以青藏高原为研究区,采用数字土壤制图方法,进行了少量调查样本支持下的永冻土活动层厚度预测指标集的建立及制图研究。利用土壤和景观环境之间关系,筛选建立了活动层厚度预测指标集:地表昼夜温差、海拔、坡度、坡向、归一化植被指数(NDVI)、母岩。利用样点个体代表性方法实现了该区永冻土活动层厚度分布制图,取得了较高的精度和分辨率,克服了永冻土活动层厚度模拟过程中数学物理模型难以进行空间扩展制图以及半经验模型制图分辨率粗的局限。针对调查样点全局代表性较差以及数量有限的局限,本研究没有采用常规的一次性全样本最优建模方式,而采取多次抽样分别建模方式,获得预测指标集和制图结果。
In the Qinghai-Tibet Plateau, a small number of soil samples were collected during surveys of permafrost for the study on building up an index set for predicting thickness of the active layer in the permafrost and for mapping using the digital soil mapping method. By taking into account the relationships between soil and landscape environment, an index set was screened out and established, encompassing, diurnal surface temperature variation, altitude, slope gradient, slope aspect, NDVI and parent rock. With the aid of the method of setting individual samples as representative, a permafrost active layer thickness distribution map of the region was plotted with quite satisfying accuracy and resolution, thus overcoming the difficulty in space extending mapping in the processes of simulating thickness of the active layer of the permafrost with mathematic-physical models and breaking through the limitation of the use of semi-empirical models in mapping being low in resolution. In view of the problem that the sampling sites in the survey were quite limited in number and not so representative of the whole region, the method of modeling based on multi- samplings separately was adopted, instead of the conventional one-time full samples method. Consequently a predictive index set and a map was acquired.