- Original article
- Open Access
Assessment of local solar resource measurement and predictions in south Louisiana
© The Author(s). 2016
- Received: 1 October 2015
- Accepted: 13 June 2016
- Published: 21 July 2016
The accessibility of reliable local solar resource data plays a critical role in the evaluation and development of any concentrating solar power (CSP) or photovoltaic (PV) project, impacting the areas of site selection, predicted output, and operational strategy. Currently available datasets for prediction of the local solar resource in south Louisiana rely exclusively on modeled data by various schemes. There is a significant need, therefore, to produce and report ground measured data to verify the various models under the specific and unique ambient conditions offered by the climate presented in south Louisiana.
The University of Louisiana at Lafayette has been recording onsite high-fidelity solar resource measurements for the implementation into predictive models and for comparison with existing datasets and modeling resources. Industry standard instrumentation has been recording direct normal irradiance (DNI), diffuse horizontal irradiance (DHI), and global horizontal irradiance (GHI), as well as meteorological weather data since 2013. The measured data was then compared statistically to several available solar resource datasets for the geographic area under consideration.
Two years of high-fidelity solar resource measurements for a location in south Louisiana that were previously not available are presented. Collected data showed statistically good agreement with several existing datasets including those available from the National Solar Radiation Database (NSRDB). High variability in year-over-year monthly DNI due to cloud cover was prevalent, while a more consistent GHI level was observed.
The analysis showed that the datasets presented can be utilized for predictive analysis on a monthly or yearly basis with good statistical correlation. High variability in year-over-year monthly DNI due to cloud cover was prevalent, with as much as a 70 % difference in monthly DNI values observed in the measured data. A more consistent GHI level was observed since the GHI is less susceptible to cloud cover transients. Collected data showed statistically good agreement with several existing datasets including those available from the NSRDB when forecasting was for monthly and yearly intervals.
- National Renewable Energy Laboratory
- Mean Bias Error
- Concentrate Solar Power
- Parabolic Trough
- Surface Radiation Budget
As an integrated part of the University of Louisiana at Lafayette (UL Lafayette) Solar Technology Applied Research and Testing (START) Lab, local solar resource measurements have been conducted onsite since the summer of 2013. The compilation of short- and long-term solar resource data has been conducted for performance evaluation of the onsite solar energy technologies as well as the generation and validation of reliable solar resource models [1, 2]. The primary solar energy asset during the period of investigation presented here has been a pilot scale (650 kWth) parabolic trough solar thermal power plant, constructed and operated by UL Lafayette, for which knowledge of the solar resource was critical. The results of the solar resource study relative to the operation of this power plant will therefore be presented. In general, there are several important roles that accurate solar resource evaluation and forecasting plays in solar power applications. In the project planning stages, understanding the quality and quantity of the resource is essential to accurately predict system performance and financial viability of any future project and can be broken down into three areas of study . Site selection, predicted annual plant output, and short-term temporal performance and operating strategy will all be grossly affected by the local short- and long-term resource availability and fluctuation . Additionally, accurate measurement and dissemination of resource data to determine short- and long-term plant performance is vital to optimize performance once operation is underway. Reliable resource measurement will therefore remain vital to the plant’s efficient operation throughout its service life.
From the initial planning stages, site selection will typically be based on datasets of historical solar resource data involving changes in weather affecting ground-level insolation from year-to-year, and therefore, more years of data are advantageous for constructing a representative annual dataset . Reliable long-term (25 year) historical datasets are rarely available. Typical meteorological year (TMY) datasets, available from several sources, including the National Renewable Energy Laboratory, can be efficiently used to compare the solar resource at alternative sites and to predict a range of annual performance values of a proposed solar energy plant. Data from individual years are useful in illuminating year-to-year variability that can be expected for the specific locale which assists in the sizing of components of solar systems accurately [6, 7].
where zenith is the topocentric solar zenith angle measured in degrees. The DNI is of particular interest to concentrating solar power (CSP) projects since the DNI is the component of the solar radiation which can be concentrated and therefore of particular interest to this study for the operational performance analysis of the solar thermal power plant at UL Lafayette. Likewise, the GHI is of particular interest to photovoltaic (PV) projects as most PV installations are non-concentrating and are not dependent on direct beam radiation but of the total hemispherical radiation intercepted on the plane of the array.
