Solar Irradiance Observations Research Articles

Wavelength calibration of a solar spectral irradiance radiometer and validation by correlation photon coincidence counting

Accurate wavelength calibration is the key to achieving accuracy observation of solar spectral irradiance. A wavelength calibration of solar spectral irradiance radiometer based on correlation photon radiation reference is proposed. The wavelengths of two channels of the solar spectral irradiance radiometer are calibrated by a Hg-Ar lamp and lasers. A collimating and focusing optics are designed to ensure that light can illuminate the slit vertically and without polarization. The optics can also keep the consistency of observing and self-calibration modes. The calibration equations of two channels are established. The wavelength deviations of two channels are −0.11 ∼ 0.07 nm and −0.085 ∼ 0.017 nm respectively. To validate the calibrated wavelength, channel 1 is selected as the inspection channel according to the spectral line number of the calibration lamp and the variation of the photon count rate. The motor position close to the calibration wavelength 763.511 nm is selected in channel 2, and the coincidence measurement is carried out to check the accuracy of wavelength calibration in channel 1. The deviation of wavelength around 408 nm is 1.139 nm. The experimental results show that the correlated photons self-calibration system can realize the calibration and validation of the system wavelength. The method can increase the number of calibration wavelengths and is suitable for both channels.

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.

A Hybrid Ensemble Learning Model for Short-Term Solar Irradiance Forecasting Using Historical Observations and Sky Images

A hybrid multi-modal ensemble learning model is proposed for short-term solar irradiance forecasting based on historical observations and sky images in this paper. In the proposed model, historical observations of solar irradiance are utilized to extract temporal characteristics, and ground-based sky images are used as an exogenous input to offer cloud cover information. A powerful ensemble learning model, Extreme Gradient Boosting (XGBoost), is employed to capture the function relationships between input features and future observations. Since mean squared error loss is sensitive to the extreme large or small historical irradiance, a novel loss function is proposed to improve robustness of XGBoost. In order to find out the best controlling parameters, Rao-1 algorithm is employed for its easy operation. To validate performance of the proposed method, a solar irradiance dataset containing three-year historical observations and ground-based sky images collected from Folsom is employed. Meanwhile, five commonly applied methods, LASSO, Ridge regression, support vector regression, boosted regression trees, the generic XGBoost, are considered as benchmarking methods. The forecasting horizons from 5 to 30 min are considered for all compared methods while two metrics, mean absolute error and root mean squared error, are computed. Experimental results prove that the proposed hybrid model has better forecasting performance compared with benchmarking methods over all forecasting horizons.

Reconciling Observations of Solar Irradiance Variability With Cloud Size Distributions

AbstractClouds cast shadows on the surface and locally enhance solar irradiance by absorbing and scattering sunlight, resulting in fast and large solar irradiance fluctuations on the surface. Typical spatiotemporal scales and driving mechanisms of this intra‐day irradiance variability are not well known, hence even 1 day ahead forecasts of variability are inaccurate. Here, we use long‐term, high‐frequency solar irradiance observations combined with satellite imagery, numerical simulations, and conceptual modeling to show how irradiance variability is linked to the cloud size distribution. Cloud shadow sizes are distributed according to a power law over multiple orders of magnitude, deviating only from the cloud size distribution due to cloud edge transparency at scales below ≈750 m. Locally cloud‐enhanced irradiance occurs as frequently as shadows, and is similarly driven mostly by boundary layer clouds, but distributed over a smaller range of scales. We reconcile studies of solar irradiance variability with those on clouds, which brings fundamental understanding to what drives irradiance variability. Our findings have implications not only for weather and climate modeling, but also for solar energy and photosynthesis by vegetation, where detailed knowledge of surface solar irradiance is essential.

