The outputs of the bioRad package will be a PPI (Plan Position Indicator) plot, showing the specified variables over Manhattan, and tabular data, which translate the NEXRAD grid cells into a user specified point density of 100 meter distances, with each point containing information about the radar variables. The main variable of interest is reflectivity, a measure of the echo intensity, or power transmitted back to the radar receiver. Reflectivity is a measure of density, and is measured on a logarithmic scale in DBZ, decibels relative to the reflectivity factor Z. A negative to under 20 range of DBZ means there is generally little density in the air, and around 20 would describe light rain (NOAA National Weather Service). The upper limit of reflectivity of birds is around 30 DBZ, so 35 DBZ will be chosen as a filter for the grid cells as a safe threshold (Stephanian, 2016).
Another variable of interest is the copolar correlation coefficient (ρHV), which measures the correlation between vertically (DBZV) and horizontally (DBZH) polarized reflectivity factors. The correlation coefficient will be one if there is no pulse to pulse variation in a sampled volume, such as objects that are homogenously dispersed, and will lower the more inhomogeneous the sample is. Because the shape and positioning of bird bodies is less homogenous than mist or rain drops, correlation coefficients of above 0.95 can be filtered out to remove meteorological observations from the study (Stephanian, 2016).
The last variable that will be examined is radial velocity, measured in meters per second. Velocity is useful for removing ground clutter, such as trees and buildings, that could be affecting results in reflectivity. As ground clutter is immobile, cells that have radial velocities within a range of -1 to 1 will be removed, although this could remove birds flying perpendicular to the radar as well (Hubbert, 2009). Additionally, velocity has been used to distinguish birds from insects, as insects generally fly at speeds less than 8 to 10 meters per second, while migrating birds typically fly at speeds over 10 meters per second. However, due to the potential for overlap, this filter for insects will not be utilized (Gauthereaux, 1998).
Bird Count Estimates
To estimate bird density, a similar methodology to Gauthreaux et. al. will be employed. Gauthreaux et. al. related maximum reflectivity values from 0.5° elevation scan angles to the number of birds crossing a 1.6 km line in an hour, which were observed by moon watching. This study resulted in a fit line that was a third order polynomial, and was found to be highly significant with an R-squared value of 0.87. Gauthereaux then went on to adjust the fit line, first converting migration traffic rates to density (birds per km3), and then converting relative decibels of reflectivity to just reflectivity (dBZ to Z). The result is a linear fit line, and when forced through zero, again generated an R-squared value of 0.87 (Gauthreaux et. al., 2008).