Validation of GOES-Based Surface Insolation Retrievals and Its Utility for Model Evaluation

Peiyang Cheng Department of Atmospheric Science, University of Alabama in Huntsville, Huntsville, Alabama

Search for other papers by Peiyang Cheng in
Current site
Google Scholar
PubMed
Close
,
Arastoo Pour-Biazar Earth System Science Center, University of Alabama in Huntsville, Huntsville, Alabama

Search for other papers by Arastoo Pour-Biazar in
Current site
Google Scholar
PubMed
Close
,
Richard T. McNider Department of Atmospheric Science, University of Alabama in Huntsville, Huntsville, Alabama

Search for other papers by Richard T. McNider in
Current site
Google Scholar
PubMed
Close
, and
John R. Mecikalski Department of Atmospheric Science, University of Alabama in Huntsville, Huntsville, Alabama

Search for other papers by John R. Mecikalski in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

Incident solar radiation at Earth’s surface, also called surface insolation, plays an important role in the Earth system as it affects surface energy balance, weather, climate, water supply, biochemical emissions, photochemical reactions, etc. The University of Alabama in Huntsville (UAH) and the NASA Short-term Prediction Research and Transition Center (SPoRT) have been generating and archiving several products, including insolation, from the Geostationary Operational Environmental Satellite (GOES) Imager for over a decade. The NASA/UAH insolation product has been used in studies to improve air quality simulations, biogenic emission estimates, correcting surface energy balance, and for cloud assimilation, but has not been thoroughly evaluated. In this study, the NASA/UAH insolation product is compared to surface pyranometer measurements from the Surface Radiation Budget Network (SURFRAD) and the U.S. Climate Reference Network (USCRN) for a 12-month period from March 2013 to February 2014. The insolation product has normalized bias values within 6% of the mean observation, a root-mean-square error between 6% and 16%, and correlation coefficients greater than 0.96 for hourly insolation estimates. It also shows better performance without the presence of clouds. However, erroneous estimates may be produced for persistent snow-covered surfaces. Further, this study attempts to demonstrate the use of such a satellite-based insolation product for model evaluation. The NASA/UAH insolation product is compared to the downward shortwave radiation from the Rapid Refresh, version 1 (RAPv1), and successfully captures the overestimation tendency in surface energy input as mentioned in previous studies. Finally, future plans for improving the retrieval algorithm and developing a GOES-16 insolation product are discussed.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Peiyang Cheng, peiyang.cheng@nsstc.uah.edu

Abstract

Incident solar radiation at Earth’s surface, also called surface insolation, plays an important role in the Earth system as it affects surface energy balance, weather, climate, water supply, biochemical emissions, photochemical reactions, etc. The University of Alabama in Huntsville (UAH) and the NASA Short-term Prediction Research and Transition Center (SPoRT) have been generating and archiving several products, including insolation, from the Geostationary Operational Environmental Satellite (GOES) Imager for over a decade. The NASA/UAH insolation product has been used in studies to improve air quality simulations, biogenic emission estimates, correcting surface energy balance, and for cloud assimilation, but has not been thoroughly evaluated. In this study, the NASA/UAH insolation product is compared to surface pyranometer measurements from the Surface Radiation Budget Network (SURFRAD) and the U.S. Climate Reference Network (USCRN) for a 12-month period from March 2013 to February 2014. The insolation product has normalized bias values within 6% of the mean observation, a root-mean-square error between 6% and 16%, and correlation coefficients greater than 0.96 for hourly insolation estimates. It also shows better performance without the presence of clouds. However, erroneous estimates may be produced for persistent snow-covered surfaces. Further, this study attempts to demonstrate the use of such a satellite-based insolation product for model evaluation. The NASA/UAH insolation product is compared to the downward shortwave radiation from the Rapid Refresh, version 1 (RAPv1), and successfully captures the overestimation tendency in surface energy input as mentioned in previous studies. Finally, future plans for improving the retrieval algorithm and developing a GOES-16 insolation product are discussed.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Peiyang Cheng, peiyang.cheng@nsstc.uah.edu

1. Introduction

As a key component of the surface energy budget, incident solar radiation (i.e., insolation) at Earth’s surface plays an important role in a wide range of studies, including weather prediction, climate monitoring, renewable energy production, water supply management, soil moisture and vegetation evapotranspiration estimation (Tarpley 1979; Gautier et al. 1980; Li et al. 1995; Otkin et al. 2005; Diak 2017; Huang and Tatcher 2017; Watanabe and Nohara 2019). Insolation also affects atmospheric chemistry by modulating photodissociation reaction rates (or photolysis rates) and influencing biogenic hydrocarbon emissions. Therefore, using a reliable insolation estimate can reduce the uncertainties in such studies (Pour-Biazar et al. 2007; Guenther et al. 2012).

Surface insolation can be measured instantaneously by ground-based pyranometers (Augustine et al. 2000, 2005; Diamond et al. 2013). These nonuniform point-source measurements are sparse and may not represent an accurate measure of the surface energy input over a large region (Gautier et al. 1980; Jacobs et al. 2002; Mecikalski et al. 2018; Paech et al. 2009; Podlasly and Berger 2002; Tarpley 1979; Li et al. 1995; Otkin et al. 2005).

To overcome the limitations of surface measurements, various physical models have been developed to retrieve surface insolation using observations from meteorological satellites (Diak and Gautier 1983; Schmetz 1989; Pinker et al. 1995; Haines et al. 2004; Diak et al. 2004; Habte et al. 2012, 2013; Gautier et al. 1980; Otkin et al. 2005; Diak 2017), including Geostationary Operational Environmental Satellite (GOES) (Haines et al. 2004) and Moderate Resolution Imaging Spectroradiometer (MODIS) (Justice et al. 1998). These models simulate radiative processes in the Earth-atmosphere column and calculate surface insolation based on shortwave radiation received by the satellite. Satellite-based surface insolation has been used in land surface carbon and water flux assessments (Jacobs et al. 2002; Anderson et al. 2011; Mecikalski et al. 2011, 2018), air quality simulations (White et al. 2018; McNider et al. 2018; Zhang et al. 2017, 2018; Tang et al. 2015; Pour-Biazar et al. 2007; Guenther et al. 2012), and agricultural forecast models (Diak et al. 1998, 2000; McNider et al. 2011, 2015). The advantage of satellite-based insolation is that the satellite offers continuous large-scale coverage over contiguous United States (CONUS) and physically describes cloud radiative effects (Gautier et al. 1980; Habte et al. 2012; Schmetz 1989). Previous studies have shown that, compared to surface measurements, standard errors for satellite-estimated insolation are typically 15%–20% on an hourly basis and about 10% for daily estimates (Jacobs et al. 2002; Schmetz 1989; Pinker et al. 1995). However, the accuracy of these retrievals decreases as the variability of cloud fields increase (Diak 2017; Gautier et al. 1980; Habte et al. 2013; Hong et al. 2016; Jacobs et al. 2002; Mecikalski et al. 2018; Paech et al. 2009; Podlasly and Berger 2002), especially for higher elevations (Otkin et al. 2005).

Since 2004, the NASA Short-term Prediction Research and Transition (SPoRT) Center has been operating the GOES Product Generation System (GPGS) to generate near-real-time GOES data products such as surface insolation, skin temperature, cloud-top temperature, cloud albedo, and surface albedo, utilizing GOES-12 and GOES-13 measurements (Haines et al. 2004). Since that time, in collaboration with the University of Alabama in Huntsville (UAH), the Imager products in this suite have been modified, improved, and reprocessed several times. The most recent improvements comprise addressing some of the key uncertainties in the retrieval code and extending the domain to cover all of CONUS. The current retrieval system uses a dynamic water vapor field as opposed to a constant value in the previous code. It also automatically accounts for the sensor degradation that was a source of concern in the previous version. Photosynthetically active radiation (PAR) was also added to the product suite (Pour-Biazar et al. 2015). The NASA/UAH insolation product has a spatial resolution of 4 km × 4 km and has been used in numerous studies to improve cloud simulations (Pour-Biazar et al. 2007; White et al. 2018), biogenic emission estimates (Zhang et al. 2018; Pour-Biazar et al. 2015), and agricultural forecast models (McNider et al. 2011, 2015). McNider et al. (2011) showed that the NASA/UAH insolation product outperforms the derived solar insolation obtained from meteorological observations. While these studies relied on limited evaluations of the NASA/UAH insolation product, a comprehensive evaluation has never been performed. Thus, it was necessary to validate the retrieval and to elucidate the limitations of the physical model. In this study, a rigorous evaluation is conducted by comparing the NASA/UAH insolation product to surface pyranometer observations to better understand the accuracy of the satellite product, identify the possible sources of uncertainty in the retrieval process, and offer possible ways to improve it.

A secondary objective of this study is to seek other potential applications of the archived NASA/UAH insolation product. For example, evaluating numerical models with respect to surface energy budget over the CONUS, or use of these data in retrospective studies. Surface radiation data from numerical models are used in various studies, including but not limited to solar power applications (Jimenez et al. 2016; Powers et al. 2017; Verbois et al. 2018), energy budget studies (Szeto et al. 2008), surface climate variability and change (Previdi et al. 2013), and air quality studies (Otte 2008). Thus, errors in insolation estimates will adversely affect these studies. Although model-based insolation estimates can be validated by surface pyranometer measurements, the NASA/UAH insolation product may serve as an alternative reference dataset that does not have the limitations of surface measurements. In this study, the NASA/UAH insolation product is therefore compared to a high-quality numerical weather prediction (NWP) model to illustrate the framework of validating model surface energy input. The choice of model was dictated by the availability of insolation in the model archived dataset, model spatial resolution being comparable with the relatively high-resolution GOES insolation retrievals, and model temporal resolution being comparable to the hourly satellite retrieval archives. Based on these constraints, archived data from the Rapid Refresh, version 1 (RAPv1; Benjamin et al. 2016), was used in the current study.