Detecting undocumented trends in solar irradiance observations

Quantifying the long-term stability of solar irradiance observations is crucial for determining how the Sun varies in time and detecting decadal climate change signals. The stability of irradiance observations is challenged by the degradation of instrumental sensitivity in space and by the post-launch corrections needed to mitigate this degradation. We propose a new framework for detecting instrumental trends based on the existing idea of comparing the solar irradiance at pairs of dates for which a proxy quantity reaches the same level. Using a parametric model, we then reconstruct the trend and its confidence interval at all times. While this method cannot formally prove the instrumental origin of the trends, the observation of similar trends with different proxies provides strong evidence for a non-solar origin. We illustrate the method with spectral irradiance observations from the Solar Radiation and Climate Experiment (SORCE) mission, using various solar proxies such as sunspot number, MgII index, F10.7 index. The results support the existence of non-solar trends that exceed the level of solar cycle variability. After correcting the spectral irradiance for these trends, we find the difference between the levels observed at solar maximum and at solar minimum to be in good agreement with irradiance models.

A Reconstruction of Total Solar Irradiance Based on Wavelet Analysis

AbstractAs an important indicator of solar activity, the long‐term total solar irradiance (TSI) observations are needed to uncover the impact of solar activity on Earth's climate. In this paper, the periodic variation of TSI and its relationships with sunspot number and Ca II K index are analyzed by using the satellite observation data from 1979 to 2015 and the wavelet method. The results of continuous wavelet analysis show that TSI, sunspot number and Ca II K index all have significant and stable oscillation periods of 9∼13 years and intermittent oscillation periods of 2∼6 months only during the time of intense solar activity. Moreover, the results of cross wavelet analysis indicate that the effect of sunspot number and Ca II K index on TSI is mainly reflected in the period of 9∼13 years with one month phase lag, and sunspot number and Ca II K index cannot explain the TSI variation in the period of 3∼6 months. Under the condition of considering the phase relationship and the model order, the TSI reconstruction model is established and the monthly TSI time series from 1907 to 1978 is restored.

The TSIS-1 Hybrid Solar Reference Spectrum.

We present a new solar irradiance reference spectrum representative of solar minimum conditions between solar cycles 24 and 25. The Total and Spectral Solar Irradiance Sensor‐1 (TSIS‐1) Hybrid Solar Reference Spectrum (HSRS) is developed by applying a modified spectral ratio method to normalize very high spectral resolution solar line data to the absolute irradiance scale of the TSIS‐1 Spectral Irradiance Monitor (SIM) and the CubeSat Compact SIM (CSIM). The high spectral resolution solar line data are the Air Force Geophysical Laboratory ultraviolet solar irradiance balloon observations, the ground‐based Quality Assurance of Spectral Ultraviolet Measurements In Europe Fourier transform spectrometer solar irradiance observations, the Kitt Peak National Observatory solar transmittance atlas, and the semi‐empirical Solar Pseudo‐Transmittance Spectrum atlas. The TSIS‐1 HSRS spans 202–2730 nm at 0.01 to ∼0.001 nm spectral resolution with uncertainties of 0.3% between 460 and 2365 nm and 1.3% at wavelengths outside that range.

SOLAR-v: A new solar spectral irradiance dataset based on SOLAR/SOLSPEC observations during solar cycle 24

Context. Solar spectral irradiance (SSI) is the wavelength-dependent energy input to the top of the Earth’s atmosphere. Solar ultraviolet (UV) irradiance represents the primary forcing mechanism for the photochemistry, heating, and dynamics of the Earth’s atmosphere. Hence, both temporal and spectral variations in solar UV irradiance represent crucial inputs to the modeling and understanding of the behavior of the Earth’s atmosphere. Therefore, measuring the long-term solar UV irradiance variations over the 11-year solar activity cycle (and over longer timescales) is fundamental. Thus, each new solar spectral irradiance dataset based on long-term observations represents a major interest and can be used for further investigations of the long-term trend of solar activity and the construction of a homogeneous solar spectral irradiance record. Aims. The main objective of this article is to present a new solar spectral irradiance database (SOLAR-v) with the associated uncertainties. This dataset is based on solar UV irradiance observations (165−300 nm) of the SOLAR/SOLSPEC space-based instrument, which provides measurements of the full-disk SSI during solar cycle 24. Methods. SOLAR/SOLSPEC made solar acquisitions between April 5, 2008 and February 10, 2017. During this period, the instrument was affected by the harsh space environment that introduces instrumental trends (degradation) in the SSI measurements. A new method based on an adaptation of the Multiple Same-Irradiance-Level (MuSIL) technique was used to separate solar variability and any uncorrected instrumental trends in the SOLAR/SOLSPEC UV irradiance measurements. Results. A new method for correcting degradation has been applied to the SOLAR/SOLSPEC UV irradiance records to provide new solar cycle variability results during solar cycle 24. Irradiances are reported at a mean solar distance of 1 astronomical unit (AU). In the 165−242 nm spectral region, the SOLAR/SOLSPEC data agrees with the observations (SORCE/SOLSTICE) and models (SATIRE-S, NRLSSI 2) to within the 1-sigma error envelope. Between 242 and 300 nm, SOLAR/SOLSPEC agrees only with the models.

