过去千年里，人类活动已经显著地改变了欧洲自然景观，尤其是由森林或其他植被类型向农业用地的转变。为了探究因森林砍伐等人类活动引发的土地利用/土地覆盖变化（Land use and land cover change, 简称LUCC）对碳循环以及气候系统的深远影响，绘制历史时期具有精确空间位置的植被分布图是关键步骤。当下具有代表性的历史植被数据集要么是农业用地数据集的衍生产品，要么是基于人口密度数据估算的，而从耕地扩张与森林砍伐关系入手重建历史自然植被分布的研究尚未有之。因此，本研究构建了一种基于现代耕地垦殖率和林地覆盖率空间相关关系的方法重建欧洲历史自然植被分布，主要步骤包括：（1）基于9套现代遥感全球土地覆盖产品中林地（含灌木）分布数据的一致性评价分析结果，制备了一套现代（AD2000）全球30″×30″比例型林地覆盖率数据集；（2）分区（欧洲北部和欧洲中、东南部） 构建现代“耕地垦殖率-林地覆盖率”空间相关关系曲线（空间分辨率为1°×1°）并建立验证数据集对其在时间维度的适用性进行验证；（3）将已收集的历史时期欧洲国家尺度耕地面积数据按照土地宜垦性权重分配到1°×1°的空间像元内；（4）应用本研究重建的“耕地垦殖率和林地覆盖率”相关关系曲线和已分配的欧洲历史时期空间分辨率为1°×1°的耕地垦殖率数据集重建欧洲历史林地覆盖率数据集；（5）在1°格网中扣减掉耕地、林地和其他利用地类型（苔藓和地衣、湿地、裸地、水体、永久冰雪以及城镇用地）得到欧洲历史草地覆盖率数据集。得到的主要结论如下：

（1）构建了基于现代“耕地垦殖率-林地覆盖率”负指数相关关系曲线重建欧洲历史时期林地覆盖率的方法。该负指数关系曲线经过已收集的欧洲国家或亚国家尺度、时间跨度最长可达千年的历史耕地垦殖率和林地覆盖率重建数据验证后，显示其在过去千年时段内同样适用。本研究发现耕地垦殖率和林地覆盖率在欧洲不同地区（欧洲北部和欧洲中、东南部）均呈现出负指数相关关系，但函数系数不同。

（2）基于构建的曲线方程和已收集的历史耕地数据集重建了欧洲中、东南部地区AD1800年以来4个时点的林、草地覆盖率格网数据集。揭示了如下的林、草分布变化特征：AD1800~AD1900年间，阿尔卑斯山以北地区的林地和草地减少较多；直至AD1900年，仅在阿尔卑斯山和巴尔干半岛西部山区依旧保存着部分原始森林和草地；AD1900年以来，法国、德国西部以及波兰中部等地区的林地覆盖率出现回升；阿尔卑斯山以北部分地区的草地覆盖率回升。

（3）重建了德国和法国过去千年12个时点的林草地分布格局，时间间隔为50~100年。重建结果显示德国、法国主要经历了11~13世纪的大规模毁林期、14~15世纪的森林恢复期、16~19世纪的二次毁林期以及20世纪以来的森林恢复期四个阶段；此外，该地区对应时段内的草地覆盖率增加或减少的幅度要远小于林地的变化，但是草地覆盖率出现转折的时间亦与社会经济活跃时期或者重大历史事件（如黑死病）出现的时间相对应。

（4）重建了北欧地区AD1700年以来6个时点的林、草地覆盖率格网数据集。揭示了如下的林、草分布变化特征：北欧的森林砍伐主要发生在南部靠近城市的地区，其北界大约位于瑞典耶夫勒北部边缘地区；草地覆盖率在过去300年内经历了“先增加后降低”的变化过程，20世纪中期以前增加的草地可能来源于森林砍伐，而之后的草地面积减少可能与该时期用于冬季饲料的饲草栽培从牧区转移到具有更高潜力的耕地有关。

（5）过去千年林草地时空变化特征分析表明，本研究重建的林地、草地覆盖时空变化与从史料中获取的林草地变迁定性描述具有较好的一致性；同时重建结果的可靠性较IPCC第五次评估报告采用的Pongratz（2008）重建的全球历史土地利用/覆盖数据集有明显提升。

