The monitoring of crop dynamics is generally important by both environmental and economic points of view, because of its impacts on ecosystems, food security and markets. There is a lot of literature regarding monitoring of crops by Earth Observation techniques for different applications: ranging from crop season monitoring at continental scale, to optimization of agricultural practices at field level. The use of EO-based monitoring techniques for crop damage assessment is more recent, especially at farm scale. The underpinning concept is that the analysis of EO data could provide evidences, at different temporal and spatial scales, of crop damage occurrence and extent, which are relevant information for different stakeholders and players (i.e. public authorities, insurance companies, consultants, farmers). In this study the application is committed to the detection and recognition of crop damages hints at within-field level, with the final aim to provide loss adjusters with added-value information to support their decision making, while estimating yield losses as due to crop lodging. The proposed methodology is configured to automatically detect within-field spatial patterns presenting homogeneous crop dynamics, i.e. characterized by a similar development trend throughout the season; such patterns could likely be associated to the occurrence/absence, or different severity, of weather related damages. A clustering procedure (fuzzy C-means) was applied to a Sentinel 2 (A and B) time series - from June to August 2017 - covering the crop season of maize in a cereal farm located in Northwest Italy; the modified soil-adjusted vegetation index (MSAVI) was used as the feature for representing crop development over time. In addition, a UAV survey was performed with a multispectral camera (MicaSense RedEdge) at the end of season, over three parcels affected by severe lodging. During the overflight a field survey (tracked by a GPS) was accomplished together with a loss-adjuster, and reference targets were placed for image correction. The UAV images were orthomosaicked, then clustered, and labelled according to the collected ground truth, and the loss-adjuster evaluations. The case at hand was particularly difficult since lodging was not characterized by the plant stems lying horizontally, as usual; the maize stems were fairly straight and grown, but bended at bottom internodes. This because the damaging wind storm event took place at an early growth stage, and plants kept on their vegetative developing, this causes crop damages to be hardly recognisable, not only within the field, but also from above. The results of clustering on multi-temporal Sentinel 2 data were compared with the lodging maps obtained from UAV, and the approach proved to be very promising in the perspective of providing added value information to loss adjusters.