This paper is organized as follows: section 2 describes the datasets used, section 3 illustrates the methodology, section 4 discusses the results in detail, and section 5 summarizes the conclusion and states future work.

2. Datasets

A 12-month period from 1 March 2013 to 28 February 2014 was selected for this study due to high data availability. The data used are described in the following sections.

a. NASA/UAH insolation product

Details of the NASA/UAH insolation product are documented in Haines et al. (2004) and Pour-Biazar et al. (2007), along with several refinements documented in Pour-Biazar et al. (2015). The algorithm used for retrieving albedo and surface insolation follows Gautier et al. (1980) and Diak and Gautier (1983). The model utilizes the GOES Imager visible channel (~0.65 µm for GOES-13) with a ~1 km spatial resolution (at nadir) and employs a clear or a cloudy atmosphere model to simulate radiative transfer processes in the Earth system. The insolation model calculates surface albedo at each pixel for a given hour by using a visible channel 20-day clear-sky composite image, which is similar to the approach referred to as “minimum albedo method” (Schmetz 1989), whereas the cloud albedo is calculated by solving a quadratic equation [refer to Eq. (4) in Diak and Gautier 1983] based on the current visible channel measurement and surface albedo. The insolation is determined as the sum of incident solar radiation at Earth’s surface from both direct and diffuse light, with the effects of attenuation by the atmosphere and clouds included. To account for atmospheric attenuation, the model applies the effects of water vapor absorption (Paltridge 1973), Rayleigh scattering for the GOES-13 visible band (Coulson 1959) and ozone absorption (Lacis and Hansen 1974). Cloud absorption is assumed to be 7% of the incident flux at cloud top for opaque clouds. Furthermore, total precipitable water data from the North American Regional Reanalysis (NARR) are used to represent the absorption effects due to water vapor. There is also an automated procedure to correct the raw data for sensor degradation.

Table 1 shows how the major components of the NASA/UAH insolation retrieval method compared to a recent upgrade by Diak (2017). Neither insolation product deals with the following uncertainties: 1) anisotropic cloud effects due to bidirectional reflectance; 2) direct and indirect topographic effects on surface insolation; 3) external aerosol data to parameterize aerosol effects explicitly (instead, a small attenuation constant is used). Diak (2017) applies a near-linear function to parameterize cloud absorption (Gautier and Landsfeld 1997). A major difference between the two approaches is that the surface albedo used in NASA/UAH product shows a large variation during the day, while Diak’s technique uses a single noontime albedo.

Table 1.

Major components of the NASA/UAH insolation product and Diak (2017).

Table 1.

b. Pyranometer data

There are several operational pyranometer networks in the United States for validating satellite-derived surface radiation products. In this study, measurements from the Surface Radiation Budget Network (SURFRAD) (www.esrl.noaa.gov/gmd/grad/surfrad/sitepage.html, last accessed 7 November 2019) and the U.S. Climate Reference Network (USCRN) (www.ncdc.noaa.gov/crn/, last accessed 7 November 2019) are used.

The SURFRAD provides continuous, high-quality measurements of surface radiation fluxes over the United States (Augustine et al. 2000, 2005). It consists of seven stations with diverse climatic conditions and provides daily files of 1-min observations. These files include not only measurements of global horizontal irradiance, but direct and diffuse surface insolation values. For a better accuracy in this study, if both direct and diffuse solar radiation data are available and have good data quality, then the total irradiance is calculated by adding the product of the direct and the cosine of the solar zenith angle to the diffuse. Otherwise, the global horizontal irradiance values are used as the total irradiance.

The USCRN is a systematic and sustained network of climate monitoring stations (Diamond et al. 2013). Currently, it consists of 114 stations in the CONUS, 21 stations in Alaska, and 2 stations in Hawaii. Since 2012, after instrument upgrades, the USCRN has started to record 5-min averages of global solar radiation, but the direct and diffuse light are not reported separately. In this study, only good quality 5-min data from stations within the CONUS domain are used. After performing additional quality control on the USCRN data, measurements from six USCRN stations were not included in the following analysis. Five of the stations may have questionable reporting time (two sites in Washington State), calibration issue (one station in North Carolina), or monitors that were not clean in some seasons (one in California and one in Maine). For the sixth station, along the western coastline at Bodega Bay, California, there existed persistent marine stratocumulus clouds during the warm seasons, which may have caused an incorrect sky conditions in the retrievals.

In this comparison, there is an implicit assumption that pyranometer measurements are the true values. However, measurement uncertainty for even well-calibrated pyranometers is about 5% (Augustine et al. 2000; Diak 2017). Figure 1 shows monthly averaged clear-sky measurements from SURFRAD and USCRN sites at Sioux Falls, South Dakota. Since these two stations are almost collocated (100 ft apart), it is reasonable to assume that these sites always have the same sky condition and receive the same amount of shortwave radiation. Results indicate that they have about 5% normalized mean difference, which is consistent with the expected calibration error. Regardless of small fluctuations of the monthly average, which are probably due to different number of clear cases at each time stamp, a very high correlation coefficient (0.999 55) is found between the measurements. Therefore, in our evaluations, if the satellite-based insolation is within the uncertainty range of surface measurements (5%), we consider that to be in “good agreement.”

Fig. 1.
Fig. 1.

Monthly averaged (July 2013) clear-sky surface insolation measured at SURFRAD and USCRN Sioux Falls, SD, station. Spatial distance between two stations is ~100 ft.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

c. NWP model data

Currently, various meteorological models are generating downward shortwave radiation data for studies like climatology, renewable energy management, air quality modeling, land–atmosphere interaction, etc. (James et al. 2017; Huang and Thatcher 2017; Mass 2012; Gagne et al. 2017; Arbizu-Barrena et al. 2017; Watanabe and Nohara 2019). The NARR and the RAP are two reputable models that provide a suitable dataset for such studies. The NARR produces long-term, high-resolution (32 km/29 layer in space and 3 hourly in time), dynamically consistent atmospheric and land surface hydrology data over the North American domain (Mesinger et al. 2006). The RAP is an hourly updated NWP model, which produces parameters on a 13-km Northern Hemisphere Lambert Conformal Conic grid (Benjamin et al. 2016). The motivation for comparing the insolation fields from a reputable numerical model to the GOES insolation retrieval is to show the utility of NASA/UAH archived data for model evaluation. The archived insolation product provides hourly area-averaged observation that is comparable to model gridcell area. Thus, if the retrievals are within the surface observation’s uncertainty, then it can complement the pyranometer observations for model evaluation. Accordingly, downward solar radiation from the RAP was utilized on a priority basis. The RAP has a comparable temporal resolution to the GOES product, and its spatial resolution is relatively closer to the satellite data compared to the NARR. Therefore, with respect to the evaluation of total solar energy input, RAP was more suitable for the current study.

The RAP is an operational regional analysis and forecast system, which replaced its predecessor, the Rapid Update Cycle (RUC), in 2012. Instead of using the RUC forecast model and the RUC three-dimensional variational data assimilation (3DVar), the RAP utilizes the Advanced Research version of the Weather Research and Forecasting (WRF) Model and the Gridpoint Statistical Interpolation assimilation system (GSI) with modifications. The RAP modeling system has been updated several times since 2012 and the current operational version is RAPv4 (https://rapidrefresh.noaa.gov/, last accessed 7 November 2019). A complete description of the RAP (v1–v3) can be found in Benjamin et al. (2016). At the time of this study, the NWS/Cooperative Program for Operational Meteorology, Education and Training (COMET) archive of RAP (RAPv1 being interpolated to a smaller 13-km grid) (http://soostrc.comet.ucar.edu/data/grib/rap/, last accessed 7 November 2019) was the only long-term, publicly accessible RAP archive that provided downward shortwave radiation flux field. Thus, the RAPv1 archived dataset was used in the current study. One caveat here is that, Benjamin et al. (2016) points out that the RAP (before RAPv3) tends to overestimate surface insolation due to insufficient model-estimated cloudiness. Therefore, comparing the NASA/UAH insolation product to the RAPv1 radiation data in this study is only for the purpose of illustrating the framework of validating surface energy input of numerical models and does not represent the current performance of the RAP system. Results from this comparison only pertains to the quality of the NWS/COMET archived data used here.

To evaluate the RAP-estimated downward shortwave radiation data using the NASA/UAH insolation product, it is necessary to ensure that both datasets are on the same grid. Considering that the NASA/UAH insolation product has a higher spatial resolution (4 × 4 km2 vs 13 × 13 km2 for RAP), the satellite-derived insolation is mapped onto the RAP 13 × 13 km2 grid using the Environmental Protection Agency’s (EPA’s) Spatial Allocator tool (https://www.cmascenter.org/sa-tools/, last accessed 7 November 2019). Evaluation of the remapped insolation showed that most of the spatial patterns were successfully captured and preserved, while some broken cloud features disappeared due to reduced resolution.

3. Methodology

a. Evaluation metrics

To assess the degree of agreement between the NASA/UAH insolation product and pyranometer measurements, some commonly used paired in space/time evaluation metrics are used. These include mean bias error (MBE), root-mean-square error (RMSE), and coefficient of determination (R2) as detailed in Boylan and Russell (2006), Simon et al. (2012), and Ali and Abustan (2014). If the satellite-based insolation is denoted by M and pyranometer-measured insolation by O, then
MBE=1N (MnOn),
RMSE=1N (MnOn)2,
R2={ [(MnM¯)(OnO¯)] (MnM¯)2 (OnO¯)2}2.

