箱式法目前是观测土壤呼吸的最主要方法，有人工操控的也有全自动的。但由于受到经费等条件的限制，以及开展较大空间尺度上多点观测的需求，高分辨率的连续自动观测系统很难获得，而分辨率较低的人工操控的观测法则是目前土壤呼吸测定的主要方法。研究表明，土壤呼吸在日、周、月和年尺度上均波动很大。中亚热带地区，由于温度和降雨的季节波动，导致森林土壤呼吸随时间显著变化，这给准确捕捉波动规律和正确估算年土壤呼吸量带来了很大不确定性。而对于多大观测频度才能满足不同的观测要求却又缺乏统一的标准，因此，探讨监测频率与模式对精确估算年土壤呼吸的影响显得极为必要。

    为此，我们利用监测步长为每周2次的长期定位土壤呼吸观测数据，根据时间序列，抽出每周1次、每两周1次、每月1次的观测数据序列，根据这些不同频度观测的数据序列，利用方程（4）建立相应的土壤呼吸与环境条件的关系式，基于连续的自动监测环境要素，估算年土壤呼吸量，并以每周2次观测频度得到的年土壤呼吸结果为标准进行对比。结果显示，只要以每2周1次均衡的监测频率（即大约每年24次）便可以得到可靠的年土壤呼吸估算量（与对照相比偏差<5%），如果监测频率低到每月1次，那么就难以获得稳定精确的年土壤呼吸值。但如果想获得多年的土壤呼吸平均值，那么每月1次的监测频率便足以满足。然而，如果采用不均衡的监测模式，比如旱季少于2周1次、雨季增加频度的监测模式等，那么即使是每年监测达到24次，也难以满足年土壤呼吸估算精度的要求。

综上，每两周一次测定就可以满足年土壤呼吸精确测定的要求，每月一次就可以满足多年平均土壤呼吸测定的需求，这个结果为准确测定土壤呼吸并减少工作量提供了方法依据。

该成果已发表在：EUROPEAN JOURNAL OF SOIL BIOLOGY, 52, 41-47. http://www.sciencedirect.com/science/article/pii/S1164556312000635

题目：Optimizing manual sampling schedule for estimating annual soil CO2 efflux in a young exotic pine plantation in subtropical China

作者：Wang Yidong; Wang Huimin; Wang Zhong-Liang; Zhang Wenjiang; Guo Changcheng; Wen Xuefa; Liu Yunfen

期刊：EUROPEAN JOURNAL OF SOIL BIOLOGY 

摘要: Optimizing a manual measurement schedule (both frequency and pattern) for estimating annual soil respiration (R-S) is an important but unresolved issue. We hypothesized that (i) an optimal sampling setup can be found to obtain a reliable annual R-S, and (ii) if the desired outcome is a multi-year mean annual R-S, a lower sampling frequency might be adequate. Here we explored these issues using a three-year chamber-based dataset, with a sampling frequency of twice per week (defined as control), in an exotic slash pine (Pinus elliottii Englem.) plantation in subtropical China. The results showed that R-S during 9:00 -11:00 a.m. represented diurnal mean R-S well. In order to obtain an annual R-S as reliable as the control (deviation within +/- 5%), the optimal measurement strategy is a biweekly sampling across a year and not a trade-off sampling pattern (monthly sampling combined with weekly sampling depending on the seasons). Furthermore, despite an obvious inter-annual variability in R-S (548.4-757.5 g C m(-2) year(-1), CV = 16.3%), a monthly sampling was sufficient to obtain an unbiased multi-year mean annual R-S (deviation within +/- 5%). Such findings are useful when easy looking for estimates of annual ecosystem carbon budgets. However, the generality needs to be examined in other ecosystems.

2012-王义东-EJSOBI2599.pdf