Abstract
Cold regions are characterized by fragile and low resilience ecosystems that are strongly affected by global warming. Temperature variation, besides regulating the dynamics of snow, glaciers, and permafrost, influences the phenology of vegetation and the productivity of mountain and arctic ecosystems. The difficulty of accessing these sites and the extreme weather conditions during most of the year make satellite remote sensing an effective strategy for monitoring cold regions. One way to collect images from space is based on Synthetic Aperture Radar(SAR)sensors. Thesesensorsallowa consistent observation of the Earth’s surface, despite clouds and darkness conditions. These features make SAR sensors useful for monitoring phenological development during the whole vegetation growing season. Nevertheless, the results are difficult to be assessed due to the contribution of soil and vegetation to the σ0 backscattering coefficients. Conversely, optical satellite sensors investigate the Earth’s surface in the portion of the electromagnetic spectrum that belongs to the visible, near-infrared, mid-infrared and thermal-infrared regions. During the growing season, these sensors provide a description of radiation variations that is related to chemical changes in plants, although the consistency of time series is severely limited in cold regions by the presence of clouds and polar night. The main objective of this thesis is to exploit the synergy between different types of sensors in order to improve the accuracy of the detection of vegetation phenology in cold regions. Hence, a multi-temporal and multi-sensor approach is presented and discussed. Specifically, this work explores the capabilities and the limitations of the SAR backscatter in the retrieval of phenological phases of mountain, high mountain and arctic grasslands. The results are evaluated together with the optical sensors, in a data integration perspective. First, the potential of Sentinel-1 dual-pol backscatter to detect phenological changes in different land-use types of South-Tyrol, an Italian Province located in the European Alps, was assessed. Then, the analysis focused on the phenological phases and mowing retrieval in mountain meadows up to 2000 m a.s.l. Next, a multi-polarization analysis of the SAR response to mountain vegetation (meadows and pastures) was carried out. Therefore, the ability to follow phenological changes in vegetation up to 2700 m a.s.l. was evaluated along with the limitations derived from the structure of high mountain plants. Finally, the backscatter signal response to grassland growth in the Arctic was analyzed together with snow seasonality and freezing-thawing conditions. Within this framework, an Arctic sampling method for the validation of satellite data was described. The major outcomes of this thesis are: (i) the SAR signal can detect phenological cycles in different vegetation cover types of mountain regions; (ii) a significant correlation between the cross-polarized channel from the SAR sensor and the NDVI optical index was found in mountain grasslands; (iii) the SAR signal is able to detect the onset, maximum and end of the growing season in mountain meadows; (iv) the SAR time series can be integrated with the optical ones in the study of vegetation dynamics in cold regions; (v) the structure of alpine and arctic vegetation limits the abilities of the backscatter to follow phenological phases; (vi) due to the structure of thevegetation,athighelevationsandinsomeArcticplantcommunities,thesoilcomponent becomes predominant over the vegetation on the backscattering coefficients; (vii) the snow dynamics, and therefore the thawing process of the soil, affect the SAR signal’s capability to monitor vegetation growth.