Here, subscript n represents the record number and the overbar indicates the average quantity. Also, MBE and RMSE values are normalized as percentages of the mean measurements to show relative bias and error.

b. Hourly insolation comparison

Hourly estimates from the NASA/UAH insolation product were compared to pyranometer data for the daytime. To locate the satellite grid locations that contain the pyranometer sites, a grid search algorithm is applied. The algorithm computes spherical distances (Δi,jk) between the center of grids and pyranometer sites and selects the nearest grid to each pyranometer. The distance Δi,jk is calculated as
Δi,jk=Rcos1[sinϕi,j×sinΦk+cosϕi,j×cosΦk×cos(λi,jΛk)],
where R is the radius of Earth, Φk and Λk are latitude and longitude of pyranometer site k, and ϕi,j and λi,j are latitude and longitude of satellite grid (i, j). To compensate for the effects of moving clouds, satellite estimates within a 3 × 3 box around this nearest pixel are examined. If both pyranometer and satellite data are available in each scene, then the satellite estimate that is closest to pyranometer measurement is extracted for subsequent comparisons. This closest value is referred to as the best-fit value in the following.

Since pyranometer and satellite sensors have different working principles, it is not easy to design a direct comparison strategy. The pyranometers measure continuous solar hemispheric broadband radiation from a point location, which represents a temporal average over the sampling period, while the GOES-13 Imager takes a snapshot within a narrow visible band, which represents a spatial average over a ~1 km2 pixel in the midlatitudes. The insolation retrieval system compensates for this discrepancy by applying space/time translation and narrow-to-broadband conversion (Gautier et al. 1980). Moreover, pyranometer data are highly sensitive to transient or subpixel clouds. Yet the impact of subpixel clouds is relatively small on satellite data, since satellite sensors essentially average these cloud effects. Such discrepancy can produce uncertainties when comparing ground-based measurements to satellite-based insolation estimates. To compensate for these factors, a reasonable averaging time for pyranometer data is needed. To select such a reasonable averaging time, high frequency pyranometer data were averaged over different time intervals (10, 20, 30, and 60 min) and compared to satellite estimates. Based on this exercise, it was decided that averaging pyranometer data over one hour is the most suitable averaging time for the 4 × 4 km2 GOES data. A similar averaging time scale was also adopted by Habte et al. (2012, 2013).

To demonstrate this, 1-min pyranometer observations from SURFRAD Goodwin Creek station is examined in Fig. 2. Figure 2a shows broken clouds passing over the site for most of the day on 13 June. Even at times, surface insolation is greater than what is expected for clear sky. This is due to scattering from surrounding overhead clouds. The satellite measurements (Fig. 2b), which spatially average the reflected sunlight, are only comparable to pyranometer temporal average. Furthermore, a neighboring satellite pixel may better represent the averaged insolation over the station if clouds are moving from that direction. On 13 June, spatial variations of satellite estimates within the surrounding pixels are relatively larger due to the partly cloudy conditions compared to cloud-free hours on 14 June. However, the pyranometer hourly averages for 13 June agree with satellite retrievals (Fig. 2b). For the second day, 14 June, the scene is mostly sunny and the hourly variation in pyranometer measurement only represents the change in zenith angle. Due to relatively small change in zenith angle near solar noon, hourly variations of pyranometer data are small from 1100 to 1300 LT. It is also evident from Fig. 2b, that under clear-sky conditions, satellite-derived insolation estimates are consistent in the surrounding pixels and agree with hourly averaged pyranometer measurements. Moreover, for most of the sites, insolation estimates fall within one standard deviation of the mean measurements, which implies the confidence in the satellite retrievals. In short, the preprocessing treatment can effectively reduce the discrepancy between two different types of datasets, which is caused by different instrument working principles, and are essential to this validation work.

Fig. 2.
Fig. 2.

A two-day comparison of satellite-estimated and pyranometer-measured insolation data. Pyranometer data are measured at SURFRAD Goodwin Creek, MS, station. It is cloudy on 13 Jun 2013, but mostly clear on 14 Jun 2013. (a) One-minute raw measurements. (b) Hourly averages of pyranometer data (black lines), hourly variations of pyranometer data (vertical black bars), satellite best-fit values (blue crosses), and ranges of satellite estimates within the surrounding 3 × 3 box of pyranometer site (vertical blue bars).

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

c. Methods for calculating total surface energy input

For comparing model shortwave surface energy input versus satellite-observed quantity, daily cumulative insolation values are averaged over the entire CONUS for each season. The averages obtained from different datasets are then compared. With this treatment, the time discrepancy between the observations (15-min offset due to GOES-13 observations at 45 min after the hour) is implicitly eliminated, and the impact of transient clouds are largely reduced as well. The idea of computing seasonal area-weighted daily cumulative insolation averages is straightforward. First, hourly seasonal averages are calculated and then accumulated daily averages are estimated. Hourly estimates are sorted according to their time stamps, and are averaged over the entire season by
Ihourlyi,j(tl)¯=n=1N[Ihourlyi,jn(tl)×Ωi,jn(tl)]n=1NΩi,jn(tl),l=1,,24.

Here, i and j are grid indices, tl is hourly time stamp (24 time stamps in total), N is the number of days in each season, Ihourlyi,jn(tl) represents hourly insolation estimate at each grid and time, Ωi,jn(tl) is the data mask (1 for daytime good quality estimates and 0 for others), and Ihourlyi,j(tl)¯ indicates seasonal hourly insolation averages for each hour (unit: W m−2).

Then the hourly averages are integrated to obtain seasonal daily cumulative insolation averages Idailyi,j¯ (unit: J m−2) at each grid cell by the simple trapezoidal rule:
Idailyi,j¯l=124[Ihourlyi,j(tl1)¯+Ihourlyi,j(tl)¯2Δtl].
Here, Δtl is one hour. Once seasonal daily cumulative insolation averages at each grid cell are obtained, the seasonal total daily cumulative insolation over the entire CONUS Itotaldaily¯ (unit: J) is computed. This is accomplished by multiplying the area of each grid cell, Ai,j, and the U.S. land mask Ψi,j (1 for grids over the CONUS and 0 otherwise):
Itotaldaily¯=i=1Ij=1J(Idailyi,j¯×Ai,jΨi,j).

Here, the U.S. land mask Ψi,j is obtained from NASA Land Data Assimilation Systems (NLDAS; available online at https://ldas.gsfc.nasa.gov/nldas/NLDASspecs.php, last accessed 7 November 2019) and is mapped onto the RAP 13-km grid.

The seasonal area-weighted daily cumulative insolation averages Iaveragedaily¯¯ (unit: J m−2) can be calculated by dividing the domain seasonal totals by the area or the domain. The corresponding hourly averages Iaveragehourly¯¯ (unit: W m−2) are obtained by dividing the total daily averages by 24. By comparing these averages associated with the satellite retrieval and the model, the overall difference between the model-estimated and satellite-derived surface energy input for each season is obtained.

4. Results

a. Evaluation of satellite retrievals

1) Hourly statistics

Seasonal comparisons between the GOES hourly insolation estimates and surface measurements are summarized in Fig. 3. The MBE values are mostly positive in the summer (with a few exceptions over northeastern region) with MBE in the range of −10 to +20 W m−2 (−3% to +6% of the mean observed values) (Figs. 3b,f). Surface insolation is underestimated in the satellite retrievals in spring and fall (with exceptions in the south) by 10 to 20 W m−2 (Figs. 3a,c), or within 6% of the mean measurements (Figs. 3e,g). Winter MBE values are all negative, with the magnitudes being on the order of −20 to −10 W m−2 for stations in the southern United States, but are as large as −100 W m−2 (−60%) in the north where snow events are favored (Figs. 3d,h). Seasonal variation of the bias (Figs. 3a–h) seems to suggest that when the solar zenith angle is small (summer) satellite retrievals show a positive bias, while for the other seasons as the solar zenith angle becomes larger the bias becomes negative. At this point, it is not clear which assumptions in the retrieval algorithm is causing this trend.

Fig. 3.
Fig. 3.

Error statistics of satellite-estimated hourly insolation being compared to pyranometer measurements from combined SURFRAD and USCRN stations for four seasons in 2013. Each row presents MBE, normalized MBE, R2, RMSE, RMSE percentage, and RMSE percentage for clear-sky cases. Each column represents spring, summer, fall, and winter, respectively.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

The magnitude of hourly RMSE percentages (with few exceptions in the north) is about 6%–16% in all seasons except winter (Figs. 3q–t), indicating that satellite estimates are consistent with surface measurements when it is warm. Also, the RMSE values tend to be larger in the west compared to the other regions in the summer and in the fall (Figs. 3r,s), which may be caused by satellite viewing angle, or the lack of terrain and elevation effects on surface insolation in the retrieval system. The RMSE values can be very large (≥20%) in the northern United States in winter (Fig. 3t), and for a smaller number of stations in spring (Fig. 3q) and fall (Fig. 3s). Furthermore, as expected, seasonal R2 values (Figs. 3i–l) and RMSE percentages (Figs. 3q–t) are negatively correlated. Cases with small RMSE (less than 12%) normally have high coefficient of determination (R2 ≥ 0.96), while large RMSE (greater than 20%) corresponds to low R2 values (below 0.86).

