基于InVEST模型的森林碳储量动态监测

Dynamic monitoring of forest carbon stocks based on the InVEST model

森林碳储量主要通过地面调查数据来估算,存在着统计工作量大,建模复杂度高等难点。如何快捷、准确地估测森林碳储量一直是国内外林业领域研究的热点和难题。结合遥感图像监测尺度大和InVEST模型输入参数少的特点,提出一种基于InVEST模型结合遥感图像估测森林碳储量的方法。该方法根据森林类型碳储量信息和相应的栅格数据,利用InVEST模型估测区域森林碳储量,然后通过比对多期遥感数据估测出的碳储量得出该区域碳储量变化,从而实现森林碳储量的动态监测。对浙江省庆元县2009年的碳储量进行了估算和绘图,根据行政区划图可估算出乡（镇）和村级的碳储量。实验分别对坑西村2009年和2014年的碳储量进行了估算,根据基于碳库差别的方法实现了碳储量动态监测。实验结果表明：1庆元县2009年的森林碳储量为3.274 3×10~7Mg;2坑西村2009年到2014年碳储量增加了1.780 3×10~4Mg。相比不对森林类型进行细分的森林平均碳密度,将各森林类型平均碳储量和森林类型年度碳汇量引入到森林碳储量监测中,使得森林碳储量的估测精度得到了提升。另外,提出的森林碳储量动态监测可实现对县、乡（镇）和村级的监测,具有模型输入数据量少和输出结果可视化的优点,以及监测手段简单便捷,操作性强等特点。

Forest carbon storage is almost estimated by the ground survey data, there is a difficulty of huge statistical work and modeling complexity. To calculate forest carbon stocks in a quick and accurate way, which has been the current research focus in forestry both inside and outside of China. A new method is proposed by combining a little input parameters of the INVEST model with a large monitoring scale of remote sensing data. First, INVEST model was used to estimate regional carbon stocks according to forest type carbon data and according to raster data. Then this data was compared to the multistage carbon data from remote sensing data to achieve a dynamic monitoring of forest carbon stocks. This paper estimated and mapped carbon stock of the Qingyuan （a county） in 2009, according to the administrative division map can estimate the carbon storage of the scale of town and village. After estimated carbon reserves of Kengxi （a village） in 2009 and 2014, carbon difference method was adopted to dynamic monitoring of carbon sink. Results showed that the carbon sink estimated with the new method for Qiyuan（a county）, Zhejiang, China was 3.274 3×107 Mg in 2009. Also, the carbon stock for Kengxi （a village） increased 1.780 3×104 Mg from 2009 to 2014. Compared to forest average carbon density, adding forest species average carbon storage and carbon sinks to forest carbon stock monitoring improved estimation precision of forest carbon stocks. This dynamic multi-dimensional method for monitoring forest carbon sinks being simple, convenient, user-friendly, and advantageous because it used less data input and visual output in the model, could be used in the county, township （town） and village forest carbon moni- toring. [Ch, 5 fig. 4 tab. 32 ref. l