土壤是一个巨大的碳库, 土壤呼吸是大气CO2的主要贡献者，故探索森林土壤呼吸动态变化特征，对于正确评价中国陆地生态系统碳收支具有重要的战略意义。但是，传统土壤呼吸监测方法无法对区域范围进行测量，往往以单点监测数据表征整个区域，其结果存在很大的不确定性。针对传统监测方法的弊端，本项目拟以浙江省天目山森林生态系统为示范，以分布式监测网络为数据采集平台，探索自制土壤呼吸仪SFO-V1001的精度控制方法，提高自制监测仪的可靠性；研究基于计算机图像分割算法的布点理论，确定监测点布置位置和布置数量；探索基于BP神经网络算法的空间插值方法，以解决非线性、海量数据空间分布问题；研究基于多尺度分布的森林土壤呼吸时空动态特征分析方法，提高土壤呼吸时空异质性分析与评价的合理性。旨改进森林土壤呼吸空间变异性分析的科学性，为进一步研究森林土壤碳的动态变化及其调控机理、评价中国陆地生态系统碳收支作必要的技术支撑。

Soil is a huge carbon pool and soil respiration is a major contributor of atmospheric CO2,so it is important significance to study the dynamic characteristics of forest soil respiration.But,Traditional monitoring methods of soil respiration are often used a single point to measure regions which may lead to huge errors.Because of the shortcomings of traditional monitoring methods, the project plans to use distributed monitoring networks as data acquisition platform to study Tianmu Mountain forest ecosystems.We study precision control methods of soil respiration monitor SFO-V1001 to improve the reliability of the monitoring device.We study the theory of the placement based on computer image segmentation algorithm to determine the location and number of monitoring points.We study the spatial interpolation method based on BP neural network algorithm to solve nonlinear problems of spatial distribution.We study the temporal and spatial characteristic of forest soil respiration based on multi-scale distribution to improve the rationality of analysis and evaluation in soil respiration.The purpose of the project is to further study the dynamic changes of forest soil carbon and its regulatory mechanisms and to provide technology for evaluating Chinese terrestrial ecosystem carbon budget.