Landsat‐based greening trends in alpine ecosystems are inflated by multidecadal increases in summer observations

Summary

This paper demonstrates a critical methodological artefact in Landsat-based vegetation monitoring: greening trends derived from annual maximum NDVI are significantly overestimated in seasonally snow-covered ecosystems because the density of satellite observations increases over the archive period. The bias is most pronounced in habitats with short growing seasons and sparse observations, such as late-snowmelting alpine areas. The authors recommend mandatory reporting of temporal sampling characteristics when publishing long-term Landsat studies to enable appropriate contextualisation and comparison between ecosystems.

Regional applicability

The findings are directly applicable to United Kingdom upland and alpine ecosystems (Scottish Highlands, Lake District, Snowdonia), particularly areas with late snowmelt that experience similar short growing seasons and historical observation gaps. However, lower-elevation UK habitats with longer growing seasons and denser Landsat coverage may experience smaller biases. The methodological caveat is globally relevant for any greening trend study using Landsat archives.

Key measures

Annual maximum normalised difference vegetation index (NDVI); temporal sampling frequency (observations per year); magnitude of greening trend bias as a percentage; bias variation along environmental gradients

Outcomes reported

The study quantified how inconsistent Landsat observation frequency over time inflates estimated greening trends in snow-covered alpine ecosystems, demonstrating that approximately 50% of greening magnitude in late-snowmelting European Alpine habitats can be attributable to this sampling bias rather than genuine vegetation change.

Topic tags