论文部分内容阅读
以探地雷达、电磁测深、钻探等技术方法获得野外数据及数字高程(DEM)遥感数据为基础,通过聚类分析和相关性分析对高程、坡度、坡向等因素对多年冻土分布的影响进行了定量化研究.利用非线性的多元自适应回归样条(MARS)方法建立了基于高程、太阳辐射的多年冻土分布模型,通过自身的交叉验证及对比年平均地温模型和逻辑回归模型的总体分类精度,说明MARS模型具有较好的分类精度.运用MARS模型模拟了整个温泉区域冻土的空间分布特征.结果表明:MARS模型分类精度较高,验证了此模型模拟温泉区域冻土分布的可行性;此模型除了考虑高程对对多年冻土分布的控制作用外,还体现了太阳辐射这一局地综合因素对多年冻土分布的调整作用,较好地模拟了高程相对较低的低山区多年冻土的存在.
Based on the field data and digital elevation (DEM) remote sensing data acquired by the methods of GPR, electromagnetic sounding and drilling, the distribution of permafrost distribution on the basis of elevation, slope and aspect was analyzed by cluster analysis and correlation analysis And the effects of these factors were studied quantitatively.Using the nonlinear Multivariate Adaptive Regression Spline (MARS) method, a model of permafrost distribution based on elevation and solar radiation was established. Through its own cross-validation and comparison with annual average geothermal model and logistic regression model , Which shows that MARS model has better classification accuracy.The MARS model is used to simulate the spatial distribution of permafrost in the whole hot spring area.The results show that the classification accuracy of MARS model is high and the model is validated to simulate the distribution of permafrost in hot spring area This model not only considers the control effect of elevation on the distribution of permafrost, but also reflects the adjustment effect of solar radiation, a comprehensive factor on the distribution of permafrost, and simulates the effect of the relatively low elevation Permafrost in low mountain areas.