Abstract Seasonal phenological dynamics of vegetation hold important clues on ecosystem performance towards management goals, such as carbon uptake, and thus should be considered in projections of their targeted services. However, in wetlands spatio-temporal heterogeneity due to mixing of open water, soil, green and dead vegetation makes it difficult to generalize ecosystem functioning across different regions. Remote sensing observations can provide spatially-explicit, cost-effective phenology indicators; however, little is known about their capacity to indicate the links between wetland ecosystem structure and function. Here we assessed this potential by comparing one-year Enhanced Vegetation Index (EVI) from satellite products at high (5m; RapidEye) and low (30m; Landsat) spatial resolutions with eddy covariance time series of net carbon exchange, field digital camera (phenocam) greenness and water temperature among three floristically similar restored wetlands in California, USA. Phenological timing differed by wetland site: depending on satellite, the range in site-median start of greening was up to 28 days, end of greening – up to 73 days, start of senescence – up to 79 days, and end of senescence – up to 10 days. Key transition dates from satellite inputs agreed with seasonal changes in net carbon exchange, phenocam greenness and water temperatures, suggesting that phenological contrasts could result in part from site differences in vegetation configuration and litter affecting the exposure of canopy, soil and water to sunlight and thus sub-canopy microclimate and ecosystem functioning. Yet, the agreement between satellite inputs was non-systematic, with the greatest disparities at the more heterogeneous, less vegetated site. Phenological model fitting uncertainty increased with greater spatial resolution, highlighting the tradeoff between the accuracy of representing vegetation and the complexity of local seasonal variation. These findings highlight the sensitivity of satellite-derived phenology to structural and functional heterogeneity of ecosystems and call for more rigorous spatially-explicit analyses to inform assessments of restoration and management outcomes.

中文翻译：

---

恢复湿地的遥感物候异质性：连接植被结构和功能

摘要 植被的季节性物候动态为生态系统表现向管理目标（如碳吸收）提供重要线索，因此应在其目标服务的预测中加以考虑。然而，在湿地中，由于开放水域、土壤、绿色和死亡植被的混合，时空异质性使得难以概括不同区域的生态系统功能。遥感观测可以提供空间明确、具有成本效益的物候指标；然而，人们对其表明湿地生态系统结构和功能之间联系的能力知之甚少。在这里，我们通过比较高（5m；RapidEye）和低（30m；RapidEye）卫星产品的一年增强型植被指数 (EVI) 来评估这种潜力。Landsat) 空间分辨率与涡流协方差时间序列的净碳交换、现场数码相机 (phenocam) 绿度和水温在美国加利福尼亚州三个植物区系相似的恢复湿地之间。不同湿地地点物候时间不同：取决于卫星，地点范围内绿化开始的中位数范围长达 28 天，绿化结束 - 长达 73 天，衰老开始 - 长达 79 天，衰老结束 - 高达到 10 天。卫星输入的关键过渡日期与净碳交换、苯酚绿度和水温的季节性变化一致，这表明物候对比可能部分是由于植被构造和凋落物的场地差异影响了树冠、土壤和水对阳光的暴露，因此亚冠层微气候和生态系统功能。然而，卫星输入之间的一致性是非系统性的，在更多样化、植被较少的地点差异最大。物候模型拟合的不确定性随着空间分辨率的提高而增加，突出了表示植被的准确性与局部季节性变化的复杂性之间的权衡。这些发现突出了卫星衍生物候学对生态系统结构和功能异质性的敏感性，并呼吁进行更严格的空间显式分析，为恢复和管理结果的评估提供信息。强调表示植被的准确性和局部季节性变化的复杂性之间的权衡。这些发现突出了卫星衍生物候学对生态系统结构和功能异质性的敏感性，并呼吁进行更严格的空间显式分析，为恢复和管理结果的评估提供信息。强调表示植被的准确性和局部季节性变化的复杂性之间的权衡。这些发现突出了卫星衍生物候学对生态系统结构和功能异质性的敏感性，并呼吁进行更严格的空间显式分析，为恢复和管理结果的评估提供信息。