Is it possible to use pyranometer data to find locations in the United States where there are significant changes in average cloudiness during the past 10 years?
The National Oceanic and Atmospheric Administration (NOAA) has developed the Climate Reference Network (CRN), a network of meteorology sites, starting with a few sites in 2002, that will form the basis for monitoring long-term changes in climate within the United States. These sites are carefully monitored and are in environmentally stable natural areas. The primary measurement is air temperature, but there is also instrumentation for measuring other parameters, including insolation. The ratio of measured daily integrated insolation to a theoretical clear sky value depends primarily on cloud cover at that site; the lower the ratio, the higher the cloudiness over the day. If this ratio changes significantly over time, an obvious interpretation is that average cloud conditions at the site have changed. Typically, this ratio will also have a seasonal component that may make long-term changes in cloudiness more difficult to detect. Other factors, such as an average change in air quality, can also affect this ratio.
Here is a graph of the observed/clear sky ratio, plus a 30-day moving average of the ratio, for the CRN site at Avondale, Pennsylvania, from when this site started reporting data in mid-2006 through the end of 2010. Although it is possible for the integrated daily insolation to exceed a clear-sky value for certain sky conditions, a much more likely explanation of the seasonal trend in the observed/clear sky ratio, which often exceeds a value of 1 especially during the winter, is that the clear sky model is not an adequate representation of insolation at this site. It should be possible to fix this problem by imposing some physically reasonable seasonal variation on the model parameters.
Assuming that a better clear sky representation is used, there may be other interesting trends in these data. One could construct monthly count histograms for the observed/clear sky ratio — these histograms may change seasonally or over time. The number of cloud-free days, where the observed/clear sky ratio is close to 1, should be of particular interest. Or, perhaps the histograms could be based on the observed/clear sky ratio just around noon, for the local hours 11:00-12:00 and 12:00-13:00. (Note: Looking for trends in the number of cloud-free days, for the entire day or just around noon, has been proposed by Forrest Mims.)
All the CRN data are available online along with README files that explain the file format. This is an interesting application of pyranometer data that requires some computer skills and a basic knowledge of statistics. The For more information about this project, contact David Brooks.