To understand the performance of the satellite-based insolation technique under different sky conditions, insolation estimates are divided into two categories based on cloud albedo estimates. To avoid misrepresenting clear-sky aerosols as clouds, a 10% satellite cloud albedo threshold was chosen to separate clear and cloudy cases. It is shown that hourly RMSE values are much smaller for clear-sky cases (Figs. 3u–x) compared to all situations (Figs. 3q–t). Clear-sky RMSE values are less than 10% in the warm seasons (Figs. 3u–w) and below 16% even in winter (Fig. 3x). Note that these errors mostly represent random errors rather than bias errors. A previous study by Pour-Biazar et al. (2015) indicated systematic positive biases in the eastern and central United States, but negative in the western United States, in the original insolation product in the warm seasons. Thus, a bias correction process was carried out to minimize these biases and to improve data quality. In general, aside from cold seasons over snow-covered regions (as will be discussed in the following), the archived GOES-based insolation retrievals are within the 5% uncertainty of surface pyranometer measurements.

2) Snow contamination issue

To find possible causes of the retrieval system’s failure in colder weather, error statistics associated with unreliable winter insolation estimates are investigated. A satellite-based insolation is marked as unreliable if compared to surface pyranometer observation has normalized MBE ≥ 20%, relative RMSE ≥ 30%, and R2 ≤ 0.8. Results at individual observational sites that have unreliable winter insolation estimates are summarized in Table 2. Furthermore, these cases were divided into two categories based on their sky conditions: 1) cases with satellite cloud albedos >10% are assigned as cloudy, and 2) the rest are marked as clear. The thresholds for marking a retrieval unreliable under clear-sky and cloudy-sky conditions are the same as those for all-sky cases. In Table 2, 34 cases are listed as unreliable since they failed the threshold test. For clear-sky conditions, 33 of those listed as unreliable pass the threshold test and can be marked as reliable insolation estimates. The remaining single case, marked as unreliable, has only 3 available records (less than ~1% of the total), which indicates the error statistics may lack credibility and can be ignored. However, all the cloudy cases underestimated surface insolation, which implies that the retrieval system is more likely to fail for cloudy calculations in the winter, when more snow events are expected.

Table 2.

Error statistics associated with unreliable satellite-derived hourly insolation estimates under all-, clear-, and cloudy-sky conditions for 2013 winter. The unreliable values are highlighted with bold font style, whose thresholds are set as normalized MBE ≥ 20%, relative RMSE ≥ 30%, or R2 ≤ 0.8.

Table 2.

Therefore, one hypothesis for the systematic underestimation of insolation in colder weather is that, if the surface is very bright (e.g., covered by snow), the minimum albedo method may not correctly identify the clear scenes, and therefore, clear and cloudy calculations may not be applied properly. An important assumption in the physical model is that the surface albedo is always smaller than cloud albedo. However, this assumption may not always hold for snow-covered surfaces (as the surface can be brighter as seen by the satellite than when covered with clouds). If the surface is covered by snow for a long period of time (comparable to the time interval set in minimum albedo method), and the snow is more reflective than clouds, then the retrieval system will treat darker clouds as the surface (based on the assumption that cloud albedo is always greater than surface albedo). In such cases, clear-sky scenes are treated as cloudy and cloudy scenes are treated as clear sky, resulting in erroneous insolation estimates in the subsequent calculations. This is documented as snow contamination in the literature (Pinker 2003; Schmetz 1989; Otkin et al. 2005; Diak 2017), which is a significant unresolved issue in the current satellite-based insolation retrieval technique.

To test this hypothesis further, a filtering algorithm was developed to remove snow-related hourly estimates. The algorithm uses measurements of surface temperature and wetness from pyranometer stations and a MODIS snow-cover product (Hall and Riggs 2016) to determine if the surface is covered by snow. Wintertime data over three surface stations over different climate regions were examined and the results are shown in Fig. 4. The figure shows that the clear-sky insolation estimates are reasonable regardless of their locations (blue dots in Figs. 4a,c,e), but they are underestimated at the La Junta and Sioux Falls sites when the scene is not cloud-free (red dots in Figs. 4c,e). After applying the filtering algorithm, there is little change in the number of data points at the Austin site (Fig. 4b), where snow is uncommon in the winter. On the other hand, at the La Junta site (Fig. 4d), filtering data points associated with snow-covered surfaces increases the agreement. Consequently, MBE and RMSE decrease from 8.5% and 32% to 0.2% and 12%, while R2 value increases from 0.75 to 0.95. The Sioux Falls site (Fig. 4f), which has the worst winter estimates among the three, shows little improvement after filtering snow-related cases. It seems that the filtering algorithm removes most data points at this site. Similarly, at some other stations where the surface is mostly covered by snow in winter, this filtering algorithm removes most of the available data points and the statistics are not reliable.

Fig. 4.
Fig. 4.

Scatterplots of winter hourly insolation measured at three USCRN sites and derived from GOES-13 images before and after filtering snow-related cases. Blue and red dots indicate clear and cloudy cases, respectively. Error statistics are also presented in each panel.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

To further demonstrate the problem caused by snow-covered surface under cloudy conditions, the NASA/UAH archived wintertime insolation, cloud albedo, and surface albedo are examined over January 2014 for an area in southern Colorado. The center of this area was highly reflective with surface albedo values greater than 50% and consistently exhibited low surface insolation. However, the high cloud albedo values remained nearly stationary for 12 days as shown in Fig. 5. Continental clouds in the atmosphere are highly variable. It is not reasonable to observe such highly reflective clouds stay in the same place for a 12-day period. The only plausible explanation is the failure of satellite-based insolation retrieval technique that is treating the snow-covered surface as cloud.

Fig. 5.
Fig. 5.

Cloud albedo plots at 1945 UTC from 1 through 12 Jan 2014 for an area in southern Colorado.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

b. Model/satellite comparisons

Having confidence in GOES-retrieved insolation over non-snow-covered surfaces (based on the analysis in the previous section), it would be reasonable to use this insolation product for model evaluation. The NASA/UAH insolation archives have been used in several retrospective studies for either model evaluation (McNider et al. 2011) or substituting for the model insolation field (Pour-Biazar et al. 2007; Zhang et al. 2018; Pour-Biazar et al. 2015; McNider et al. 2015). The following comparison shows the value of this insolation product for evaluating other model archived data.

Before comparing the NASA/UAH insolation product to hourly downward shortwave radiation data from numerical models, one should be aware that there might be a 15-min offset. The GOES-13 images are at 45 min past the hour, while the model data (RAPv1 in this case) are on the hour. However, the large-scale (on the order of 100 km or larger) insolation patterns and cloud fields should not change much within such a short period of time as indicated in Fig. 6. The figure shows that the RAPv1 can capture major cloud features; however, model cloud coverage is underestimated when compared to the satellite retrieval. For example, a large-scale cloud cluster (green boxes in Fig. 6) can be observed over parts of the southern United States in both products, but the cluster shows fewer details and less cloudiness in the model (Figs. 6a,b) than the NASA/UAH insolation product (Figs. 6c,d). In addition, the model does not accurately capture the scattered clouds over the Sierra Nevada (black arrows in Fig. 6). Moreover, the difference plot (Figs. 6e,f) indicates that the RAPv1 consistently overestimates incoming solar energy under clear-sky conditions. The magnitude of overestimation is about 50 W m−2 in the west and 100 W m−2 in the east, which seems to be a common feature.

Fig. 6.
Fig. 6.

Plots of hourly insolation from the RAPv1 and the NASA/UAH insolation product, and their difference at 1800 and 2000 UTC 15 Jul 2013. The green boxes and black arrows refer to two of the areas where the RAPv1 insolation differs from the GOES. Also note the time for the NASA/UAH insolation product is 15 min earlier than the RAPv1.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

Seasonal averages indicate that daily bias of the NASA/UAH insolation product is generally less than ~1.5 MJ m−2 (~6% of the mean) compared to USCRN observations at the Austin, La Junta, and Sioux Falls stations (Fig. 7) except at the Sioux Falls site in the winter (due to surface snow cover) (Fig. 7l). During the spring and summer periods in this study, the RAPv1 on the average overestimates incident solar energy by over 4 MJ m−2 (up to 6.63 MJ m−2) for these three stations. While the difference is lower in fall and winter, there still exists about 3 MJ m−2 overestimation in daily total insolation at the Austin and La Junta stations. Moreover, the model tends to overestimate surface insolation starting near noon and persisting for the rest of the day. As mentioned in section 2c, the RAP (before version 3) has the tendency to overestimate surface insolation due to insufficient clouds (Benjamin et al. 2016). As evident in the figure, the NASA/UAH insolation product may be an alternative to surface pyranometer measurements for evaluating the model surface shortwave radiation.

Fig. 7.
Fig. 7.

Seasonal averaged hourly insolation from the RAPv1, the NASA/UAH insolation product, and USCRN measurements at three USCRN stations. Each column represents spring, summer, fall, and winter, respectively. Difference of seasonal averaged daily total insolation between estimates and pyranometer measurements are also shown in each panel. Notice that due to a 15-min offset between the RAPv1 and NASA/UAH insolation product, pyranometer measurements at these sites are averaged accordingly (on the hour for the RAPv1 and at 45 min past the hour for the NASA/UAH insolation product).

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

Additionally, hourly and daily insolation from both the GOES-13 and the RAPv1 are compared against pyranometer measurements to demonstrate the scatter in the data. As evident in Fig. 8, compared with surface measurements, the GOES insolation product shows less scatter for the summer than the other seasons. There is more scatter for the winter when the satellite retrieval tends to underestimate insolation (as well as some of the spring cases) as previously discussed. For the RAP–pyranometer comparison, however, there is more scatter in the top-left corner of each panel, indicating the model consistently overestimates insolation for all seasons. The positive bias is largest in the summer and smallest in the winter. Such features are more obvious for daily cumulative results (Fig. 9). Daily cumulative results greatly reduce the scatter for the GOES retrieval over all seasons, although the underestimation pattern remains for winter and a few spring cases. The overestimation issue of the RAPv1 is even more noticeable in the daily cumulative scatterplots, indicating that the positive bias is consistent. Some of the scatter is thought to be due pyranometer measurement issues (e.g., sensor calibration or reflective surrounding). For completeness, however, unreliable pyranometer data were not removed from these scatterplots.

