Typical Meteorological Year (TMY) data files were first created from long-term data files in the NSRDB database for analysis of building performance. At the time, computers were much slower and had less memory than today’s computers. Users need a data set with a recording period of 1 year, and the simulation results can be derived using 30 years of available data in the NSRDB database. Many meteorological data parameters have a greater impact on building performance than anthropogenic solar radiation, and the TMY dataset was created as typical meteorological data in the NSRDB database.
Each TMY data file is composed of the most representative 12 months of data selected from the database year. Sandia National Laboratories selects a typical month to create the original file based on 9 daily indicators. The 9 daily indices include maximum dry bulb temperature and dew point temperature, minimum dry bulb temperature and dew point temperature, average dry bulb temperature and dew point temperature, maximum wind speed, average wind speed, and total horizontal irradiance (see Table 1 for details). The final selection of months takes into account the monthly mean and median of the nine indices (as shown in Table 1), and the persistence of weather patterns (Marion and Urban, June 1995). The selected 12 months form a representative TMY file. The file needs to be revised at the beginning and end of each month to eliminate the transitional data caused by the selection of adjacent months from different years.
The raw TMY data files were created from measured GHISOLMET data and ERSATZ model data from 1952-1975. The TMY2 data files were created from the NSRDB database from 1961-1990, with 93% of the data values already modeled. TMY3 data files are created from the NSRDB database from 1991-2005 and the NSRDB database from 1961-1990 (if data at this location is included). In the TMY2 data file, the DNI was added to the weighted index, which increased the comparison of the annual average DNI in the file with the long-term average DNI in the NSRDB database file by a factor of approximately 2. The wind speed weighting was reduced, and the persistence criteria in the TMY2 and TMY3 data files also changed slightly (Wilcox and Marion May 2008). Table 1 illustrates the weighted differences used by the Sandia National Laboratory (TMY) method and the National Renewable Energy Laboratory (TMY2 and TMY3) methods. Note that for the TMY2 and TMY3 datasets, half are weighted to solar irradiance values and the other half are meteorological variables. In the raw TMY data files, the estimated uncertainties of the monthly mean daily total GH1 and DNI values obtained from the measured SOLMET data were ±7.5% and ±10%, respectively. Likewise, the estimated uncertainties for the monthly mean daily total GHI and DNI obtained from the ERSATZ model data are ±10% and ±20%, respectively (SOLMET in 1978). In the TMY2 file, the months between May 1982 and December 1984 were not included in the analysis due to significant differences between the aerosols from the eruption of El Chichon in Mexico and typical values object. For the TMY3 file, the months between June 1991 and December 1994 were not included due to the effects of the eruption of Mount Pinatubo in the Philippines. Therefore, 83% of the data in the TMY3 file is derived from 11.5 years of data.
- Limitations of TMY2 and TMY3 files
TMY files are created to reflect typical meteorological years and atypical solar years. Due to the limited number of years covered in most TMY3 data files, there is no guarantee that the files will accurately reflect average GHI or DNI measurements for the entire historical dataset. For example, Figures 2 and 3 show that the annual mean values of TMY for GHI and DNI are different from the mean values of the NSRDB database; in Groton-New London, Connecticut, USA, the GHI value of the annual mean TMY is lower than the mean value of each year in the NSRDB database. Average annual GHI value. In Paso Robles, California, the opposite is true. The GHI every 12 months is lower than the annual average GHI of TMY3.
Figures 2 and 3 also show the DNIs of Groton-New London and Paso Robles. Both examples illustrate that even when the 50% indicator weighting is GHI and DNI, there is no guarantee that the annual mean irradiance value obtained in the TMY file is close to the true long-term mean solar irradiance. Such extreme cases are rare, but if TMY files are used to estimate the efficacy of solar systems, long-term data sets must be compared with TMY averages, especially with TMY3 data files created from only 11 years of data.
The TMY dataset deliberately excludes extreme cases. As a result, little data is available when trying to understand resource variability. For meteorological variables, it takes about 30 years of data to fully characterize the solar irradiance at a location. All short-term weather changes, such as El Niño and La Niña, are included in the 30-year data, and even short-term weather changes due to 11- or 22-year sunspot cycles. Short-term meteorological events lasting several years must affect observations. Whereas for short-term datasets (eg, 15 years), weather cycles (eg, El Niño) may distort the statistical characteristics of shorter datasets, and the percentage of total records affected by these infrequent events increases (Vignola and McDaniels, 1993 April).
Read more: What is a solar radiation dataset?