Degradation Correction of TSIS SIM

An understanding of solar variability over a broad spectral range and broad range of timescales is needed by scientists studying Earth’s climate. The Total and Spectral Solar Irradiance Sensor (TSIS) Spectral Irradiance Monitor (SIM), is designed to measure solar spectral irradiance (SSI) with unprecedented accuracy from 200 nm to 2400 nm. SIM started daily observations in March 2018. To maintain its accuracy over the course of its anticipated 5-year mission and beyond, TSIS SIM needs to be corrected for optical degradation, common for solar viewing instruments. The differing long-term trends of various independent solar-irradiance records attest to the challenge at hand. The correction of TSIS SIM for optical degradation is based on piecewise linear fits that bring the three instrument channels into agreement. It is fundamentally different to the correction applied to the TSIS SIM predecessor on SORCE. The correction facilitates reproducibility, uncertainty estimation and is measurement-based. Corrected, integrated TSIS SIM SSI agrees with independent observations of total solar irradiance to within 45 ppm as well as various solar-irradiance models. TSIS SIM SSI is available at: http://lasp.colorado.edu/lisird/ .

The LASCO Coronal Brightness Index

We present the construction of a new white-light coronal brightness index (CBI) from the entire archive of observations recorded by the Large Angle Spectrometric Coronagraph (LASCO) C2 camera between 1996 and 2017, comprising two full solar cycles. We reduce all fully calibrated daily C2 observations of the white light corona into a single daily coronal brightness observation for every day of observation recorded by the instrument, with mean daily brightness values binned into 0.1 Rsun radial x 1 degree angular regions from 2.4 -- 6.2 Rsun for a full 360-degrees. As a demonstration of the utility of the CBI, we construct a new solar irradiance proxy that correlates well with a variety of direct solar irradiance observations, with correlations shown to be in the range of 0.77-0.89. We also present a correlation mapping technique to show how irradiance correlations depend on, and relate to, coronal structure/locations, and to demonstrate how the LASCO CBI can be used to perform long-term "spatial correlation" studies to investigate relationships between the solar corona and any arbitrary concurrent geophysical index. Using this technique we find possible relationships between coronal brightness and plasma temperature, interplanetary magnetic field magnitude and (very weakly) proton density.

Neural Network for Solar Irradiance Modeling (NN-SIM)

An understanding of solar variability over a broad range of wavelengths and timescales is needed by scientists studying Earth’s climate. While the current understanding of solar irradiance from measurements and models is maturing, there remain notable areas of discrepancy that highlight a lack of understanding of the variability in solar spectral irradiance (SSI) on 27-day solar-rotational timescales and longer, and in total solar irradiance (TSI) at solar-cycle timescales and longer. The sources of instrumental noise and instability suspected behind differences in independent measurement records are actively debated. Furthermore, estimates from solar-irradiance empirical-proxy models and semi-empirical models also differ from each other and from the observations by varying degrees. To investigate whether models and observations can be brought into closer agreement we developed a novel, data-driven, solar-irradiance model using an ensemble of feed-forward artificial neural networks. Key features of our model architecture include a non-linear relationship between solar-activity proxy and irradiance with a high degree of freedom that comes from the incorporation of a greater number of solar-activity proxies than previous proxy models. Furthermore, we utilize a recent re-analysis of solar spectral irradiance (SSI) observations, stemming from a new degradation-correction methodology, to develop our model. Our approach, the Neural Network for Solar Irradiance Modeling (NN-SIM), reconstructs total solar irradiance and SSI from 205 nm to 2300 nm and from 1979 to the present day. We find close agreement between NN-SIM and various observational records as well as independent models. NN-SIM is available at .

