We present a detailed study of the Delta -variance as a method to qunatify
molecular cloud structure. The Delta -variance was introduced by Stutzki et
al. (1998) to analyze the drift behaviour of scalar functions and is used
to characterize the spatial structure of observed molecular cloud images. F
or fractional Brownian motion structures (fBm-fractals), characterized by a
power law power spectrum and random phases, the Delta -variance allows to
determine the power spectral index beta. We present algorithms: to determin
e the a-variance for discretely sampled maps and study the influence of whi
te noise, beam smoothing and the finite spatial extent of the maps. We find
that for images with beta > 3, edge effects can bias the structure paramet
ers when determined by means of a Fourier transform analysis. In contrast,
the Delta -variance provides a reliable estimate for the spectral index bet
a, if detcrnlined in the spatial domain. The effects of noise and beam smoo
thing are analytically represented in a lending order approximation. This a
llows to use the Delta -variance of observed maps even at scales where the
influence of both effects becomes significant, allowing to derive the spect
ral index beta over a wider range and thus moro reliably than possible othe
rwise. The Delta -variance is applied to velocity integrated spectral line
maps of several clouds observed in rotational transitions of (CO)-C-12 and
(CO)-C-13. We find that the spatial structure of the emission is well chara
cterized by a power law power spectrum in all cases. For linear scales larg
er than similar to0.5 pc the spectral index is remarkably uniform for the d
ifferent clouds and transitions observed (2.5 less than or equal to beta le
ss than or equal to 2.8). Significantly larger values (beta greater than or
similar to 3) are found for observations made with higher linear resolutio
n toward the molecular cloud MCLD 123.5+24.9 in the Polaris Flare, indicati
ng a smoother spatial structure of the emission at small scales (<0.5 pc).