Fig. 8.
Fig. 8.

Scatterplots of hourly insolation showing pyranometer measurements against (left) GOES product and (right) RAPv1 estimates. All data points are binned with 5 W m−2 bin width to reduce point density. The color indicates number of points in each bin. Each row (from top to bottom) represents spring, summer, fall, and winter, respectively. Black and red lines are of 1:1 reference and best fit, respectively. The best-fit equations of original data points are also presented in each panel.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

Fig. 9.
Fig. 9.

Scatterplots of daily cumulative insolation showing pyranometer measurements against (left) GOES product and (right) RAPv1 estimates. Each row (from top to bottom) represents spring, summer, fall, and winter, respectively. Black and red lines are of 1:1 reference and best fit, respectively. The best-fit equations are also presented in each panel.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

As shown in Fig. 10, the NASA/UAH insolation product is a reliable approximation of the ground measurements when most of the domain is snow-free. The GOES product overestimates daily total insolation by about 0.04 MJ m−2 on the average. However, the RAPv1 appears to overestimate surface insolation throughout the day. The total daily overestimation by RAPv1 is 4.66 MJ m−2 per day, which is more than 100 times the satellite overestimation. This excessive solar energy input in the model can be a large source of uncertainty in subsequent climatology, energy production, or air quality studies relying on model data. The model shows a tendency to overestimate surface insolation with larger magnitude in the afternoon hours, which corresponds to the feature shown in Fig. 7.

Fig. 10.
Fig. 10.

Averaged hourly insolation from the RAPv1 (red), the NASA/UAH insolation product (blue), and pyranometers at all USCRN locations over a 6-month warm period (1 May through 31 Oct 2013). Differences of daily cumulative insolation between estimates from the RAPv1/GOES and measurements from the USCRN are also shown.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

Finally, to quantify how much incident shortwave radiation is reaching the surface in the NASA/UAH insolation product and the numerical model, seasonal area-weighted daily total insolation averages of both datasets at each model grid box over the CONUS are calculated (Fig. 11). The difference between these two datasets is also presented. It should be noted, however, that the satellite retrievals in the northern United States for snow-covered surfaces are not reliable (the northern part in Fig. 11h). As evident from the figure, the RAPv1 overestimates surface insolation systematically over the whole domain in spring, summer, and fall. Additionally, the GOES insolation is likely to have higher MBE values over the mountainous regions, such as the Rocky Mountains and the Appalachian Mountains. One possible reason is that the satellite-based retrieval system ignores topography effects on solar radiation.

Fig. 11.
Fig. 11.

Seasonal area-weighted daily cumulative insolation averages from the RAPv1 and GOES, and the difference between them. Each column represents spring, summer, fall, and winter, respectively.

Citation: Journal of Atmospheric and Oceanic Technology 37, 4; 10.1175/JTECH-D-19-0058.1

Quantitatively, differences of seasonal area-weighted daily cumulative insolation averages between the RAPv1 and the satellite retrieval are about 6.2, 5.3, and 3.4 MJ m−2 in spring, summer, and fall, while differences of hourly averages are 71.8, 61.7, and 39.4 W m−2, respectively. To put these energy differences into perspective, the radiative forcing of carbon dioxide (CO2) is about 1.68 W m−2 as stated in Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (Myhre et al. 2013). Thus, the average model error in surface incident energy for this study period is about 34 times the radiative forcing of CO2.

5. Conclusions

This study presented the first comprehensive evaluation of the NASA/UAH GOES insolation product that had been used in many previous investigations. Hourly insolation estimates from the NASA/UAH GOES-based insolation product have been compared to surface pyranometer measurements from the SURFRAD and USCRN for a 12-month period from 1 March 2013 to 28 February 2014 on a seasonal basis. It is difficult to make such comparison since 1) the satellite retrieval represents spatial averages while surface measurements are point measurements, and 2) various sources of uncertainty exist in both the satellite retrieval and surface measurements. However, the preprocessing treatment in this study addressed these discrepancies and effectively reduced the uncertainties.

Results indicate that the NASA/UAH insolation is closer to the pyranometer measurements when the surface is not persistently covered with snow. Additionally, the results from this study are comparable to Diak (2017), although the retrievals and the details of the evaluation process are different. For hourly estimates, compared to pyranometer measurements, GOES-based insolation seasonal MBE values are generally within ±20 W m−2 (±6% of mean surface measurements), which tend to be positive in the summer but negative in the other seasons. This seasonal overestimation/underestimation tendency is thought to be caused by variations in the solar zenith angle. Seasonal RMSE percentages are about 6%–16% and R2 values are mostly greater than 0.96, indicating the accuracy of the NASA/UAH insolation product. The results also suggest that the hourly insolation estimates are closer to surface pyranometer measurements under clear-sky conditions than cloudy cases. This is expected as the pyranometer measurements exhibit large temporal variation under partly cloudy sky, while the satellite detects the average reflection from a 4 × 4 km2 area. The NASA/UAH insolation product may produce erroneous estimates during the cold months in the northern United States when the surface is persistently covered with snow over two weeks. The snow-filtering algorithm has shown the potential to improve insolation estimates by removing snow-affected cases, but it may not work as expected for all locations. Furthermore, by separating unreliable winter results into clear and cloudy conditions, it became obvious that cloudy estimates are more likely affected by snow than clear-sky cases. This perhaps can be a guide on where to improve the insolation retrieval over snow-covered surfaces in future work. A detailed examination of the NASA/UAH products has shown the evidence of constant cloud albedo values over regions with high surface reflection, which implies that the albedo module of the retrieval system may fail to distinguish clouds from very bright surfaces (e.g., snow-covered surface). This is consistent with the previous studies identifying snow contamination as a major issue for physical-based satellite insolation retrieval techniques (Cano et al. 1986; Ohmura et al. 1998; Schmetz 1989; Pinker 2003; Otkin et al. 2005; Diak 2017). The NASA/UAH daily total insolation was only off by ~0.04 MJ m−2 from the USCRN observations, meaning that it can be used as an alternative to the surface insolation measurements.

To test if the NASA/UAH insolation product is suitable for evaluating model surface energy input, the downward shortwave radiation data from the NWS/COMET archive of the WRF-based RAPv1 (as a typical NWP model) was used and compared to the satellite insolation retrieval. The results indicated that the archived model data 1) systematically overestimated the surface insolation, and 2) the positive bias in the model was more pronounced in the afternoon than the midmorning period. This is consistent with the previous finding that the RAP models prior to RAPv3 tend to underestimate cloud formation and overestimate surface insolation (Benjamin et al. 2016). Quantitative comparison of seasonal area-weighted daily cumulative insolation averages between the RAPv1 and the NASA/UAH insolation product indicates that the RAPv1 overestimates daily averaged incident solar radiation by 6.2, 5.3, and 3.4 MJ m−2 in spring, summer, and fall, respectively. These are equivalent to 71.8, 61.7, and 39.4 W m−2 for hourly estimates, respectively. This excessive energy is more than 30 times the radiative forcing of CO2 and can be a serious problem when the model results are used in energy balance studies. With the satellite–model comparison, one may easily determine whether the model surface energy input is on the reasonable order or not, and the results can provide more spatial details than surface pyranometer measurements.

The snow contamination issue remains a major factor causing the failure of satellite-based insolation retrieval techniques and needs to be addressed. Additionally, topography effects, surface elevation, cloud bidirectional reflectance, and aerosol parameterizations are not explicitly dealt with in the satellite-based insolation retrieval system. Therefore, efforts can also be made to incorporate one or more of these factors into the system. The priority for these improvements should be weighed against the computational limitations as indicated in Diak (2017). While this study evaluated the archived data that are mostly based on GOES-13, the physical model is undergoing an upgrade and will continue to produce the insolation product based on observations by GOES-R series of satellites (currently GOES-16 and GOES-17). The Advanced Baseline Imager (ABI; Schmit et al. 2017) on board GOES-16 is providing data at higher temporal and spatial resolution. The upgrade will take advantage of these advancements and will address some of the issues indicated above.

In general, the results here suggest that the satellite-derived insolation estimates can reasonably represent surface measurements in most cases (except for the snow-covered surfaces). The framework of how the NASA/UAH insolation product can be used for model evaluation is also demonstrated. This satellite–model comparison may provide more details than pyranometer-model comparison, since the density of surface monitors is currently limited. Furthermore, the satellite insolation retrieval may be used in retrospective studies to reduce errors in the model insolation field.

Acknowledgments

The authors would like to acknowledge Drs. Kevin Doty, Yuling Wu, and Andrew White for their contributions to this study. The authors would also like to thank two of the three reviewers for providing helpful comments and suggestions. One reviewer, in particular, provided an extensive and detailed review to help improve this manuscript. The findings presented here were accomplished under partial support from NASA Science Mission Directorate Applied Sciences Program and the Texas Commission on Environmental Quality (TCEQ). Note the results in this study do not necessarily reflect policy or science positions by the funding agencies.

REFERENCES

  • Ali, M. H., and I. Abustan, 2014: A new novel index for evaluating model performance. J. Nat. Resour. Dev., 4, 19, https://doi.org/10.5027/JNRD.V4I0.01.

