Accurate and frequently updated land cover and crop type maps are essential for environmental monitoring, agricultural management, and informed decision-making. While multi-temporal SAR and multispectral data from various satellite sensors have been widely assessed towards this end, the potential of spaceborne hyperspectral Environmental Mapping and Analysis Program (EnMAP) data for such applications, remains largely unexplored. This study evaluates the added value of multi-temporal EnMAP observations combined with Sentinel-1 and Sentinel-2 time series for joint land cover and crop type mapping in a complex landscape of Western Greece. A detailed reference dataset comprising 32 classes (14 non-crop and 18 crop categories) was developed, and progressively richer sensor configurations were evaluated using a Random Forest classifier. Different representations of the high-dimensional EnMAP feature space were assessed, including original bands, spectral indices, Principal Component Analysis (PCA), and Minimum Noise Fraction (MNF) transformations. All experimental configurations achieved Overall Accuracy (OA) rates above 88% and average F1-score above 78%. Incorporating Sentinel-1 temporal metrics increased OA from 88.13%, using Sentinel-2 spectro-temporal metrics alone, to 89.40%, while the addition of EnMAP features further improved performance across all feature representations. The highest OA (90.27%) and average F1-score (83.45%) were achieved by combining Sentinel-2 and Sentinel-1 metrics with the MNF-transformed EnMAP features. Overall, multi-temporal EnMAP information enhanced notably the discrimination of spectrally complex and overlapping classes, particularly those with lower baseline performance, while dimensionality reduction techniques provided the greatest classification benefits. These results encourage further research on hyperspectral time series over broader spatial extents and their integration into operational land cover and crop type mapping frameworks.