Modeling Quiet Solar Luminosity Variability from TSI Satellite Measurements and Proxy Models during 1980–2018

A continuous record of direct total solar irradiance (TSI) observations began with a series of satellite experiments in 1978. This record requires comparisons of overlapping satellite observations with adequate relative precisions to provide useful long term TSI trend information. Herein we briefly review the active cavity radiometer irradiance monitor physikalisch-meteorologisches observatorium davos (ACRIM-PMOD) TSI composite controversy regarding how the total solar irradiance (TSI) has evolved since 1978 and about whether TSI significantly increased or slightly decreased from 1980 to 2000. The main question is whether TSI increased or decreased during the so-called ACRIM-gap period from 1989 to 1992. There is significant discrepancy between TSI proxy models and observations before and after the gap, which requires a careful revisit of the data analysis and modeling performed during the ACRIM-gap period. In this study, we use three recently proposed TSI proxy models that do not present any TSI increase during the ACRIM-gap, and show that they agree with the TSI data only from 1996 to 2016. However, these same models significantly diverge from the observations from 1981 and 1996. Thus, the scaling errors must be different between the two periods, which suggests errors in these models. By adjusting the TSI proxy models to agree with the data patterns before and after the ACRIM-gap, we found that these models miss a slowly varying TSI component. The adjusted models suggest that the quiet solar luminosity increased from the 1986 to the 1996 TSI minimum by about 0.45 W/m2 reaching a peak near 2000 and decreased by about 0.15 W/m2 from the 1996 to the 2008 TSI cycle minimum. This pattern is found to be compatible with the ACRIM TSI composite and confirms the ACRIM TSI increasing trend from 1980 to 2000, followed by a long-term decreasing trend since.

Comparison of Decadal Trends among Total Solar Irradiance Composites of Satellite Observations

We present a new analysis of the two-decade-old controversy over interpretation of satellite observations of total solar irradiance (TSI) since 1978 and the implications of our findings for TSI as a driver of climate change. Our approach compares the methods of constructing the two most commonly referenced TSI composites (ACRIM and PMOD) that relate successive observational databases and two others recently constructed using a novel statistical approach. Our primary focus is on the disparate decadal trending results of the ACRIM and PMOD TSI composite time series, namely, whether they indicate an increasing trend from 1980 to 2000 and a decreasing trend thereafter (ACRIM) or a continuously decreasing trend since 1980 (PMOD). Construction of the four-decade observational TSI composites from 1978 to the present requires the use of results from two less precise Earth Radiation Budget experiments (Nimbus7/ERB and ERBS/ERBE) during the so-called ACRIM-Gap (1989.5–1991.8), between the end of the ACRIM1 and the beginning of the ACRIM2 experiments. The ACRIM and PMOD composites used the ERB and ERBE results, respectively, to bridge the gap. The well-established paradigm of positive correlation between Solar Magnetic Field Strength (SMFS) and TSI supports the validity of the upward trend in the ERB results and the corresponding decadal upward trend of the ACRIM composite during solar cycles 21 and 22. The ERBE results have a sensor degradation caused downward gap trend, contrary to the SMFS/TSI paradigm, that biased the PMOD composite decadal trend downward during solar cycles 21 and 22. The different choice of gap bridging data is clearly the cause of the ACRIM and PMOD TSI trending difference, agreeing closely in both magnitude and direction. We also analyze two recently proposed statistical TSI composites. Unfortunately their methodology cannot account for the gap degradation of the ERBE experiment and their resulting uncertainties are too large to uniquely distinguish between the trending of the ACRIM and PMOD composites. Our analysis supports the ACRIM TSI increasing trend during the 1980 to 2000 period, followed by a long-term decreasing trend since.