    • Search Google Scholar
    • Export Citation
  • Anderson, M. C., and Coauthors, 2011: Mapping daily evapotranspiration at field to continental scales using geostationary and polar orbiting satellite imagery. Hydrol. Earth Syst. Sci., 15, 223239, https://doi.org/10.5194/hess-15-223-2011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Arbizu-Barrena, C., J. A. Ruiz-Arias, F. J. Rodríguez-Benítez, D. Pozo-Vázquez, and J. Tovar-Pescador, 2017: Short-term solar radiation forecasting by advecting and diffusing MSG cloud index. Sol. Energy, 155, 10921103, https://doi.org/10.1016/j.solener.2017.07.045.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Augustine, J. A., J. J. DeLuisi, and C. N. Long, 2000: SURFRAD—A national surface radiation budget network for atmospheric research. Bull. Amer. Meteor. Soc., 81, 23412357, https://doi.org/10.1175/1520-0477(2000)081<2341:SANSRB>2.3.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Augustine, J. A., G. B. Hodges, C. R. Cornwall, J. J. Michalsky, and C. I. Medina, 2005: An update on SURFRAD—The GCOS surface radiation budget network for the continental United States. J. Atmos. Oceanic Technol., 22, 14601472, https://doi.org/10.1175/JTECH1806.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2016: A North American hourly assimilation and model forecast cycle: The Rapid Refresh. Mon. Wea. Rev., 144, 16691694, https://doi.org/10.1175/MWR-D-15-0242.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boylan, J. W., and A. G. Russell, 2006: PM and light extinction model performance metrics, goals, and criteria for three-dimensional air quality models. Atmos. Environ., 40, 49464959, https://doi.org/10.1016/j.atmosenv.2005.09.087.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cano, D., J. M. Monget, M. Albuisson, H. Guillard, N. Regas, and L. Wald, 1986: A method for the determination of the global solar radiation from meteorological satellite data. Sol. Energy, 37, 3139, https://doi.org/10.1016/0038-092X(86)90104-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coulson, K. L., 1959: Characteristics of the radiation emerging from the top of a Rayleigh atmosphere—II: Total upward flux and albedo. Planet. Space Sci., 1, 277284, https://doi.org/10.1016/0032-0633(59)90032-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., 2017: Investigations of improvements to an operational GOES-satellite-data-based insolation system using pyranometer data from the U.S. Climate Reference Network (USCRN). Remote Sens. Environ., 195, 7995, https://doi.org/10.1016/j.rse.2017.04.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., and C. Gautier, 1983: Improvements to a simple physical model for estimating insolation from GOES data. J. Climate Appl. Meteor., 22, 505508, https://doi.org/10.1175/1520-0450(1983)022%3C0505:ITASPM%3E2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., M. C. Anderson, W. L. Bland, J. M. Norman, J. M. Mecikalski, and R. M. Aune, 1998: Agricultural-management decision aids driven by real-time satellite data. Bull. Amer. Meteor. Soc., 79, 13451355, https://doi.org/10.1175/1520-0477(1998)079<1345:AMDADB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., W. L. Bland, J. R. Mecikalski, and M. C. Anderson, 2000: Satellite-based estimates of longwave radiation for agricultural applications. Agric. For. Meteor., 103, 349355, https://doi.org/10.1016/S0168-1923(00)00141-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., J. R. Mecikalski, M. C. Anderson, J. M. Norman, W. P. Kustas, R. D. Torn, and R. L. DeWolf, 2004: Estimating land surface energy budgets from space: Review and current efforts at the University of Wisconsin—Madison and USDA–ARS. Bull. Amer. Meteor. Soc., 85, 6578, https://doi.org/10.1175/BAMS-85-1-65.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diamond, H. J., and Coauthors, 2013: U.S. Climate Reference Network after one decade of operations status and assessment. Bull. Amer. Meteor. Soc., 94, 485498, https://doi.org/10.1175/BAMS-D-12-00170.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gagne, D. J., II, A. McGovern, S. E. Haupt, and J. K. Williams, 2017: Evaluation of statistical learning configurations for gridded solar irradiance forecasting. Sol. Energy, 150, 383393, https://doi.org/10.1016/j.solener.2017.04.031.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gautier, C., and M. Landsfeld, 1997: Surface solar radiation flux and cloud radiative forcing for the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP): A satellite, surface observations, and radiative transfer model study. J. Atmos. Sci., 54, 12891307, https://doi.org/10.1175/1520-0469(1997)054<1289:SSRFAC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gautier, C., G. Diak, and S. Masse, 1980: A simple physical model to estimate incident solar radiation at the surface from GOES satellite data. J. Appl. Meteor. Climatol., 19, 10051012, https://doi.org/10.1175/1520-0450(1980)019<1005:ASPMTE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guenther, A. B., X. Jiang, C. L. Heald, T. Sakulyanontvittaya, T. Duhl, L. K. Emmons, and X. Wang, 2012: The Model of Emissions of Gases and Aerosols from Nature version 2.1 (MEGAN2.1): An extended and updated framework for modeling biogenic emissions. Geosci. Model Dev., 5, 14711492, https://doi.org/10.5194/gmd-5-1471-2012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Habte, A., M. Sengupta, and S. Wilcox, 2012: Surface radiation from GOES: A physical approach. 27th European Photovoltaic Solar Energy Conf. and Exhibition, Frankfurt, Germany, WIP Renewable Energy, 3052–3059.

  • Habte, A., M. Sengupta, and S. Wilcox, 2013: Validation of GOES-derived surface radiation using NOAA’s physical retrieval method. National Renewable Energy Laboratory Tech. Rep. NREL/TP-5500-57442, 20 pp., https://www.nrel.gov/docs/fy13osti/57442.pdf.

    • Crossref
    • Export Citation
  • Haines, S. L., R. J. Suggs, and G. J. Jedlovec, 2004: The Geostationary Operational Environmental Satellite (GOES) product generation system. NASA Tech. Memo. NASA TM-2004-213286, 52 pp., https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20050019524.pdf.

  • Hall, D. K., and G. A. Riggs, 2016: MODIS/Terra snow cover daily L3 global 0.05deg CMG, version 6. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 7 November 2019, https://doi.org/10.5067/MODIS/MOD10C1.006.

    • Crossref
    • Export Citation
  • Hong, Y., G. Liu, and J. L. F. Li, 2016: Assessing the radiative effects of global ice clouds based on CloudSat and CALIPSO measurements. J. Climate, 29, 76517674, https://doi.org/10.1175/JCLI-D-15-0799.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, J., and M. Thatcher, 2017: Assessing the value of simulated regional weather variability in solar forecasting using numerical weather prediction. Sol. Energy, 144, 529539, https://doi.org/10.1016/j.solener.2017.01.058.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jacobs, J. M., D. A. Myers, M. C. Anderson, and G. R. Diak, 2002: GOES surface insolation to estimate wetlands evapotranspiration. J. Hydrol., 266, 5365, https://doi.org/10.1016/S0022-1694(02)00117-8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • James, E. P., S. G. Benjamin, and M. Marquis, 2017: A unified high-resolution wind and solar dataset from a rapidly updating numerical weather prediction model. Renewable Energy, 102, 390405, https://doi.org/10.1016/j.renene.2016.10.059.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jimenez, P. A., and Coauthors, 2016: WRF-Solar: Description and clear-sky assessment of an augmented NWP model for solar power prediction. Bull. Amer. Meteor. Soc., 97, 12491264, https://doi.org/10.1175/BAMS-D-14-00279.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Justice, C. O., and Coauthors, 1998: The Moderate Resolution Imaging Spectroradiometer (MODIS): Land remote sensing for global change research. IEEE Trans. Geosci. Remote Sens., 36, 12281249, https://doi.org/10.1109/36.701075.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lacis, A. A., and J. E. Hansen, 1974: A parameterization for absorption of solar radiation in the Earth’s atmosphere. J. Atmos. Sci., 31, 118133, https://doi.org/10.1175/1520-0469(1974)031<0118:APFTAO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, Z., C. H. Whitlock, and T. P. Charlock, 1995: Assessment of the global monthly mean surface insolation estimated from satellite measurements using global energy balance archive data. J. Climate, 8, 315328, https://doi.org/10.1175/1520-0442(1995)008<0315:AOTGMM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mass, C., 2012: Nowcasting: The promise of new technologies of communication, modeling, and observations. Bull. Amer. Meteor. Soc., 93, 797809, https://doi.org/10.1175/BAMS-D-11-00153.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNider, R. T., J. R. Christy, D. Moss, K. Doty, C. Handyside, A. Limaye, A. G. Y. Garcia, and G. Hoogenboom, 2011: A real-time gridded crop model for assessing spatial drought stress on crops in the southeastern United States. J. Appl. Meteor. Climatol., 50, 14591475, https://doi.org/10.1175/2011JAMC2476.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNider, R. T., and Coauthors, 2015: An integrated crop and hydrologic modeling system to estimate hydrologic impacts of crop irrigation demands. Environ. Modell. Software, 72, 341355, https://doi.org/10.1016/j.envsoft.2014.10.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNider, R. T., and Coauthors, 2018: Examination of the physical atmosphere in the Great Lakes region and its potential impact on air quality—Overwater stability and satellite assimilation. J. Appl. Meteor. Climatol., 57, 27892816, https://doi.org/10.1175/JAMC-D-17-0355.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mecikalski, J. R., D. M. Sumner, J. M. Jacobs, C. S. Pathak, S. J. Paech, and E. M. Douglas, 2011: Use of visible geostationary operational meteorological satellite imagery in mapping reference and potential evapotranspiration over Florida. Evapotranspiration, 446, 229254, https://doi.org/10.5772/14478.

