European beech spring phenological phase prediction with UAV-derived multispectral indices and machine learning regression

Aug 4, 2026

Acquiring phenological event data underpins research on climate-change impacts in forest dynamics and the risks linked to early leaf-out in young trees. While mapping phenological timing via Earth-observation data can add crucial spatial context, converting ground-based observations into reliable training ground truth remains challenging. This study evaluates predicting high-resolution European beech spring phenological phases using UAV-borne multispectral indices and machine-learning regression with systematic feature selection. The best-performing Green Chromatic Coordinate with GAM boosting model generalizes across sites, achieving robust unseen-dataset performance with limitations for near-infrared oversaturation indices.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages23

Reading Time58 min

AccessFree to download

Introduction

Significance of phenology and climate-change relevance

Challenges in ground truthing and remote sensing

Study objective and approach

UAV-based multispectral indices and machine learning

Feature selection and model evaluation

[www. nature.com/scientificreports](www. nature.com/scientificreports) OPEN European beech spring phenological phase prediction with UAV‑derived multispectral indices and machine learning regression Stuart Krause1,2 & Tanja Sanders1* Acquiring phenological event data is crucial for studying the impacts of climate change on forest dynamics and assessing the risks associated with the early onset of young leaves. Large‑scale mapping of forest phenological timing using Earth observation (EO) data could enhance our understanding of these processes through an added spatial component. However, translating traditional ground‑ based phenological observations into reliable ground truthing for training and validating EO mapping applications remains challenging. This study explored the feasibility of predicting high‑resolution phenological phase data for European beech (Fagus sylvatica) using unoccupied aerial vehicle (UAV)‑ based multispectral indices and machine learning. Employing a comprehensive feature selection process, we identified the most effective sensors, vegetation indices, training data partitions, and machine learning models for phenological phase prediction. The model that performed best and generalized well across various sites utilized Green Chromatic Coordinate (GCC) and Generalized Additive Model (GAM) boosting. The GCC training data, derived from the radiometrically calibrated visual bands of a multispectral sensor, were predicted using uncalibrated RGB sensor data. The final GCC/GAM boosting model demonstrated capability in predicting phenological phases on unseen datasets within a root mean squared error threshold of 0.5. This research highlights the potential interoperability among common UAV‑mounted sensors, particularly the utility of readily available, low‑cost RGB sensors. However, considerable limitations were observed with indices that implement the near‑infrared band due to oversaturation. Future work will focus on adapting models to better align with the ICP Forests phenological flushing stages. Keywords Phenology, UAV, Machine learning, Intensive forest monitoring First proposed by the Swedish botanist in his work Philosphia Botinica in 17511, the concept of gathering data on the timing of leaf opening, flowering, fruiting, and leaf fall alongside climatological observations “so as to show how areas differ” is still relevant today2. Historically phenological observations assisted in agriculture by means of predicting the timing of cultivation practices3 and emerged as a scientific discipline in the last 100 years4. Recently recognized as bioindicators of climate change5, phenological data proofs a sensitive proxy for climate change investigation4 due to the observed relationship between phenological timing and changing climate. In particular, spring phenology mirrors the changing temperatures6. Understanding phenological variations at the stand level provides insights into early spring flushing advances and the risk of late frost damage. This historical foundation sets the stage for contemporary research, which is now increasingly focused on the impacts of recent climatic shifts as highlighted by the Intergovernmental Panel on Climate Change (IPCC) . The IPCC reported a 1.53°C increase in average land temperature for the period 2006–2015 in comparison to the 1850–1900 period (IPCC, 2018). Warmer temperatures, alongside changing precipitation patterns, altered the growing seasons, causing increased tree mortality (IPCC, 2018); however, warmer temperatures may also lead to increased carbon storage due to longer growing seasons7. These extended by approximately 10–20 days in 1Thünen Institute of Forest Ecosystems, Alfred-Möller-Str. 1, Haus 41/42, 16225 Eberswalde, Germany. 2Department of Geography, University of Bonn, Meckenheimer Allee 166, 53115 Bonn, Germany. *email: tanja.sanders@thuenen.de [www. nature.com/scientificreports/](www. nature.com/scientificreports/) recent decades, and pr

FAQs

Why is predicting European beech spring phenological phases important?

It helps quantify climate-change impacts on forest dynamics and assess risks such as late frost damage to newly unfolded leaves, especially in younger stands.

What data sources are used for the prediction in this study?

UAV-derived multispectral indices are used, with training derived from calibrated visual bands; Green Chromatic Coordinate training is predicted from uncalibrated RGB sensor data.

Which model configuration performed best and how well did it generalize?

The GCC with GAM boosting approach performed best and generalized across multiple sites, predicting unseen datasets with root mean squared error below 0.5.

Aug 4, 2026

Upload by River Wang