In this paper, we present the R package TrendTM dedicated to the trend estimation of high dimensional time series matrices. The main features of this package is the possibility to take into account different types of temporal structures on the data (none, smooth or periodic) and to select, according to the user's wishes, either the rank $k$ of the matrix, the parameter $\tau$ linked to the temporal structure or both. Then, a two-stage heuristic is provided for the joint selection of the rank $k$ and of the tuning parameter $\tau$. Moreover, the TrendTM function is applied to time series clustering and principal component analysis on real datasets. The package is available on the CRAN.