Heterogeneous agricultural landscapes are beneficial for biodiversity conservation. However, croplands' study remains challenging due to their dynamics at different spatial scales (plot, landscape and region) and temporal scales (intra- and inter-annual dynamics). Most of the studies have focused on wooded and semi-natural elements rather than crops in connectivity models, or the input crop characteristics are static. Detecting the spatio-temporal changes in cropland is essential to understand the ecological functions of the agricultural mosaic. The recent synthetic aperture radar (SAR) Sentinel-1 (S-1) and optical Sentinel-2 (S-2) time series offer a great opportunity to monitor cropland thanks to their high spatial and temporal resolutions. In this study, we assessed the structural changes of wheat and rapeseed crops using S-1 & 2 time series in Amiens and Pleine-Fougères. These two French temperate agricultural landscapes, present a gradient from open field to wooded landscapes. Firstly, we assessed the potential of single S-1 data, single S-2 data and their joint use to characterize rapeseed and wheat crops (biomass, LAI and phenological stages) using field data inventories. Random Forest (RF) algorithm has been used to identify phenological stages of wheat and rapeseed crops from the Sentinel images and general linear models (GLM) to predict biomass, LAI and height. Secondly, we studied the influence of landscapes metrics on crops characteristics using GLM. The first results show that the Sentinel time series have a good potential to identify phenological stages of wheat and rapeseed. S-1 was better than S-2 to identify phenological stages of wheat, whereas S-2 was better than S-1 to identify phenological stages of rapeseed. A significant correlation (r²=0.64) was found between the dry and wet biomass of wheat and VH polarization of S-1. Finally, we will present how a carabid species involved in pest regulation respond to these dynamics.