Solar Ultraviolet Irradiance Observations of the Solar Flares During the Intense September 2017 Storm Period

AbstractA large outburst of flares occurred between 4–10 September 2017 when new magnetic flux emerged into and strengthened an existing active region, National Oceanic and Atmospheric Administration Region 12673. This intense solar storm period included X9.3 (6 September) and X8.2 (10 September) flares, the largest flares that have occurred during Solar Cycle 24, as well as 39 M‐class flares and three additional X‐class flares. Another X‐class flare from this active region was observed on the farside of the Sun from Mars Atmosphere and Volatile EvolutioN prior to the September events, along with other large M‐class flares, showing the potential for how farside irradiance monitoring can improve flare prediction at Earth for 1‐ to 13‐day forecasts. This September 2017 flare period is similar to other famous storm periods such as the 18 October to 5 November 2003 Halloween storm that produced 14 X‐class flares and 137 M‐class flares and the 6–10 September 2005 period that had 11 X‐class and 68 M‐class flares. All of these storm periods occurred in the declining phase of the solar cycle when solar activity had decreased significantly from solar maximum levels. This paper focuses on a number of solar irradiance observations at ultraviolet (0–190 nm) wavelengths during the September 2017 storm period and the advantages that an ensemble of measurements and models have for studying solar flares.

Did Smoke From City Fires in World War II Cause Global Cooling?

AbstractBetween 3 February and 9 August 1945, an area of 461 km2 in 69 Japanese cities, including Hiroshima and Nagasaki, was burned during the U.S. B‐29 Superfortress air raids. In the previous 5 years, 205 km2 in German cities were destroyed, so the smoke that was generated was spread out over a much longer period of time than that from Japan in 1945. Observations of solar irradiance show reductions consistent with the hypothesis that smoke was injected into the stratosphere by the city fires. Historical simulations from the Coupled Model Intercomparison Project 5, with no smoke in their forcing, showed no postwar cooling. Global average surface air temperature observations during and following World War II are problematic, because of issues with measuring sea surface temperatures, but there were no large volcanic eruptions, El Niño, or La Niña during this period to confuse the record. Nevertheless, 1945 and 1946 global average land surface air temperatures were not significantly lower than the average for 1940–1944. Estimates of the amount of smoke generated by the fires are somewhat uncertain. Although the climate record is consistent with an expected 0.1–0.2 K cooling, because of multiple uncertainties in smoke injected to the stratosphere, solar radiation observations, and surface temperature observations, it is not possible to formally detect a cooling signal from World War II smoke.

Revised composite extraterrestrial spectrum based on recent solar irradiance observations

A revision of a previous composite spectrum (Gueymard, 2004) is undertaken here, with a focus on the 200–4000 nm waveband, where most of the sun’s energy is concentrated. The methodology is based on 31 sources of spectral data covering various parts of the spectrum. The data sources include observations from surface observatories, aircraft, high-altitude balloons, satellites, as well as modeled estimates of the solar flux. All existing spectra are downscaled to a single, relatively low spectral resolution to allow their direct comparison. Three successive filtering steps make it possible to eliminate outliers at each of the 2165 wavelengths under scrutiny here. The mean irradiance at each wavelength is then obtained by averaging the values of the surviving data points. The 2004 spectrum is used below 200 nm and above 4000 nm to create a complete spectrum from 0.5 nm to near-infinity. In a final step, the irradiance values are properly scaled so as to integrate to the revised solar constant value of 1361.1 W m−2. Compared to the earlier 2004 spectrum, this new composite differs mostly in the UV and visible parts of the spectrum. Based on the present findings, a revision of the outdated ASTM E490 standard extraterrestrial spectrum is recommended.

A reevaluation of the solar constant based on a 42-year total solar irradiance time series and a reconciliation of spaceborne observations

A reevaluation of the solar constant is undertaken here to take into account the progress in space radiometry that has occurred since the early 2000s. Various sources of spaceborne total solar irradiance (TSI) observations are investigated here, including the long-term ACRIM and PMOD composites, as well as recent observations from the SORCE-TIM, TCTE-TIM, and PICARD-PREMOS instruments. A proxy model is constructed using daily data of sunspot number, radio flux at 10.7 cm, and MgII index, as predictors for TSI over the 42-year period 1976–2017. These daily estimates are used to fill in 9.7% of missing TSI observations during that period. By comparison with these proxy estimates, the PMOD composite appears generally more reliable than the ACRIM composite before 2003, and particularly before 1981. The 42-year time span is separated into nine periods, each defining the revised TSI daily values from one or more sources that are selected based on the trend of their resemblance with the proxy model. A final correction is added to emulate the highly accurate absolute calibration of PREMOS. Based on the resulting TSI reconstruction, a revised solar constant value of 1361.1 W/m2 is obtained, with a standard uncertainty of 0.5 W/m2. The revised solar constant is ≈5 W/m2 less than the previous values promulgated in ASTM and ISO standards. A revision of these standards is thus highly recommended.

