RPI Analysis

The initial RPI analysis effort examined hourly raptor migration count data from seven hawk watch sites in Canada and the United States. Select a site to view a summary of the initial RPI analysis for that site:

The second stage of RPI analysis examines hourly raptor migration count data from the following ten sites:

Long-term series of data from additional selected sites are currently under analysis. The models used to generate annual indices and to calculate long term trends have been described by Farmer et al. (Auk, 2007) In summary, the procedure described by Chris Farmer (North American Monitoring Coordinator, HMSA) standardizes the count day/season at each monitoring site and also allows compensation for missing days and additional covariates (e.g. weather.)

For each species, the model:

  • (1) Identifies the daily window during which 95% of the migrants are counted;
  • (2) Selects the seasonal passage window when the middle 95% of the individuals were counted across all years;
  • (3) Calculates an index of passage rate [date-adjusted estimated geometric means];
  • (4) Obtains trends in annual passage rates by fitting a polynomial regression model to the time series of annual indices; and
  • (5) Re-parameterizes the year terms to determine the significance of these trend estimates (fide Francis and Hussell 1998; Farmer et al. 2007)

RPI has transferred a series of robust data sets to electronic form for analysis. The completed analyses include the initial seven sites from the Northeast/Great Lakes Region, twelve sites from the Intermountain West Region, four sites from the Gulf of Mexico, and seven spring watchsites.