The variance of a remotely sensed image is determined by the interaction of
scene properties with the spatial characteristics of the sensor. image var
iance is related to information content, and therefore determines the abili
ty to extract useful information about scene conditions. Mie describe a tec
hnique to estimate image variance at multiple spatial resolutions. The meth
od is useful for comparing the capabilities of sensors with differing spati
al responses.
The point-spread function (PSF) and the variogram quantify the spatial char
acteristics of the sensor and image, respectively. A geostatistical model b
ased on these two elements relates the punctual variogram of a scene with t
he regularized variogram of an image. This model forms the basis for a nume
rical approach to approximate the punctual variogram from regularized obser
vations. The resulting estimate of the punctual variogram allows analytical
determination of image variance at different spatial resolutions.
Analysis of simulated images confirms the utility of this algorithm. Varian
ce of coarse-resolution images may be estimated reliably from fine-resoluti
on data. Simulations of multiscale variability show that the method handles
more complex types of scene variability as well. The geostatistical varian
ce estimation algorithm better characterizes the relationship between varia
nce and spatial resolution than do simpler methods, such as averaging block
s of pixels. Specifically, methods which do not account for overlap of adja
cent placements of the sensor PSF tend to overestimate the variance of the
resulting images. The algorithm presented here can be used to evaluate the
utility of different sensors for particular applications, when the relation
ship between spatial resolution and image information content is important.