The solar resource and meteorological data in a SAM weather file may have been developed from ground measurements, data from a satellite, or a combination of the two. A computer model is usually involved in preparing the weather file, and may be used to fill gaps in the data, determine typical-year months, or calculate values.
|NSRDB 1998 - 2014 (Current)||North America, Central America, Caribbean, northern South America, India, Bangladesh, Sri Lanka and parts of Pakistan, Afghanistan, China, Myanmar and surrounding countries.||Single-year and typical-year hourly data (30-minute solar irradiance data available). 4 km grid cells. SAM CSV format. Free.|
|NSRDB 1991 - 2010 Update (TMY3)||United States||Typical-year hourly data (TMY). 1,454 locations. SAM-ready TMY3 format. Free.|
|NSRDB 1961 - 1990 (TMY2)||United States||Typical-year hourly data (TMY). 237 locations. SAM-ready TMY2 format. Included with SAM. Free.|
|NSRDB 1961 - 1990 Historical||United States||Single-year hourly data. 237 locations. SAM-ready TMY3 format. Free|
|NREL Solar Prospector||Continental United States||Single-year and typical-year hourly data (TMY, TGY, TDY). Satellite-derived data. Approx. 10-km grid cells. SAM-ready TMY2 format. Free.|
|AUSTELA||Australia||Single-year hourly data (select years between 1996 and 2011). Twelve locations. SAM-ready TMY3 format. Free.|
|U.S. DOE EnergyPlus||Global||Typical-year hourly data from various sources. Over 2,100 locations. SAM-ready EPW format. Free.|
|Clean Power Research||North America||Single-year data (1998 - ). Satellite-derived data. 10 km grid cells. SAM-ready TMY3 format. $$ (some free data).|
|SolarGIS||Global||Single-year and typical-year data (1994 - ). Satellite-derived data. 250 m grid cells. $$.|
|Meteonorm||Global||Single-year and typical-year data. Model-generated data from various sources. Various formats. $$.|
|White Box Technologies||Global||Single-year and typical-year data. Over 4,000 locations. Data from ASHRAE IWEC2, NSRDB, CWEC2, and CZ2010. $$.|
|Deutscher Wetterdienst WebWerdis||Germany and other countries||Single-year data. Free and $$.|
Note that historical single-year solar data only from NSRDB 1991 - 2005 Update (TMY3) is available from the NSRDB website. These files contain comma-separated files (.csv) with the specific-year historical solar data that NREL used to develop the NSRDB 1991 - 2005 Update typical-year files. These files do not contain meteorological data, and are not in the TMY3 file format, so they cannot be used in SAM without adding meteorological data and converting the data into a file format that SAM can read.
SAM can read weather files in the following text formats.
A SAM weather file is a text file that contains one year's worth of data describing the solar resource, wind speed, temperature, and other weather characteristics at a particular location. SAM's performance models can run hourly or subhourly simulations. SAM determines the simulation time step from the weather file's temporal resolution. For example, SAM interprets weather file with 8,760 rows to be for hourly simulations. Most available weather files contain hourly data, assuming that there are 8,760 in one year. A weather file may contain typical-year data that represents long-term historical data or single-year data for a particular year.
The information here supplements the descriptions and instructions in SAM's Help system. The Help system is available in SAM from the software's Help menu, or by pressing the F1 key in Windows or Command-? in OS X.
When you run a simulation in SAM, you must choose a weather file for your assumption of the renewable energy resource. SAM comes with a set of weather files that you can use to get started with your analysis:
When you install SAM, it creates folders named solar_resource and wind_resource containing weather files for the solar and wind performance models.
Solar Resource Data 101 Webinar
For SAM's solar performance models (detailed photovoltaic, PVWatts, concentrating solar power CSP, solar water heating), you choose the weather file on the Location and Resource input page. These models use data from the weather file as follows:
The files in SAM's solar resource library are from the following sources:
SAM's wind performance model uses wind speed and temperature data at three heights above the ground, along with wind direction and atmospheric pressure data. It also uses the site elevation data from the weather file.
The biomass power model uses data from several databases of feedstock data for the United States. It also uses meteorological data from a solar resource file to model thermal performance of a steam power cycle. You can use a weather file from any of the sources listed above for the Solar models.