The most recent NSRDB data, shown in the NSRDB data viewer utilizes the latest version of the SUNY model. This data provides monthly average and annual average daily total solar resource averaged over surface areas of 4 km in size. The data are generated using the PATMOS-X algorithms for cloud identification and properties, the MMAC radiative transfer model for clear sky calculations and the SASRAB model for cloud sky calculations . The data are averaged from hourly model output over 8 years (2005–2012) with each year downloadable from the website for any user identified location. For all of the models, it should be noted that physics-based solar radiation models grounded in measurements can be no more accurate than the data used to generate the model and cannot be validated or verified to a level of accuracy greater than that of the measurements .
Weather station setup
Statistical metrics for comparison of various modeled DNI data with measured data
TMY3 hourly (W/m2)
SA hourly (W/m2)
TMY3 daily (kWh/m2/day)
SA daily (kWh/m2/day)
TMY3 monthly (kWh/m2/day)
SA monthly (kWh/m2/day)
NASA monthly (kWh/m2/day)
Statistical metrics for comparison of various modeled GHI data with measured data
TMY3 hourly (W/m2)
SA hourly (W/m2)
TMY3 daily (kWh/m2/day)
SA daily (kWh/m2/day)
TMY3 monthly (kWh/m2/day)
SA monthly (kWh/m2/day)
Maximum quantities of irradiance and insolation for different cases
For reference, Djebbar et al.  reported DNI average hourly MBE and RMSE of 28.5 and 133.7 W/m2 (67.2 %) for SUNY V3 beta models from an averaged result of three sites across Canada. Here, the percentage data can be taken as equivalent to the NRMSE. Also reported were the daily results of 8.4 MBE and 30.4 RMSE (52.1 %) and monthly results of 8.5 MBE and 15.1 RMSE (25.8 %) (all in kWh/m2/day). Additionally, values were computed for GHI average hourly MBE and RMSE of 5.6 and 86.5 W/m2 (27.8 %) for SUNY V3 beta models from an averaged result of 18 sites across Canada. Also reported were the daily results of 0.1 MBE and 0.5 RMSE (15 %) and the monthly results of 0.1 MBE and 0.2 RSME (6.7 %) (all in kWh/m2/day). Virtually, all the results from the Louisiana location models compare favorably to the results obtained by Djebbar for the Canadian location.
Impact to CSP operation
Impact to PV operation
The current primary generating method for solar energy in Louisiana is through PV conversion. This generation is through residential and commercial rooftop installations, as there are currently no industrial scale installations of solar power plants of any kind in Louisiana. The implications of the solar resource data with regards to PV residential and commercial installations lie with the GHI metric. As most PV installations are non-concentrating, the GHI component of the solar resource is the primary source of comparative data. The GHI values for south Louisiana vary less from the higher solar resource areas of the Southwestern United States than the DNI values, due to the fact that the GHI is less sensitive to transient cloud cover. For example, based on NSRDB data, the ratio of Louisiana DNI to that of the most solar-rich areas of the country is about 0.6, while the ratio of GHI for the same areas is about 0.75. For this same reason, there should be less variance from year-to-year for a given month in a given location. The results of the comparison of modeled to measured values have limited efficacy due to the limited data acquired thus far, yet it can be clearly seen that the GHI modeled data provides a better forecast than the DNI, when comparing the NRMSE, MAE, and the K-S test. There was a positive MBE, indicating a slight over-prediction of solar resource by the models; however, the bias was small.