    • Search Google Scholar
    • Export Citation
  • Mecikalski, J. R., W. B. Shoemaker, Q. Wu, M. A. Holmes, S. J. Paech, and D. M. Sumner, 2018: High-resolution GOES insolation–evapotranspiration data set for water resource management in Florida: 1995–2015. J. Irrig. Drain. Eng., 144, 04018025, https://doi.org/10.1061/(ASCE)IR.1943-4774.0001312.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, https://doi.org/10.1175/BAMS-87-3-343.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Myhre, G., and Coauthors, 2013: Anthropogenic and natural radiative forcing. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 659–740.

  • Ohmura, A., and Coauthors, 1998: Baseline Surface Radiation Network (BSRN/WCRP): New precision radiometry for climate research. Bull. Amer. Meteor. Soc., 79, 21152136, https://doi.org/10.1175/1520-0477(1998)079<2115:BSRNBW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Otkin, J. A., M. C. Anderson, J. R. Mecikalski, and G. R. Diak, 2005: Validation of GOES-based insolation estimates using data from the U.S. Climate Reference Network. J. Hydrometeor., 6, 460475, https://doi.org/10.1175/JHM440.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Otte, T. L., 2008: The impact of nudging in the meteorological model for retrospective air quality simulations. Part II: Evaluating collocated meteorological and air quality observations. J. Appl. Meteor. Climatol., 47, 18681887, https://doi.org/10.1175/2007JAMC1791.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Paech, S. J., J. R. Mecikalski, D. M. Sumner, C. S. Pathak, Q. Wu, S. Islam, and T. Sangoyomi, 2009: A calibrated, high-resolution GOES satellite solar insolation product for a climatology of Florida evapotranspiration. J. Amer. Water Resour. Assoc., 45, 13281342, https://doi.org/10.1111/j.1752-1688.2009.00366.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Paltridge, G. W., 1973: Direct measurement of water vapor absorption of solar radiation in the free atmosphere. J. Atmos. Sci., 30, 156160, https://doi.org/10.1175/1520-0469(1973)030<0156:DMOWVA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., 2003: Surface radiation budgets in support of the GEWEX Continental-Scale International Project (GCIP) and the GEWEX Americas Prediction Project (GAPP), including the North American Land Data Assimilation System (NLDAS) project. J. Geophys. Res., 108, 8844, https://doi.org/10.1029/2002JD003301.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., R. Frouin, and Z. Li, 1995: A review of satellite methods to derive surface shortwave irradiance. Remote Sens. Environ., 51, 108124, https://doi.org/10.1016/0034-4257(94)00069-Y.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Podlasly, C., and F. H. Berger, 2002: Insolation estimates for the LITFASS area derived from high resolution satellite data. Theor. Appl. Climatol., 73, 8795, https://doi.org/10.1007/s00704-002-0695-3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pour-Biazar, A., and Coauthors, 2007: Correcting photolysis rates on the basis of satellite observed clouds. J. Geophys. Res., 112, D10302, https://doi.org/10.1029/2006JD007422.

    • Search Google Scholar
    • Export Citation
  • Pour-Biazar, A., R. T. McNider, A. White, D. S. Cohan, and R. Zhang, 2015: Incorporating space-borne observations to improve biogenic emission estimates in Texas. Texas Air Quality Research Program Rep., 13 pp., http://aqrp.ceer.utexas.edu/projectinfoFY14_15%5C14-017%5C14-017%20QAPP.pdf.

  • Powers, J. G., and Coauthors, 2017: The Weather Research and Forecasting Model: Overview, system efforts, and future directions. Bull. Amer. Meteor. Soc., 98, 17171737, https://doi.org/10.1175/BAMS-D-15-00308.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Previdi, M., K. L. Smith, and L. M. Polvani, 2013: The Antarctic atmospheric energy budget. Part I: Climatology and intraseasonal-to-interannual variability. J. Climate, 26, 64066418, https://doi.org/10.1175/JCLI-D-12-00640.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmetz, J., 1989: Towards a surface radiation climatology: Retrieval of downward irradiances from satellites. Atmos. Res., 23, 287321, https://doi.org/10.1016/0169-8095(89)90023-9.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., P. Griffith, M. M. Gunshor, J. M. Daniels, S. J. Goodman, and W. J. Lebair, 2017: A closer look at the ABI on the GOES-R series. Bull. Amer. Meteor. Soc., 98, 681698, https://doi.org/10.1175/BAMS-D-15-00230.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Simon, H., K. R. Baker, and S. Phillips, 2012: Compilation and interpretation of photochemical model performance statistics published between 2006 and 2012. Atmos. Environ., 61, 124139, https://doi.org/10.1016/j.atmosenv.2012.07.012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Szeto, K. K., H. Tran, M. D. MacKay, R. Crawford, and R. E. Stewart, 2008: The MAGS water and energy budget study. J. Hydrometeor., 9, 96115, https://doi.org/10.1175/2007JHM810.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tang, W., and Coauthors, 2015: Influence of satellite-derived photolysis rates and NOx emissions on Texas ozone modeling. Atmos. Chem. Phys., 15, 16011619, https://doi.org/10.5194/acp-15-1601-2015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tarpley, J. D., 1979: Estimating incident solar radiation at the surface from geostationary satellite data. J. Appl. Meteor., 18, 11721181, https://doi.org/10.1175/1520-0450(1979)018<1172:EISRAT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Verbois, H., R. Huva, A. Rusydi, and W. Walsh, 2018: Solar irradiance forecasting in the tropics using numerical weather prediction and statistical learning. Sol. Energy, 162, 265277, https://doi.org/10.1016/j.solener.2018.01.007.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Watanabe, T., and D. Nohara, 2019: Prediction of time series for several hours of surface solar irradiance using one-granule cloud property data from satellite observations. Sol. Energy, 186, 113125, https://doi.org/10.1016/j.solener.2019.05.004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • White, A. T., A. Pour-Biazar, K. Doty, B. Dornblaser, and R. T. McNider, 2018: Improving cloud simulation for air quality studies through assimilation of geostationary satellite observations in retrospective meteorological modeling. Mon. Wea. Rev., 146, 2948, https://doi.org/10.1175/MWR-D-17-0139.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, F., and X. Wu, 2013: Radiometric calibration accuracy of GOES sounder infrared channels. IEEE Trans. Geosci. Remote Sens., 51, 11871199, https://doi.org/10.1109/TGRS.2012.2219625.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, R., A. Cohan, A. Pour-Biazar, and D. S. Cohan, 2017: Source apportionment of biogenic contributions to ozone formation over the United States. Atmos. Environ., 164, 819, https://doi.org/10.1016/j.atmosenv.2017.05.044.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Zhang, R., A. T. White, A. Pour-Biazar, R. T. McNider, and D. S. Cohan, 2018: Incorporating GOES satellite photosynthetically active radiation (PAR) retrievals to improve biogenic emission estimates in Texas. J. Geophys. Res. Atmos., 123, 13091324, https://doi.org/10.1002/2017JD026792.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save
  • Ali, M. H., and I. Abustan, 2014: A new novel index for evaluating model performance. J. Nat. Resour. Dev., 4, 19, https://doi.org/10.5027/JNRD.V4I0.01.

