This article presents an approach coupling space filling curves and Markov chains for analysing spatial data about hedge localisation. Due to the data spatial heterogeneity, an adaptative Hilbert curve is used, allowing space linearisation with respect to data local density. In order to exploit the resulting sequence, we need to characterise both the distance between a point and its predecessor on the curve and the local density. We propose therefore to compute an access time for a point from its predecessor based on the cutting depth notion. This variable, together with the hedge characterising variables are then analysed through a Markov model. Results obtained on a dataset of about 10000 hedge segments from a zone in the Durance low valley are presented and discussed.