The geothermal model is unique among the SAM performance model in that it performs a simulation over the entire plant life rather than using a single-year simulation to represent the system's performance over a multi-year period. The heat resource for a geothermal system varies over periods of months as the resource degrades, unlike solar and wind systems whose resource can varies diurnally. The geothermal model accesses a database of temperature and depth data for the geothermal resource (United States locations only), and uses meteorological data from a weather file to model thermal performance of a steam power cycle that can be from any of the data sources listed above for the solar models.
TMY is an acronym for typical meteorological year. A typical-year file contains one year of hourly data that represents historical weather data over a multi-year period. It is typical because it uses one year's worth of typical data to represent the original data over the entire period. There are different methods for determining the typical values (see below for details). You can think of typical-year data as the average of the original data over the historical period, but it is more accurate to say that the data is typical because the methods involve more than just calculating average values. A TMY file is meteorological because the method for developing the files involves analysis of solar resource data and weather data (wind speed, ambient temperature, etc.). Some data sources provide different kinds of typical-year data files, such as typical GHI year (TGY) and typical DNI year (TDY) files, which were developed by analyzing the solar resource data but not the weather data -- these are also typical-year data files, but use a different definition of "typical".
The term "TMY" can be confusing because sometimes people use it incorrectly to describe a file format or data source. Here are some examples of how to describe weather files to make the distinction clear:
The weather file you choose for an analysis in SAM is your assumption of the renewable resource available at the system's location. You also make assumptions about the physical characteristics of its design, project financial structure, and installation and operating costs. Each assumption has a degree of uncertainty that contributes to the uncertainty of the results of your analysis.
When you use SAM to predict the electrical output of a renewable energy system, uncertainty in the weather data is one of the main sources of uncertainty in your SAM results. When you run SAM with one of the NSRDB weather files or weather files from one of the other sources mentioned here, you are trying to predict the future using historical data. That's inherently uncertain, so you should be prepared to present SAM results as ranges of probable values rather than as a specific values.
Here are some general tips to help you do so:
For a good discussion of weather files and uncertainty in energy modeling, see Stoffel et. al. 2012 ( PDF 7.5 MB). Although this handbook was written for an audience of concentrating solar power analysts, the information is also useful for anyone using SAM to model photovoltaic systems, and to some degree, wind systems.
When you model a renewable energy system in SAM, it is usually for a renewable energy system that will generate electricity over a period of many years. You may be using SAM to estimate how much electricity a parabolic trough system will generate hourly, monthly, and annually, on average, over a 30-year period. Or, you may be trying to determine what electricity power price a wind farm or utility-scale PV project needs to meet a desired internal rate of return over a 25-year project life. Or, you could be trying to estimate how much a customer will save on her monthly electric bill over the next ten years if you install a roof-top PV system on her home. To save simulation time and complexity, SAM only simulates the system's performance over a single year. So, for these kinds of analyses, you need to use a weather file that represents the resource over many years.
A typical year file uses a single year of hourly data to represent the renewable resource and weather conditions over a multi-year period. The typical year methodology involves analyzing a multi-year data set and choosing a set of 12 months from the multi-year period that best represent typical conditions over the long term period. For example, a typical year file developed from a set of data for the years 1998-2005, might use data from 2000 for January, 2003 for February, 1999 for March etc. Annual simulation results from typical year weather data are suitable for long-term economic analysis.
Single year data represents the weather at a location for a specific year. Single year data is appropriate for analysis of a system's performance in a particular year when you are not using the results to predict economic value over many years. Single-year data may also be appropriate for analyses involving time-dependent electricity pricing or electric loads for a given year.
You can also use a set of single-year weather files for many years to calculate a system's hourly output over a long term period. This is the approach SAM uses for P50/P90 analysis.
A P50/P90, or probability exceedance analysis, involves running a set of single-year simulations to calculate annual output values, and then from those values determining the annual output value that was exceeded 50% of the time (the P50 value) and the value that was exceeded 90% of the time (P90 value). A P50/P90 analysis in SAM requires single-year data for at least 15 years, but ideally more. For example, a P50 value of 30 MWh would mean that based on the historical data, there is a 50% likelihood that the system's annual output will exceed 30 MWh. Similarly, a P90 value of 30 MWh, would mean that there is a 90% likelihood that the output will exceed 30 MWh.
