Reliability of net ecosystem productivity estimation in an Alaskan boreal forest from satellites

我们把西伯利亚东部的落叶松林和阿拉斯加的云杉林作为永久冻土上的代表性森林。落叶松和云杉分别占西伯利亚东部和阿拉斯加面积的80%和70%。我们估计净生态系统生产力（NEP）和森林火灾对NEP的影响，使用地球观测卫星上的传感器。我们评估了森林的碳预算，没有受到森林火灾，阿拉斯加黑森林火灾的严重程度，以及西伯利亚东部的树冠和地表火灾。基于苔藓和森林组成的实地光谱观测，我们建立了一种方法来估计森林地面植被以及冠层净初级生产力（NPP）和叶面积指数的遥感数据，我们应用到阿拉斯加云杉和西伯利亚落叶松森林。我们估计冠层叶面积指数和森林地面植被组成。基于这些关系和野外观测，我们评估了冠层叶面积指数和冠层NPP，森林植被组成和森林地面NPP之间的关系，和NEP的地理分布。此外，我们还测量了森林地面的光谱反射率，包括苔藓，地衣和烧伤疤痕，以评估火灾后烧伤疤痕和植被恢复的影响。在阿拉斯加黑云杉林中，藓类植物的面积分数是由50 m × 50 m面积上平均的绿色和红色带的比值用线性方程估算的（决定系数R^2=0.6）。由于苔藓植物的大部分保留下来，云杉的恢复将占主导地位。由于苔藓的一小部分仍然存在，阔叶树，如桦树将占主导地位。估计的残存藓类植物的面积分数是估计火灾后NEP的主要指标。

We focused on a larch forest in eastern Siberia and a spruce forest in Alaska as representative forests that exist on permafrost. Larch and spruce occupies 80% and 70% of the area in eastern Siberia and Alaska, respectively. We estimated net ecosystem productivity (NEP) and forest fire effects on the NEP using sensors onboard earth observation satellites. We evaluated carbon budgets of forests that were not damaged by forest fires, fire severity of Alaskan black forest, and crown and surface fires in eastern Siberia. Based on field spectral observations of mosses and forest component, we built a methodology for estimating forest floor vegetation as well as canopy net primary productivity (NPP) and leaf area index from the remotely sensed data, and we applied to Alaskan spruce and Siberian larch forests. We estimated canopy leaf area index and forest floor vegetation composition. Based on these relationships and field observations, we evaluated the relationship between canopy leaf area index and canopy NPP, forest floor vegetation composition and forest floor NPP, and geographical distribution of NEP. Further, we measured spectral reflectances of forest floor including mosses, lichens, and burnt scars to evaluate effects of burnt scars and vegetation recovery after the fire. In the Alaskan black spruce forest, fractional area of mosses were estimated from the ratio of green and red bands averaged in 50m×50m area using a linear equation (coefficient of determination R^2=0.6). As the many portion of the mosses remained, recovery of spruce will be predominant. As small part of mosses remained, broad leaf trees such as birch will be predominant. The estimated fractional area of remained mosses was the main index for estimating NEP after the fire.