X img

官方微信

img

群号:冰川冻土交流群

QQ群:218834310

高级检索
作者投稿 专家审稿 编辑办公 编委办公 主编办公

冰川冻土 ›› 2021, Vol. 43 ›› Issue (5): 1458-1467.doi: 10.7522/j.issn.1000-0240.2021.0096

• 寒区工程与灾害 • 上一篇    下一篇

青藏工程走廊活动层厚度预测模型与分布特征研究

刘志云1(),黄川1,于晖2,钟振涛1,崔福庆1()   

  1. 1.长安大学 地质工程与测绘学院,陕西 西安 710054
    2.中交第一公路勘察设计研究院有限公司 西安中交公路岩土工程有限责任公司,陕西 西安 710075
  • 收稿日期:2021-07-09 修回日期:2021-09-29 出版日期:2021-10-31 发布日期:2021-09-09
  • 通讯作者: 崔福庆 E-mail:dcdgx33@chd.edu.cn;cfq731@chd.edu.cn
  • 作者简介:刘志云,副教授,主要从事冻土工程热灾害防治研究. E-mail: dcdgx33@chd.edu.cn
  • 基金资助:
    国家自然科学基金项目(51574037);中国交建科技研发项目(2018-ZJKJ-PTJS03)

Study on the prediction model and distribution characteristics of active layer thickness along the Qinghai-Tibet engineering corridor

Zhiyun LIU1(),Chuan HUANG1,Hui YU2,Zhentao ZHONG1,Fuqing CUI1()   

  1. 1.College of Geology Engineering and Geomatics,Chang’an University,Xi’an 710054,China
    2.Xi’an Zhongjiao Highway Geotechnical Engineering Co. Ltd. ,CCCC First Highway Consultants Co. Ltd. ,Xi’an 710075,China
  • Received:2021-07-09 Revised:2021-09-29 Online:2021-10-31 Published:2021-09-09
  • Contact: Fuqing CUI E-mail:dcdgx33@chd.edu.cn;cfq731@chd.edu.cn

摘要:

为探究青藏工程走廊沿线多年冻土区活动层厚度分布情况,结合青藏公路、青藏铁路沿线300个钻孔点的活动层厚度监测数据,基于年平均地表温度、平均植被指数、等效纬度、纬度、高程和含冰量等参数建立了活动层厚度的经验公式、随机森林和径向基函数(radial basis function, RBF)神经网络预测模型。各预测模型结果表明,活动层厚度与各预测因子间具有极强的非线性关系;RBF神经网络预测模型具有最高的预测精确度,拟合优度R2达到0.84。运用RBF神经网络预测模型和高精度遥感数据绘制活动层厚度分布图,分布图显示研究区内活动层厚度主要为2~4 m,总面积为5 468.3 km2,面积占比为47.27%,主要分布于楚玛尔平原至北麓河盆地和唐古拉山区南部至头二九山区;活动层厚度大于4 m次之,总面积为3 382.3 km2,面积占比为29.24%,整体分布偏向南部地区,主要分布于布曲河谷地至头二九山区。并对研究区活动层厚度与含冰量、地温关系进行了研究,结果表明活动层厚度随含冰量增加而减小、随地温升高而增加。

关键词: 青藏工程走廊, 活动层厚度, 含冰量, 随机森林, RBF神经网络

Abstract:

In order to explore the distribution characteristics of active layer thickness (ALT) in permafrost areas along the Qinghai-Tibet engineering corridor, combined with the ALT monitoring data of 300 drilling points along the Qinghai-Tibet Highway and Qinghai-Tibet Railway, Using annual average surface temperature, average vegetation index, equivalent latitude, latitude, elevation and ice content as analysis parameters, prediction models based on the empirical formula, random forest and radial basis function (RBF) neural network method are developed. The results of each prediction model show that the ALT has a strong non-linear relationship with each prediction factor; the RBF neural network prediction model has the highest prediction accuracy, and the goodness of fit (R2reaches 0.84. Utilizing the developed RBF neural network prediction model and high-precision remote sensing data, the ALT distribution map of whole Qinghai-Tibet engineering corridor has been obtained. The distribution map shows that the thickness of active layer in the study area is mainly 2~4 m, with area of 5 468.3 km2, accounting for 47.27% of the total area, and mainly distributes in Chumar River Plain, Beiluhe Basin, Tanggula Mountain, and Touerjiu Mountain; subsequently, the conditions of ALT larger than 4m has the area of 3 382.3 km2, accounting for 29.24% of total, the overall distribution of which is biased towards the southern region and mainly in Buqu River valley to Touerjiu Mountain. Finally, the relations among the ALT, ice content and ground temperature of the study area have been researched. The results show that the ALT decreases with the increase of ice content and increases with the increase of ground temperature.

Key words: Qinghai-Tibet engineering corridor, active layer thickness, ice content, random forest, RBF neural network

中图分类号: 

  • P642.14