Assimilation of repeated woody biomass observations constrains decadal ecosystem carbon cycle uncertainty in aggrading forests
Assimilation of repeated woody biomass observations constrains decadal ecosystem carbon cycle uncertainty in aggrading forests
复制标题
DOI：
10.1002/2016jg003520
复制
发表时间：
2017-03
期刊：
影响因子：
--
通讯作者：
T. Smallman;J. Exbrayat;Maurizio Mencuccini;A. Bloom;M. Williams
中科院分区：
文献类型：
--
作者：
T. Smallman;J. Exbrayat;Maurizio Mencuccini;A. Bloom;M. Williams
森林碳汇强度受植物生长、死有机物矿化和干扰的控制。在整个景观中，遥感可以提供有关森林地上状态的信息，并且可以将这些信息与模型联系起来，以估计接近稳定状态的森林中的碳循环。对于森林退化来说，这种方法更具挑战性，而且尚未得到证实。在这里，我们应用贝叶斯方法，将一个简单的模型与一系列数据联系起来，以评估两个退化森林的信息内容。我们将使用当地观测的高信息含量分析与使用逐渐稀疏的遥感信息（重复、单一和无木质生物量观测）的检索进行比较。两个森林的净生物群落生产力被限制为净汇，在输入范围内变化<2 Mg C ha−1 yr−1。然而，特定碳库的封存随着同化的生物量信息而变化。与使用单一或无种群信息的分析相比，重复生物量观测的同化减少了所有生态系统碳库（而不仅仅是木材）的不确定性和/或偏差。作为验证，我们的重复生物量分析解释了一个森林中凋落物动态变化的 78-86%，而在第二个森林中，死亡有机物总量估计值处于观测不确定性范围内。与生物量信息较少的分析相比，重复生物量分析中检索到的生态系统性状的不确定性降低了 50%。这项研究量化了重复木质观测在限制木材和死亡有机物动态方面的重要性，强调了拟议的遥感任务的好处。
Forest carbon sink strengths are governed by plant growth, mineralization of dead organic matter, and disturbance. Across landscapes, remote sensing can provide information about aboveground states of forests and this information can be linked to models to estimate carbon cycling in forests close to steady state. For aggrading forests this approach is more challenging and has not been demonstrated. Here we apply a Bayesian approach, linking a simple model to a range of data, to evaluate their information content, for two aggrading forests. We compare high information content analyses using local observations with retrievals using progressively sparser remotely sensed information (repeated, single, and no woody biomass observations). The net biome productivity of both forests is constrained to be a net sink with <2 Mg C ha−1 yr−1 variation across the range of inputs. However, the sequestration of particular carbon pool(s) varies with assimilated biomass information. Assimilation of repeated biomass observations reduces uncertainty and/or bias in all ecosystem C pools not just wood, compared to analyses using single or no stock information. As verification, our repeated biomass analysis explains 78–86% of variation in litter dynamics at one forest, while at the second forest total dead organic matter estimates are within observational uncertainty. The uncertainty of retrieved ecosystem traits in the repeated biomass analysis is reduced by up to 50% compared to analyses with less biomass information. This study quantifies the importance of repeated woody observations in constraining the dynamics of both wood and dead organic matter, highlighting the benefit of proposed remote sensing missions.