SAM's P50/P90 feature requires a set of single-year weather files that you put in a folder on your computer. You can download NSRDB Historical Single Year Data for the 237 locations in the TMY2 data collection.
For a description of SAM's P50/P90 methodology, see Dobos et. al. 2012 (PDF 423 KB).
For a system with an address in Golden, Colorado on the NREL campus, you could establish a range of reasonable values for its annual output using three typical-year data files: 1) Boulder, the nearest location in the NSRDB 1961 - 1991 collection (TMY2), 2) Denver/Centenial Boulder, the nearest location in the NSRDB 1991 - 2010 update (TMY3), and 3) the Solar Prospector data for the 10 km square that contains the address. The map in Figure 1 shows the system location and the three data sites. The Boulder TMY2 site is about 30 km (19 miles) north of the system location. The TMY3 site is within a few kilometers of the system location, and the system location is inside the borders of the Solar Prospector 10 km square. (The 10 km square is approximate, and looks like a rectangle because of the map projection.) Table 1 shows the annual output that SAM estimates for a 4 kW system for each dataset (using PVWatts to model a fixed array tilted at latitude with a 0.77 DC-to-AC derate factor). These values provide a basis for estimating a reasonable range of annual output values for the system. For this example, you could say that based on historical weather data and the system's physical description and model assumptions, it would be reasonable to expect the system's output to be between 5.5 MWh/year and 5.9 MWh/year.
Figure 1. Map showing three solar resource data sites available for a system location on the NREL campus in Golden, Colorado
|Weather File||Annual Ouput (kWh)|
|Boulder, CO TMY2||5,835|
Ground-measured data is data taken from devices that measure solar radiation, temperature, wind speed, humidity, and atmospheric pressure. These devices take measurements at a specific location and store data at some time resolution, typically every minute, ten minutes, or fifteen minutes. Preparing this data for use in a weather data file involves collecting the data, checking it for data quality issues, filling data gaps, and calculating hourly values from the higher temporal resolution data. Ground measured data provides accurate historical information, but is limited to the location of the measurement devices.
Satellite-derived data is data generated by a computer model that processes data from digital images of the earth's surface taken from a satellite. Satellite-derived data can provide historical information about large geographic areas.
SAM's SAM CSV (solar) and SRW (wind) weather file formats are basic comma-separated formats that make it relatively easy to create weather files with your own data. SAM's Weather File Checker macro helps you identify formatting problems with files in the SAM CSV format. The Weather File Converter macro converts files in the TMY3, TMY2, and EPW formats into the SAM CSV format.In some cases, you may have weather data from a resource measurement program, meteorological stations, or some other source. SAM can read a weather file with data from any source, as long as it is in one of the formats described here without any formatting errors, gaps in the data, or invalid values. You can create the file in a spreadsheet or database program, or write your own software to create the file.
Dobos, A.; Gilman, P.; Kasberg, M. (2012). "P50/P90 Analysis for Solar Energy Systems Using the System Advisor Model." NREL Conference Paper Preprint No. CP-6A20-54488. (PDF 432 KB)
Habte, A.; Lopez, A.; Sengupta, M.; Wilcox, S. (2014). "Temporal and Spatial Comparison of Gridded TMY, TDY, and TGY Data Sets." NREL Report No. TP-5D00-60866. (PDF 17.4 MB)
National Solar Radiation Database. (1992). "National Solar Radiation Data Base User's Manual (1961-1990)." [TMY2 format description.]
Sengupta, M.; Habte, A.; Kurtz, S.; Dobos, A.; Wilbert, S.; Lorenz, E.; Stoffel, T.; Renne, D.; Myers, D.; Wilcox, S.; Blanc, P.; Perez, R. (2015). "Best Practices Handbook for the Collection and Use of Solar Resource Data for Solar Energy Applications." NREL Report No. TP-5D00-63112. (PDF 8.9 MB)
Wilcox, S.; Marion, W. (2008). "Users Manual for TMY3 Data Sets (Revised)." 58 pp.; NREL Report No. TP-581-43156. ( PDF 1.7 MB)