Abstract:

Phenology is a subdiscipline of biology that focuses on the timing of periodic biological events in organisms and their relationships with environmental factors, particularly climatic variables, such as temperature, precipitation, and photoperiod. It primarily investigates the temporal patterns of biological processes, such as growth, development, reproduction, dormancy, and senescence in plants, animals, and microorganisms across different timeframes. Among its branches, vegetation phenology, which refers specifically to plants, is a key component that deals with the timing of physiological stages in plant life cycles throughout the growing season. Research on vegetation phenology is essential for understanding how ecosystems function, particularly in relation to the carbon and water cycles. It plays a critical role in quantifying the exchange of energy and matter between the biosphere and atmosphere and in evaluating the ecological consequences of climate variability and long-term climate change. Phenological shifts serve as sensitive bioindicators of environmental change; thus, vegetation phenology has become a key focus in global change research. It is now widely used to track the effects of climate anomalies on terrestrial ecosystems, agricultural productivity, and biodiversity patterns. Vegetation phenology can be further divided into structural phenology, which reflects the visible morphological changes in plants, such as leaf-out, flowering, and senescence, and functional phenology, which describes the dynamic physiological and metabolic processes underlying plant growth and ecological functions of plants. Although distinct, these two aspects are closely interrelated and together provide a comprehensive picture of plant-environment interactions. The combination of both perspectives enhances our ability to monitor vegetation health, detect stress, and assess species' adaptive responses to environmental variability. With the accelerating pace of global climate change, there is an increasing demand for high-resolution, multi-scale monitoring of vegetation phenology. Accurate and timely phenological data are required to model ecosystem responses, improve the parameterization of climate and ecological models, and inform management strategies in agriculture, forestry, and conservation sectors. The ability to detect changes in the timing of phenological events allows for early warnings of ecological disruption and can inform decisions regarding resource allocation and the protection of biodiversity. Currently, phenology monitoring is conducted using various approaches, each with its own advantages and limitations. Ground-based phenological observations offer detailed and accurate measurements at the canopy and individual plant levels. These methods provide high temporal fidelity and can capture subtle biological changes. However, the spatial coverage of ground stations is often limited by human, material, and financial resources, making it difficult to scale up to regional or global assessments of the data. Near-surface phenology monitoring using automatic time-lapse digital cameras (phenocams), eddy covariance technology, and unmanned aerial vehicles (UAVs) has emerged as a valuable extension of human observation. These tools can bridge the gap between plot-level ground observations and broad-scale satellite-based observations. They enhance spatial representativeness while retaining fine temporal and spatial resolutions. Moreover, near-surface methods support the development of robust retrieval algorithms and phenological models by providing dense time series data for various ecosystems. In recent years, they have been increasingly deployed for long-term ecological research and agricultural applications. Remote sensing-based phenological inversion offers a unique macroscopic perspective. Satellite platforms provide repeated and consistent observations over large spatial extents and long periods. This makes it possible to analyze cumulative phenological trends and interannual variability on regional and global scales. Remotely sensed phenological data are particularly valuable for detecting shifts in phenological changes under the influence of global warming. However, the trade-off between spatial and temporal resolution remains a key challenge. Higher-resolution imagery often has a lower revisit frequency, whereas sensors with frequent coverage may lack sufficient spatial detail. In this context, this study reviews the foundational applications of ground-based observations, near-surface sensing, and satellite remote sensing for vegetation phenology monitoring. It compiles and discusses the available phenological data from national ground-based monitoring networks in China and outlines the inversion principles, technical advantages, and practical applications of widely used observation tools, including phenocams, eddy covariance towers and UAVs. This review further introduces commonly used satellite platforms, such as MODIS, Sentinel-2, and the Landsat series, along with associated vegetation indices, such as the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Leaf Area Index (LAI). These indices are essential proxies for characterizing vegetation status and extracting phenological metrics. Based on these tools and data, this study examined the latest methodological advances and representative case studies for extracting phenology parameters from different remote sensing sources. This study also discusses the data processing approaches used for remote-sensing time-series analysis in phenology monitoring. Classical methods, such as Savitzky-Golay filtering, asymmetric Gaussian fitting, and logistic curve modeling are described in terms of their principles, strengths, and limitations. In addition, more recent techniques, such as the wWHd method, a spatially weighted Whittaker smoother with a dynamic regularization parameter λ, and Gaussian Process Regression (GPR), are highlighted. These advanced methods allow for a more accurate representation of multi-season growth patterns and improve curve fitting under noisy or uncertain data. New research progress helps broaden innovative thinking in this field. Furthermore, this study introduced a variety of software tools designed for vegetation phenology extraction. These include open-source code libraries and graphical user interface-based software applications. Several tools can handle dual growing seasons or extract multiple phenological phases within a single growing season. The comparison covers their core features, ease of use, compatibility with different data formats, and supported application domains. The increasing availability of modular and user-friendly tools has significantly lowered the barriers to entry for researchers in multiple disciplines. In the final section, we present a forward-looking perspective on the future trends in vegetation phenology monitoring. This emphasizes the critical role of multisource data fusion, which combines ground-based, near-surface, and satellite observations to enhance spatial coverage, accuracy, and consistency. Additionally, this study explored the potential of unconventional data sources, such as mobile phone applications and social media, to complement formal monitoring systems and offer novel insights into vegetation changes from the perspective of citizen-science. Finally, this review highlights the increasing importance of machine and deep learning in phenological modeling and inversion. These techniques offer powerful tools for identifying patterns in high-dimensional data and simulating complex ecological processes. In particular, under the backdrop of big data, artificial intelligence based approaches hold great promise for intelligent phenology recognition, spatiotemporal prediction, and ecosystem response modeling. The progress of future vegetation phenological information inversion may rely on multi-source data fusion and the application of machine learning, particularly deep learning.

Key words: vegetation phenology, remote sensing technology, phenology inversion, machine learning, multi-scale monitoring

摘要：

植被物候是指植物在生长季节中各个生理阶段的时间变化，其研究对揭示生态系统碳水循环机制、评估气候变化的生态效应具有重要意义。随着全球气候变化的加剧，高精度、多尺度的植被物候监测已成为生态学、气候学和环境科学研究中的重要课题。植被物候遥感反演独特的宏观视角不仅能通过长时序数据追溯物候变化的累积效应，而且能为全球变暖背景下植被动态监测提供核心的技术支撑。文章综述了地面观测、近地面传感与卫星遥感技术在植被物候监测中的应用基础，并探讨了不同遥感数据源在植被物候参数提取中的最新研究进展；阐述了面向植被物候监测的遥感时间序列观测数据处理方法的基本原理、优势以及可能存在的不确定性来源，并深入探讨了植被物候信息提取的技术方法；文章进一步给出了植被物候提取的常用软件包，详细讨论了它们的功能和优劣；最后展望了植被物候监测技术的发展方向，强调了多源数据融合在提升物候监测空间覆盖与精度方面的重要性，指出新型观测手段与社交媒体等多元数据源的应用潜力，同时分析了机器学习方法在物候反演、模型构建及复杂生态过程建模中的应用前景。

关键词: 植被物候, 遥感技术, 物候反演, 机器学习, 多尺度监测