    • Search Google Scholar
    • Export Citation
  • Anderson, M. C., and Coauthors, 2011: Mapping daily evapotranspiration at field to continental scales using geostationary and polar orbiting satellite imagery. Hydrol. Earth Syst. Sci., 15, 223239, https://doi.org/10.5194/hess-15-223-2011.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Arbizu-Barrena, C., J. A. Ruiz-Arias, F. J. Rodríguez-Benítez, D. Pozo-Vázquez, and J. Tovar-Pescador, 2017: Short-term solar radiation forecasting by advecting and diffusing MSG cloud index. Sol. Energy, 155, 10921103, https://doi.org/10.1016/j.solener.2017.07.045.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Augustine, J. A., J. J. DeLuisi, and C. N. Long, 2000: SURFRAD—A national surface radiation budget network for atmospheric research. Bull. Amer. Meteor. Soc., 81, 23412357, https://doi.org/10.1175/1520-0477(2000)081<2341:SANSRB>2.3.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Augustine, J. A., G. B. Hodges, C. R. Cornwall, J. J. Michalsky, and C. I. Medina, 2005: An update on SURFRAD—The GCOS surface radiation budget network for the continental United States. J. Atmos. Oceanic Technol., 22, 14601472, https://doi.org/10.1175/JTECH1806.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Benjamin, S. G., and Coauthors, 2016: A North American hourly assimilation and model forecast cycle: The Rapid Refresh. Mon. Wea. Rev., 144, 16691694, https://doi.org/10.1175/MWR-D-15-0242.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boylan, J. W., and A. G. Russell, 2006: PM and light extinction model performance metrics, goals, and criteria for three-dimensional air quality models. Atmos. Environ., 40, 49464959, https://doi.org/10.1016/j.atmosenv.2005.09.087.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cano, D., J. M. Monget, M. Albuisson, H. Guillard, N. Regas, and L. Wald, 1986: A method for the determination of the global solar radiation from meteorological satellite data. Sol. Energy, 37, 3139, https://doi.org/10.1016/0038-092X(86)90104-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Coulson, K. L., 1959: Characteristics of the radiation emerging from the top of a Rayleigh atmosphere—II: Total upward flux and albedo. Planet. Space Sci., 1, 277284, https://doi.org/10.1016/0032-0633(59)90032-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., 2017: Investigations of improvements to an operational GOES-satellite-data-based insolation system using pyranometer data from the U.S. Climate Reference Network (USCRN). Remote Sens. Environ., 195, 7995, https://doi.org/10.1016/j.rse.2017.04.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., and C. Gautier, 1983: Improvements to a simple physical model for estimating insolation from GOES data. J. Climate Appl. Meteor., 22, 505508, https://doi.org/10.1175/1520-0450(1983)022%3C0505:ITASPM%3E2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., M. C. Anderson, W. L. Bland, J. M. Norman, J. M. Mecikalski, and R. M. Aune, 1998: Agricultural-management decision aids driven by real-time satellite data. Bull. Amer. Meteor. Soc., 79, 13451355, https://doi.org/10.1175/1520-0477(1998)079<1345:AMDADB>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., W. L. Bland, J. R. Mecikalski, and M. C. Anderson, 2000: Satellite-based estimates of longwave radiation for agricultural applications. Agric. For. Meteor., 103, 349355, https://doi.org/10.1016/S0168-1923(00)00141-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diak, G. R., J. R. Mecikalski, M. C. Anderson, J. M. Norman, W. P. Kustas, R. D. Torn, and R. L. DeWolf, 2004: Estimating land surface energy budgets from space: Review and current efforts at the University of Wisconsin—Madison and USDA–ARS. Bull. Amer. Meteor. Soc., 85, 6578, https://doi.org/10.1175/BAMS-85-1-65.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Diamond, H. J., and Coauthors, 2013: U.S. Climate Reference Network after one decade of operations status and assessment. Bull. Amer. Meteor. Soc., 94, 485498, https://doi.org/10.1175/BAMS-D-12-00170.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gagne, D. J., II, A. McGovern, S. E. Haupt, and J. K. Williams, 2017: Evaluation of statistical learning configurations for gridded solar irradiance forecasting. Sol. Energy, 150, 383393, https://doi.org/10.1016/j.solener.2017.04.031.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gautier, C., and M. Landsfeld, 1997: Surface solar radiation flux and cloud radiative forcing for the Atmospheric Radiation Measurement (ARM) Southern Great Plains (SGP): A satellite, surface observations, and radiative transfer model study. J. Atmos. Sci., 54, 12891307, https://doi.org/10.1175/1520-0469(1997)054<1289:SSRFAC>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gautier, C., G. Diak, and S. Masse, 1980: A simple physical model to estimate incident solar radiation at the surface from GOES satellite data. J. Appl. Meteor. Climatol., 19, 10051012, https://doi.org/10.1175/1520-0450(1980)019<1005:ASPMTE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guenther, A. B., X. Jiang, C. L. Heald, T. Sakulyanontvittaya, T. Duhl, L. K. Emmons, and X. Wang, 2012: The Model of Emissions of Gases and Aerosols from Nature version 2.1 (MEGAN2.1): An extended and updated framework for modeling biogenic emissions. Geosci. Model Dev., 5, 14711492, https://doi.org/10.5194/gmd-5-1471-2012.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Habte, A., M. Sengupta, and S. Wilcox, 2012: Surface radiation from GOES: A physical approach. 27th European Photovoltaic Solar Energy Conf. and Exhibition, Frankfurt, Germany, WIP Renewable Energy, 3052–3059.

  • Habte, A., M. Sengupta, and S. Wilcox, 2013: Validation of GOES-derived surface radiation using NOAA’s physical retrieval method. National Renewable Energy Laboratory Tech. Rep. NREL/TP-5500-57442, 20 pp., https://www.nrel.gov/docs/fy13osti/57442.pdf.

    • Crossref
    • Export Citation
  • Haines, S. L., R. J. Suggs, and G. J. Jedlovec, 2004: The Geostationary Operational Environmental Satellite (GOES) product generation system. NASA Tech. Memo. NASA TM-2004-213286, 52 pp., https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20050019524.pdf.

  • Hall, D. K., and G. A. Riggs, 2016: MODIS/Terra snow cover daily L3 global 0.05deg CMG, version 6. NASA National Snow and Ice Data Center Distributed Active Archive Center, accessed 7 November 2019, https://doi.org/10.5067/MODIS/MOD10C1.006.

    • Crossref
    • Export Citation
  • Hong, Y., G. Liu, and J. L. F. Li, 2016: Assessing the radiative effects of global ice clouds based on CloudSat and CALIPSO measurements. J. Climate, 29, 76517674, https://doi.org/10.1175/JCLI-D-15-0799.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Huang, J., and M. Thatcher, 2017: Assessing the value of simulated regional weather variability in solar forecasting using numerical weather prediction. Sol. Energy, 144, 529539, https://doi.org/10.1016/j.solener.2017.01.058.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jacobs, J. M., D. A. Myers, M. C. Anderson, and G. R. Diak, 2002: GOES surface insolation to estimate wetlands evapotranspiration. J. Hydrol., 266, 5365, https://doi.org/10.1016/S0022-1694(02)00117-8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • James, E. P., S. G. Benjamin, and M. Marquis, 2017: A unified high-resolution wind and solar dataset from a rapidly updating numerical weather prediction model. Renewable Energy, 102, 390405, https://doi.org/10.1016/j.renene.2016.10.059.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jimenez, P. A., and Coauthors, 2016: WRF-Solar: Description and clear-sky assessment of an augmented NWP model for solar power prediction. Bull. Amer. Meteor. Soc., 97, 12491264, https://doi.org/10.1175/BAMS-D-14-00279.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Justice, C. O., and Coauthors, 1998: The Moderate Resolution Imaging Spectroradiometer (MODIS): Land remote sensing for global change research. IEEE Trans. Geosci. Remote Sens., 36, 12281249, https://doi.org/10.1109/36.701075.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lacis, A. A., and J. E. Hansen, 1974: A parameterization for absorption of solar radiation in the Earth’s atmosphere. J. Atmos. Sci., 31, 118133, https://doi.org/10.1175/1520-0469(1974)031<0118:APFTAO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Li, Z., C. H. Whitlock, and T. P. Charlock, 1995: Assessment of the global monthly mean surface insolation estimated from satellite measurements using global energy balance archive data. J. Climate, 8, 315328, https://doi.org/10.1175/1520-0442(1995)008<0315:AOTGMM>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mass, C., 2012: Nowcasting: The promise of new technologies of communication, modeling, and observations. Bull. Amer. Meteor. Soc., 93, 797809, https://doi.org/10.1175/BAMS-D-11-00153.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNider, R. T., J. R. Christy, D. Moss, K. Doty, C. Handyside, A. Limaye, A. G. Y. Garcia, and G. Hoogenboom, 2011: A real-time gridded crop model for assessing spatial drought stress on crops in the southeastern United States. J. Appl. Meteor. Climatol., 50, 14591475, https://doi.org/10.1175/2011JAMC2476.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNider, R. T., and Coauthors, 2015: An integrated crop and hydrologic modeling system to estimate hydrologic impacts of crop irrigation demands. Environ. Modell. Software, 72, 341355, https://doi.org/10.1016/j.envsoft.2014.10.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McNider, R. T., and Coauthors, 2018: Examination of the physical atmosphere in the Great Lakes region and its potential impact on air quality—Overwater stability and satellite assimilation. J. Appl. Meteor. Climatol., 57, 27892816, https://doi.org/10.1175/JAMC-D-17-0355.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mecikalski, J. R., D. M. Sumner, J. M. Jacobs, C. S. Pathak, S. J. Paech, and E. M. Douglas, 2011: Use of visible geostationary operational meteorological satellite imagery in mapping reference and potential evapotranspiration over Florida. Evapotranspiration, 446, 229254, https://doi.org/10.5772/14478.

    • Search Google Scholar
    • Export Citation
  • Mecikalski, J. R., W. B. Shoemaker, Q. Wu, M. A. Holmes, S. J. Paech, and D. M. Sumner, 2018: High-resolution GOES insolation–evapotranspiration data set for water resource management in Florida: 1995–2015. J. Irrig. Drain. Eng., 144, 04018025, https://doi.org/10.1061/(ASCE)IR.1943-4774.0001312.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, https://doi.org/10.1175/BAMS-87-3-343.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Myhre, G., and Coauthors, 2013: Anthropogenic and natural radiative forcing. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 659–740.

  • Ohmura, A., and Coauthors, 1998: Baseline Surface Radiation Network (BSRN/WCRP): New precision radiometry for climate research. Bull. Amer. Meteor. Soc., 79, 21152136, https://doi.org/10.1175/1520-0477(1998)079<2115:BSRNBW>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Otkin, J. A., M. C. Anderson, J. R. Mecikalski, and G. R. Diak, 2005: Validation of GOES-based insolation estimates using data from the U.S. Climate Reference Network. J. Hydrometeor., 6, 460475, https://doi.org/10.1175/JHM440.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Otte, T. L., 2008: The impact of nudging in the meteorological model for retrospective air quality simulations. Part II: Evaluating collocated meteorological and air quality observations. J. Appl. Meteor. Climatol., 47, 18681887, https://doi.org/10.1175/2007JAMC1791.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Paech, S. J., J. R. Mecikalski, D. M. Sumner, C. S. Pathak, Q. Wu, S. Islam, and T. Sangoyomi, 2009: A calibrated, high-resolution GOES satellite solar insolation product for a climatology of Florida evapotranspiration. J. Amer. Water Resour. Assoc., 45, 13281342, https://doi.org/10.1111/j.1752-1688.2009.00366.x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Paltridge, G. W., 1973: Direct measurement of water vapor absorption of solar radiation in the free atmosphere. J. Atmos. Sci., 30, 156160, https://doi.org/10.1175/1520-0469(1973)030<0156:DMOWVA>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., 2003: Surface radiation budgets in support of the GEWEX Continental-Scale International Project (GCIP) and the GEWEX Americas Prediction Project (GAPP), including the North American Land Data Assimilation System (NLDAS) project. J. Geophys. Res., 108, 8844, https://doi.org/10.1029/2002JD003301.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., R. Frouin, and Z. Li, 1995: A review of satellite methods to derive surface shortwave irradi