CLEAR-SKY RADIATIVE AND TEMPERATURE EFFECTS OF DIFFERENT AEROSOL CLIMATOLOGIES IN THE COSMO MODEL

We estimated the effects of the different aerosol climatologies in the COSMO mesoscale atmospheric model using long-term aerosol measurements and the accurate global solar irradiance observations at ground at the Moscow State University Meteorological Observatory (Russia) and Lindenberg Observatory (Germany) in clear sky conditions. The differences in aerosol properties have been detected especially during winter months. There is a better agreement of MACv2 aerosol climatology with measurements forMoscowconditions compared with Tegen aerosol climatology. However, we still have a systematical negative bias of about 2-3% in global solar irradiance at ground for both sites. A noticeable sensitivity of air temperature at2 metersto the net radiation changes of about1°Cper 100 Wm-2 due to aerosol has been evaluated, which approximately is around -0.2 – -0.3°C, when accounting for real aerosol properties.

An Evaluation of Satellite Estimates of Solar Surface Irradiance Using Ground Observations in San Antonio, Texas, USA

Estimates of solar irradiance at the earth’s surface from satellite observations are useful for planning both the deployment of distributed photovoltaic systems and their integration into electricity grids. In order to use surface solar irradiance from satellites for these purposes, validation of its accuracy against ground observations is needed. In this study, satellite estimates of surface solar irradiance from Geostationary Operational Environmental Satellite (GOES) are compared with ground observations at two sites, namely the main campus of the University of Texas at San Antonio (UTSA) and the Alamo Solar Farm of San Antonio (ASF). The comparisons are done mostly on an hourly timescale, under different cloud conditions classified by cloud types and cloud layers, and at different solar zenith angle intervals. It is found that satellite estimates and ground observations of surface solar irradiance are significantly correlated (p < 0.05) under all sky conditions (r: 0.80 and 0.87 on an hourly timescale and 0.94 and 0.91 on a daily timescale, respectively for the UTSA and ASF sites); on the hourly timescale, the correlations are 0.77 and 0.86 under clear-sky conditions, and 0.74 and 0.84 under cloudy conditions, respectively for the UTSA and ASF sites, and mostly >0.60 under different cloud types and layers for both sites. The correlations under cloudy-sky conditions are mostly stronger than those under clear-sky conditions at different solar zenith angles. The correlation coefficients are mostly the smallest with solar zenith angle in the range of 75–90° under all sky, clear-sky and cloudy-sky conditions. At the ASF site, the overall bias of GOES surface solar irradiance is small (+1.77 Wm−2) under all sky while relatively larger under clear-sky (−22.29 Wm−2) and cloudy-sky (+40.31 Wm−2) conditions. The overall good agreement of the satellite estimates with the ground observations underscores the usefulness of the GOES surface solar irradiance estimates for solar energy studies in the San Antonio area.

Probabilistic Solar Forecasting Using Quantile Regression Models

In this work, we assess the performance of three probabilistic models for intra-day solar forecasting. More precisely, a linear quantile regression method is used to build three models for generating 1 h–6 h-ahead probabilistic forecasts. Our approach is applied to forecasting solar irradiance at a site experiencing highly variable sky conditions using the historical ground observations of solar irradiance as endogenous inputs and day-ahead forecasts as exogenous inputs. Day-ahead irradiance forecasts are obtained from the Integrated Forecast System (IFS), a Numerical Weather Prediction (NWP) model maintained by the European Center for Medium-Range Weather Forecast (ECMWF). Several metrics, mainly originated from the weather forecasting community, are used to evaluate the performance of the probabilistic forecasts. The results demonstrated that the NWP exogenous inputs improve the quality of the intra-day probabilistic forecasts. The analysis considered two locations with very dissimilar solar variability. Comparison between the two locations highlighted that the statistical performance of the probabilistic models depends on the local sky conditions.