The main objective of this study is to identify grassland management practices using time series of remote sensing images. The accelerating agricultural intensification has strongly reduced grassland surfaces in some areas, generating important changes in their management and affecting environmental and socio-economic systems. Therefore, the identification of grassland management practices in farming systems is a key issue for sustainable agriculture. To this end, the leaf area index (LAI) estimated from remote sensing images was used since its temporal evolution is informative about farming practices. We evaluate the performances of two common classification algorithms using time profiles of LAI derived from simulated data and high spatial resolution satellite images. We show that they exhibit limited performances, mainly because they rely on criteria that are not suited for the comparison of time series. We then suggest the use of more advanced classification tools that work in a transformed space designed by a kernel function. We show that a kernel based on time warping measurements which are suited for the comparison of time series, perform better than classical ones based on Gaussian functions. This is a promising result for the analyzing of the future SENTINEL data that will be embedded in many time series.