Narrowband-filter imaging spectrophotometry is a powerful tool for studying
the spatial behavior of physical parameters, such as interstellar reddenin
g, excitation, electron temperature, and densities or chemical abundances a
cross extended objects such as nebulae and galaxies. We explore the hazards
and technical limits of the method of spectral imaging with narrowband int
erference filters. We discuss the specific application of statistical error
s involved in the estimation of abundance gradients from nebular line flux
measurements and the impact of those errors on interpretation. We demonstra
te that a sample of at least 16 H II regions is needed for a meaningful and
robust description of radial abundance gradients in disk galaxies. We also
discuss the systematic errors introduced by the uncertainty related to the
subtraction of the underlying stellar continuum from monochromatic emissio
n-line images; we demonstrate that the quality of subtraction of the underl
ying continuum is the main limit on the precision of imaging spectrophotome
try with narrowband filters. Finally, we investigate the impact of selectio
n effects on the derivation of physical parameters from measurements involv
ing the ratio of two or more spectral lines.