The natural vegetation has been substantially changed over the last millennium, especially from forest (other natural vegetation) to agricultural areas in Europe. To study the impact of land use and land cover change (LUCC) caused by human activities such as deforestation on the carbon cycle and the climate system, it is essential to construct spatially precise maps in natural vegetation cover. The current representative historical vegetation cover data sets are either derivatives of agricultural datasets or are produced based on the deforestation-population relationship. Almost no attempts have been made to use the inherently linked relationship between cropland expansion and deforestation to reconstruct historical vegetation cover. Therefore, this study developed a novel approach based on the relationship between cropland expansion and deforestation to derive geographically explicit information on forest cover by the following steps: (1) To intergrate a modern (AD2000) global percentage forest data set with a spatial resolution of 30″×30″ based on the results of the consistency evaluation and analysis of forest (including shrubs) data in 9 sets of modern remote sensing global land cover products; (2) To develop the relationship model to simulate the historical forest fraction on a 1° pixel scale based on the modern spatial relationship between cultivation intensity and forest fraction in different regions of Europe (northern Europe and central and southeastern Europe); (3) To generate the historical cropland data allocated from a national scale to a 1° pixel based on the land suitability for cultivation; (4) To renconstruct the historical forest cover using the relationship between cultivation intensity and forest fraction and the historical cultivation intensity dataset with a spatial resolution of 1°×1°; (5) To produce the grassland cover dataset by subtracting cropland, forest and other landuse types (moss and lichens, wetlands, bare land, water bodies, permanent ice and snow, and urban area) on a 1° grid. The results show that:

(1)   This study developed a method for reconstructing historical forest cover in Europe based on the modern negative exponential relationship between cultivation intensity and forest fraction. The negative exponential relationship has a high applicability over the last thousand years after verification by existing cropland cultivation and forest fraction reconstruction results in Europe spanning centuries or even millennia on a national or sub-national scale. The cultivation intensity and forest fraction showed a negative exponential correlation in different region of Europe (northern Europe and central and southeastern Europe), but the functional coefficients were different.

(2)   This study reconstructed a grid dataset of forest and grassland coverage in central and southeastern Europe dating back to AD1800 and revealed the following characteristics of changes: from AD1800 to AD1900, the area of forests and grasslands in the north of the Alps declined severely; in AD1900, only the Alps and the western Balkans still preserved virgin forests and grasslands; since AD1900, forest has recovered in northern France, western Germany, central Poland, and grassland has recovered in the Alps north.

(3)   The forest and grassland cover of Germany and France were reconstructed at 12 time points in the past millennium, with time intervals ranging from 50 to 100 years. The reconstruction results showed that Germany and France went through four major changes: severe deforestation from the 11^th to the 13^th centuries, forest restoration from the 14^th to the 15^th centuries, and second deforestation from the 16^th to the 19^th centuries, forest restoration since the 20th century. During the same time period, the range of change in grassland coverage was much smaller than the range of change in forest area, but the turning point in grassland coverage also corresponds to a period of social and economic activity or the occurrence of major historical events, such as black death.

(4)   This study reconstructed a grid dataset of forest and grassland coverage in Northern Europe since AD1700, revealing the following characteristics of changes: deforestation in northern Europe occurs mainly in the southern regions close to the cities, with the northern boundary approximately at the northern boundary of Gavle in Sweden. Furthermore, the change in grassland coverage in this area over the last 300 years demonstrated a trend of decreasing and then increasing. The increase in grassland before the mid-twentieth century could be attributed to deforestation, while the subsequent decrease in grassland area could be attributed to the shift in forage cultivation for winter fodder from pastoral areas to cropland with higher potential during this time period.

(5)   The analysis of the temporal and spatial change characteristics of reconstructed forest and grassland in this study shows that the temporal and spatial changes of forest land and grassland area are highly consistent with the qualitative understanding of forest and grassland changes obtained from historical records. Simultaneously, the reliability of the reconstruction results is significantly improved compared with the global historical land use/cover dataset reconstructed by Pongratz (2008) used in the IPCC Fifth Assessment Report.