Comparison of traditional ground-based observations and digital remote sensing of phenological transitions in a floodplain forest

Phenological observations are important as indicators of global warming and as estimation tools for the terrestrial carbon balance in vulnerable ecosystems, such as the last fragments of floodplain forests in the Czechia. The aim of this paper was to compare ground-based phenological observations of three dominant species (European hornbeam, English oak and narrow-leaved ash) in this ecosystem, with the seasonal trajectory of the greenness index (Gcc) and thresholds extracted from images taken by phenocameras located on a meteorological mast. The average annual air temperature in the studied years 2014–2017 was 1 °C higher than the long-term average, and the precipitation deficit reached ⅓ of annual rainfall. We found a high proportion of above-average warm days in the warmest part of the growing season. Above-average air temperatures significantly accelerated the onset of budbreak in ash. Yet a higher proportion of above-average air temperatures prolonged the period between budbreak to fully developed leaf area, especially in ash and oak. In 2017, rapid cooling after exceptionally warm temperatures at the onset of spring had a detrimental effect on the stand productivity and showed a marked effect on the phenological shifts. The period when leaf area developed was in the range of DOY 66-286 for hornbeam, DOY 79-329 for oak and DOY 88-321 for ash in 2014–2017. The seasonal trajectory of Gcc showed differences between tree species that corresponded to the dynamics of the onset of phenophases observed in the field. According to image analyses, the phenophase of greenup and maturity for hornbeam and ash had minimal uncertainty. In contrast, the uncertainty was high in the determination of phenophases for oak. Our observations show that the modern method of phenological observation by phenocameras is suitable for mixed forests, but classical ground-based observations by a phenologist are still crucial in order to verify the results.