Methodology#

PV Atlas takes a computational modeling approach to studying the effect of climate on PV performance modeling and analysis. By combining high-resolution gridded weather datasets with detailed PV system performance models, we can simulate realistic PV system production datasets corresponding to a wide diversity of climates and system configurations. These simulated production datasets can then be examined to answer many questions about PV performance and analysis.

Running simulations at such a scale requires three key components: weather data spanning a diverse range of climates, customizable and scalable PV modeling software, and sufficient computing resources to perform the simulations and analyze the results. The latter is provided by a high-performance computing (HPC) cluster at Sandia National Laboratories.

Weather data#

Large-scale climatological studies are typically based on gridded satellite-derived or reanalysis weather datasets. Although generally somewhat less accurate than measurements from ground stations, the high spatial resolution and data availability of modern gridded datasets make them invaluable in climate-based analysis studies.

Several large-scale climatological datasets are available, including MERRA-2 and ERA5. These datasets have the advantage of global coverage and long data histories. However, PV Atlas uses the NSRDB PSM3 dataset due to its high spatial resolution (approximately 4/2 km) and its focus on accuracy for PV modeling applications.

The high-resolution of the NSRDB gridded weather and irradiance data requires a large amount of storage and computational capacity to process. To give a sense of scale, efficient HDF5 files with one year of 30-minute data covering the full spatial extent of the NSRDB are about 1.5 TB in size.

Software#

At the center of PV Atlas is a PV system performance modeling engine using pvlib-python, an open-source PV modeling toolbox written in and used via the Python programming language. pvlib-python’s unparalleled flexibility and scalability makes large simulation-based studies like PV Atlas possible.

The details of the PV performance models vary across analyses according to the needs of each simulation. For details, see the publications referenced in each chapter.

Outputs#

The geographic data visualized in the interactive maps are available for public download in the form of geo-referenced GeoTIFF files. GeoTIFF is a standard format for these kinds of rasterized geographic datasets. Making the datasets public allows users to import the analysis results into Python, GIS tools, or other computing environments for further processing.

For details, see Data Access.

References#

  1. Kevin S. Anderson, Clifford W. Hansen, William F. Holmgren, Adam R. Jensen, Mark A. Mikofski, and Anton Driesse. pvlib python: 2023 project update. Journal of Open Source Software, 8(92):5994, 2023. doi:10.21105/joss.05994.

  2. William F. Holmgren, Clifford W. Hansen, and Mark A. Mikofski. pvlib python: a python package for modeling solar energy systems. Journal of Open Source Software, 3(29):884, September 2018. doi:10.21105/joss.00884.

  3. Manajit Sengupta, Yu Xie, Anthony Lopez, Aron Habte, Galen Maclaurin, and James Shelby. The national solar radiation data base (NSRDB). Renewable and Sustainable Energy Reviews, 89:51–60, June 2018. doi:10.1016/j.rser.2018.03.003.