PV does not require a minimum radiation level for start-up; therefore, the power produced is primarily a function of the system efficiency, the insolation and panel orientation. Actual PV production data was not available at the time of this study; however, it will be presented in future work.
This paper has provided the results of 2 years of high-fidelity solar resource measurements for a location in south Louisiana that was previously not available. The data was compared to modeled data from various sources. The measured data showed that high peak irradiance values were available for single days (10 kWh/m2/day) and monthly averages (6 kWh/m2/day), higher than the maximums produced by models available for the same location. In addition, high variability in year-over-year monthly DNI due to cloud cover was prevalent, with as much as a 70 % difference in monthly DNI values observed in the measured data. When comparing the measured data to the modeled data, an over 100 % difference could be seen in the expected DNI. A more consistent global horizontal irradiance (GHI) level was observed since the GHI is less susceptible to cloud cover transients. Collected data showed statistically good agreement with several existing datasets including those available from the National Solar Radiation Database (NSRDB) when forecasting was for monthly and yearly intervals.
The measurements presented here were made in Crowley, LA, and therefore, the radiation measurements are relevant to that location only. The size of the area to which the measurements are relevant is limited by the variance in local weather conditions and will fluctuate with weather patterns, which would dictate a need for a unique station for each commercial or industrial scaled deployment of solar power in order to capture real-time performance. Over longer time scales (months and years), larger areas between stations can be allowed while maintaining fidelity and definition of measurement.
Additional seasonal comparisons can be made to determine the stability of long-term seasonal predictions which can be important in operational strategies. Furthermore, due to the inherent variability in the year-to-year data, the question of how many years it will take before the solar radiation components stabilize and converge to their long-term value is an appropriate one. This question was addressed by Gueymard and Wilcox  among others, and as data is added to the START lab database, an evaluation of this metric can be made accurately. It was found that for the Eugene, Oregon, area the DNI anomaly levels only converged within ±10 % after 5 years of data and approached ±5 % after 10 years of data. The GHI anomaly was within 5 % after 2 years of data and was within 2 % after 15 years of data availability. Also addressed by Gueymard and Wilcox was the magnitude of the inter-annual variation, which was found to fluctuate from region to region. Additionally, a relationship between the monthly averaged expected DNI values and the clear sky days will be investigated along with other cloud cover data to gain a better understanding of the local weather influence on anticipated solar resource. Finally, long-term data on actual solar thermal collector and PV array output will be aggregated and correlated with predicted values for the further validation and refinement of operational models.
This work was made possible by the funding and support from Cleco Power LLC and the University of Louisiana at Lafayette. These NASA SRB dataset were obtained from the NASA Langley Research Center Atmospheric Science Data Center Surface meteorological and Solar Energy (SSE) web portal supported by the NASA LaRC POWER Project.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Raush J, Chambers T (2013) Demonstration of pilot scale large aperture parabolic trough organic Rankine cycle solar thermal power plant in Louisiana. J Power Energy Eng 1:29–39View ArticleGoogle Scholar
- Raush JR (2014) Chambers TL. Initial field testing of concentrating solar photovoltaic (CSPV) thermal hybrid solar energy generator utilizing large aperture parabolic trough and spectrum selective mirrors 3:123–131. doi:10.11648/j.ijrse.20140306.12 Google Scholar
- Stoffel T, Renné D, Myers D, Wilcox S (2010) Concentrating solar power best practices handbook for the collection and use of solar resource data. Tech Rep. doi: NREL/TP-550-47465.Google Scholar
- Chambers T, Raush J, Massiha G (2013) Pilot solar thermal power plant station in southwest Louisiana. Int J Appl Power Eng 2;(1):45–52. Google Scholar
- Burnett D, Barbour E, Harrison GP (2014) The UK solar energy resource and the impact of climate change. Renew Energy 71:333–343. doi:10.1016/j.renene.2014.05.034 View ArticleGoogle Scholar
- Wilcox S, Marion W (2008) Users manual for TMY3 data sets., doi: NREL/TP-581-43156View ArticleGoogle Scholar
- Wilcox S (2012) National solar radiation database 1991–2010 update : user’s manual., doi: NREL/TP-5500-54824Google Scholar
- Goswami DY, Kreith F, Kreider JF (2000) Principles of solar engineering, 2nd edn. Taylor & Francis Group, New YorkGoogle Scholar
- National renewable energy laboratory concentrating solar power projects in the United States. http://www.nrel.gov/csp/solarpaces/by_country_detail.cfm/country = US %28%22_self%22 %29.
