Integrating in–situ data and remote sensing for the spatiotemporal assessment of alpine vegetation

Abstract

Alpine plant communities are structured by fine–scale interactions among soil properties, topography and microclimatic conditions, but these gradients are difficult to monitor consistently across space and time. Field observations provide the ecological reference needed to interpret vegetation patterns, whereas satellite data may offer spatially continuous and repeatable indicators whose ecological meaning must be tested against in–situ measurements. Here, we assessed whether Sentinel–2 derived indicators reproduce the main ecological gradients observed from vegetation, soil and microclimatic field data in alpine grasslands. We surveyed 40 vegetation plots in the southwestern Cantabrian Mountains, measuring soil and topographic variables, and monitoring soil temperature and water potential at four representative strip grassland sites from 2021 to 2025. Satellite–derived indicators of vegetation productivity, moisture and land surface temperature were evaluated against field observations across multiple spatial scales. Soil texture and spatial structure explained a moderate fraction of vegetation compositional variation, while summer Soil–Adjusted Vegetation Index (SAVI) was the only remote sensing indicator significantly associated with community composition. This shows that satellite data reproduced part of the field–detected productivity gradient, but not the full floristic complexity of these communities. In multi–date field–satellite comparisons, remote sensing indicators showed significant but scale–dependent relationships with soil temperature and water stress. Land Surface Temperature (LST) emerged as the most robust predictor of short–term ecosystem functioning, consistently explaining both soil temperature and water–stress dynamics across spatial scales, whereas SAVI and NDMI provided complementary information related to vegetation productivity and moisture conditions. Remote sensing provides therefore spatially scalable proxies of selected ecological signals in heterogeneous alpine landscapes. However, their interpretation depends on field–based calibration, the scale of observation and requires caution when transferring the observed relationships beyond monitored communities. Acknowledging these limitations, this approach supports patial extrapolation of alpine ecosystems assessments for habitat mapping, monitoring and conservation under global change.