV system located in Maui, HI, USA.Herein, we will present the result of our proof-of-concept study where we investigated the diagnosability of the voltage response of PV charged batteries under 18 different cloud coverages scenarios and for more than 10,000 different degradations. We also evaluated the effect of training conditions on clear-sky irradiance data by analyzing the resulting differences from training on 700,000 different voltage curves for a single day vs. 45,000 different voltage curves for of each of the 18 days used for cloud coverage estimation. In addition, we considered and quantified the effect of cell-to-cell variations. Training was performed on five state-of-the-art machine learning algorithms for battery diagnosis which included decision tree ensemble methods and neural networks. Finally, because the diagnosis was done outside of constant current, the capacity- and time-based information were decorrelated and compared. While time-base diagnosis is the most interesting for deployed systems, it was shown to be less accurate than its capacity counterpart for the tested algorithms. Yet, the accuracy of the diagnosis is still satisfactory for days where cloud effect is limited (average RMSE of 2.5% compared to 1.3% for capacity based if 50% or more clear skies).Summarizing, we will show that the degradation of PV-connected batteries could be diagnosed opportunistically without the need for maintenance cycles using well-trained state-of-art machine learning algorithms. Calculated RMSEs were below 2% for days with 50% or more clear skies considering more than 10,000 different degradation paths with 25% or less of the three thermodynamic degradation modes. In addition to the results themselves, and before real PV battery degradation data becomes available, this work is showing the significant benefits of using synthetic data to understand the expected variations and to start preparing adapted tools for when data become available.

Relationship between ozone and biologically relevant UV at 4 NDACC sites.

Clouds and aerosols, as well as overhead ozone, can have large effects on ultraviolet (UV) irradiances. We use statistical methods to remove cloud effects and mean aerosol effects from spectral UV irradiance measurements to investigate the relationship between UV and total column ozone. We show that for fixed solar zenith angles (SZA), seasonal changes in ozone lead to marked changes in clear-sky UV irradiances. Such effects are larger at mid-latitudes than in the tropics. At mid-latitudes, the minimum ozone amount over the course of a year can be about 50 percent of its maximum, with the lowest values in autumn and the highest values in spring. These seasonal ozone changes lead to UV Index (UVI) values in autumn that can exceed those in spring at the same SZA by nearly a factor of two. Differences are even larger for UV spectra weighted by the action spectra for DNA-damaging UV, and for cutaneous previtamin D production. In some cases, the seasonal increase exceeds a factor of 4. The analysis experimentally demonstrates the limits of applicability of the concept of constant Radiative Amplification Factors (RAFs) for estimating effects of changes in ozone for some weighting functions. Changes in DNA-weighted UV and erythemally weighted UV are well represented by the published RAFs. However, there are large SZA dependencies in the case of UVB and vitamin D-weighted UV. For all weightings considered, RAFs calculated from the observations as a function of SZA show similar dependencies between sites, in good agreement with published values, independently of the ozone data source. High quality measurements show that natural variations in ozone are responsible for huge variations in biologically damaging UV, with seasonal changes at fixed solar zenith angles sometimes exceeding a factor of four. The measured changes from thousands of spectra agree well with calculations over a wide range of solar zenith angles.

Evaluation of aerosol optical depths and clear-sky radiative fluxes of the CERES Edition 4.1 SYN1deg data product