- U.S. Energy Information Agency State Electricity Profiles 2010. http://www.eia.gov/electricity/state/pdf/sep2010.pdf. Accessed 1 Jan 2013
- Nonnenmacher L, Kaur A, Coimbra CFM (2014) Verification of the SUNY direct normal irradiance model with ground measurements. Sol Energy 99:246–258. doi:10.1016/j.solener.2013.11.010 View ArticleGoogle Scholar
- Perez R, Ineichen P, Moore K et al (2002) A new operational model for satellite-derived irradiances: description and validation. Sol Energy 73:307–317. doi:10.1016/S0038-092X(02)00122-6 View ArticleGoogle Scholar
- Whitlock CH, Charlock TP, Staylor WF, Pinker RT, Laszlo I, Ohmura A, Gilgen HKT, DiPasquale RC, Moats CD, LeCroy SR, Ritchey NA (1995) First global WRCP shortwave surface radiation budget dataset. Bull Am Meteorol Soc 76:905–922View ArticleGoogle Scholar
- Sengupta M, Habte A, Gotseff P et al (2014) A physics-based GOES satellite product for use in NREL’ s National Solar Radiation Database preprintGoogle Scholar
- Gueymard C (2009) Direct and indirect uncertainties in the prediction of tilted irradiance for solar engineering applications. Sol Energy 83:432–444. doi:10.1016/j.solener.2008.11.004 View ArticleGoogle Scholar
- Kipp & Zonen (2008) CHP-1 Pyrheliometer instruction manual.Google Scholar
- Kipp & Zonen (2013) CMP series pyranometer instruction manualGoogle Scholar
- NREL: Measurement and Instrumentation Data Center (MIDC) Home Page. https://www.nrel.gov/midc/. Accessed 6 May 2016
- Gueymard C, Wilcox SM (2011) Assessment of spatial and temporal variability in the US solar resource from radiometric measurements and predictions from models using ground-based or satellite data. Sol Energy 85:1068–1084. doi:10.1016/j.solener.2011.02.030 View ArticleGoogle Scholar
- Zhang J, Hodge B-M, Florita., et al (2013) Metrics for evaluating the accuracy of solar power forecasting. 3rd Int Work Integr Sol Power into Power Syst 1–10.Google Scholar
- Stoffel T (2013) A review of measured/modeled solar resource uncertainty. Presented at the 2013 Sandia PV Performance Modeling Workshop Getting to PV Performance Model Input Uncertainty Measurements.Google Scholar
- Djebbar R, Morris R, Thevenard D et al (2012) Assessment of SUNY version 3 global horizontal and direct normal solar irradiance in Canada. Energy Procedia 30:1274–1283. doi:10.1016/j.egypro.2012.11.140 View ArticleGoogle Scholar
- Reda I, Andreas A (2004) Solar position algorithm for solar radiation applications. Sol Energy 76:577–589. doi:10.1016/j.solener.2003.12.003 View ArticleGoogle Scholar
- Marion WF, Dobos AP, Marion WF, Dobos AP (2013) Rotation angle for the optimum tracking of one-axis trackers rotation angle for the optimum tracking of one-axis trackersView ArticleGoogle Scholar