The use of time series of vegetation indices obtained from satellites has become a highly relevant source of data in studies of land degradation assessment and monitoring. However, information about future vegetation dynamics, which is key for early warnings oriented at land management decisions, is still lacking. Trend-cycle was recently proposed as an indicator that represents a smoothed version of a seasonally adjusted data series, which provides information on long-term movements (trend) while including changes in direction underlying the series (cycles). The aim was to estimate the direction and speed of change of the Normalized Difference Vegetation Index (NDVI) trend-cycles as a complementary information of the rhythm of change between cyclic phases of vegetation productivity. In particular, we estimate the first and second derivative of the end-point of the trend-cycle function, as a measure of the direction the function is going and the speed of change, respectively. The potential advantage of our proposal is the fast processing for large areas and its sensitivity to capturing shifts in temporal dynamics for short time series data. This information can be used as a proxy to build scenarios of the future behaviour of vegetation dynamics, which is a relevant issue to move forward in the development of early warning tools for adaptive land management.