Abstract. Aerosol optical depths (AODs) used for the Edition 4.1 Clouds and the Earth's Radiant Energy System (CERES) Synoptic 1∘ (SYN1deg) product are evaluated. AODs are derived from Moderate Resolution Imaging Spectroradiometer (MODIS) observations and assimilated by an aerosol transport model (the Model for Atmospheric Transport and Chemistry – MATCH). As a consequence, clear-sky AODs closely match with those derived from MODIS instruments. AODs under all-sky conditions are larger than AODs under clear-sky conditions, which is supported by ground-based AErosol RObotic NETwork (AERONET) observations. When all-sky MATCH AODs are compared with Modern-Era Retrospective analysis for Research and Applications (Version 2; MERRA-2) AODs, MATCH AODs are generally larger than MERRA-2 AODs, especially over convective regions (e.g., the Amazon, central Africa, and eastern Asia). This variation is largely due to the differing methods of assimilating the MODIS AOD data product and the use of quality flags in our assimilation. Including AODs with larger retrieval uncertainty makes AODs over the convective regions larger. When AODs are used for clear-sky irradiance computations and computed downward shortwave irradiances are compared with ground-based observations, the computed instantaneous irradiances are 1 %–2 % larger than observed irradiances. The comparison of top-of-atmosphere clear-sky irradiances with those derived from CERES observations suggests that AODs used for surface radiation observation sites are 0.01–0.03 larger, which is within the uncertainty of instantaneous MODIS AODs. However, the comparison with AERONET AODs suggests that AODs used for computations over desert sites are 0.08 larger. The cause of positive biases in downward shortwave irradiance and in AOD for the desert sites is possibly due to the dust particle size and distribution, as defined by the MATCH transport model, and the transfer of that information into the radiative transfer model.

Effects of spatial scale of atmospheric reanalysis data on clear-sky surface radiation modeling in tropical climates: A case study for Singapore

Solar resource assessments most generally require atmospheric information, which is customarily acquired from gridded datasets. The spatial scale mismatch problem, i.e., the difference in spatial representativeness of gridded data and in situ measurements, therefore becomes relevant. This study examines how the gridded data used as inputs to clear-sky radiation models can affect their performance at urban scale. The tropical island of Singapore is selected for the case study. Aerosol optical depth at 550 nm (AOD550), Å ngström exponent (AE), and precipitable water (PW) from both the MERRA-2 reanalysis and ground-based stations (AERONET and SuomiNet) are collected between 2013–2020. Firstly, it is found that, relatively to the AERONET ground truth, the bias in MERRA-2’s AOD550 is more prominent than that in AE or PW. Next, the bias propagation from the gridded inputs (AOD550, AE, and PW) to clear-sky radiation predictions is explored using various models. The estimated clear-sky direct normal irradiance (DNIcs) is more sensitive to AOD550 variation than the clear-sky global horizontal irradiance (GHIcs). Six clear-sky radiation models, five of which accept MERRA-2 gridded inputs, are compared with each other, and with the in situ irradiance measurements recorded at 9 sites. The inter-model difference across Singapore is remarkably consistent because the whole island fits inside a single MERRA-2 grid cell. Under high-AOD550 situations, however, the inter-model deviation becomes large for both GHIcs and DNIcs. The conventional model-versus-measurement comparison shows that each model achieves very different site-to-site performance, largely because the spatially-averaged inputs cannot fully represent the micro-climatic variability. Relatively speaking, no clear-sky radiation model significantly outperforms its peers. The simple MAC2 model and the empirical (locally derived) Yang GHIcs-only model are recommended for Singapore.

Improved Clear Sky Model from In Situ Observations and Spatial Distribution of Aerosol Optical Depth for Satellite-Derived Solar Irradiance over the Korean Peninsula

In solar resource assessment, the climatological environment of the target area is objectively quantified by the cloudiness or clear sky index, which is defined as the ratio of global horizontal irradiance to clear sky solar insolation. The clear sky model calculates incoming solar irradiance on the ground surface considering several atmospheric parameters such as water vapor and aerosol optical depth. This study investigated the importance of aerosol optical depth for deriving clear sky irradiance in radiative transfer models and examined its viability in a universal or community model for public use. The evaluation was conducted based on ground observations at the Korea Institute of Energy Research (KIER) station from January to December 2021. The original simulation was performed using the monthly mean of aerosol optical depth obtained from the Aerosol Robotic Network station; the mean absolute error was 29.9 W m−2. When the daily mean of in situ observations at KIER was incorporated into the clear sky model, the mean absolute error was reduced to 9.7 W m−2. Our results confirm that the clear sky model using gridded datasets of aerosol optical depth is suitable for use as a universal or community model.