The effect of data resolution on the area threshold method is studied
using a stochastic model of radar snapshots of tropical rainfall taken
during the Global Atmospheric Research Program Atlantic Tropical Expe
riment. The results indicate that significant biases and random error
can arise when using radar or satellite data having spatial resolution
s different from those used to calibrate the area threshold method. Al
though the use of a threshold decreases the bias and random error in t
he fractional coverage of rain rates, the larger calibration coefficie
nts associated with larger thresholds tend to increase the bias and ra
ndom error in the resulting areal mean rain-rate estimates.