X img





作者投稿 专家审稿 编辑办公 编委办公 主编办公

冰川冻土 ›› 2013, Vol. 35 ›› Issue (1): 74-83.doi: 10.7522/j.issn.1000-0240.2013.0009

• 冰冻圈与全球变化 • 上一篇    下一篇

基于Wavelet-ANFIS和MODIS地表温度产品的青藏高原0 cm土壤温度估算方法

黄培培1, 南卓铜1,2   

  1. 1. 中国科学院 寒区旱区环境与工程研究所, 甘肃 兰州 730000;
    2. 中国科学院 寒区旱区环境与工程研究所 冻土工程国家重点实验室, 甘肃 兰州 730000
  • 收稿日期:2012-08-17 修回日期:2012-12-29 出版日期:2013-02-25 发布日期:2013-07-22
  • 通讯作者: 南卓铜,E-mail:nztong@lzb.ac.cn E-mail:nztong@lzb.ac.cn
  • 作者简介:黄培培(1986-),女,河南永城人,2009年毕业于河南大学,现为硕士研究生,主要从事于GIS、RS的应用研究.E-mail:huangyingbing@yeah.net
  • 基金资助:

    冻土工程国家重点实验室开放基金项目(SKLFSE201009); 科技基础性工作专项(2008FY110200)资助

Estimation of 0-cm Soil Temperature over the Tibetan Plateau Based on the Wavelet Analysis and Adaptive Network-fuzzy Inference System

HUANG Pei-pei1, NAN Zhuo-tong1,2   

  1. 1. Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou Gansu730000, China;
    2. State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou Gansu 730000, China
  • Received:2012-08-17 Revised:2012-12-29 Online:2013-02-25 Published:2013-07-22


0 cm土壤温度是冻土模型的上边界条件, 连续的、 高质量的青藏高原0 cm土壤温度数据是进行准确冻土模拟的必要条件. 然而受复杂下垫面的影响, 遥感手段无法获取可靠的0 cm土壤温度. 利用自适应网络模糊推理系统(ANFIS)结合青藏高原实测资料建立遥感地表温度产品(LST)与0 cm土壤温度的关系, 以实现通过LST估算青藏高原逐日0 cm土壤温度. 研究了ANFIS的各种参数组合, 发现筛选合适的小波函数、 小波窗口、 小波层数建立起来的Wavelet-ANFIS模型能较准确实现估算0 cm土壤温度的目的. 验证表明, 估算结果与气象站点实测0 cm土壤温度绝对误差在2 K以下, 相关系数0.98以上. 考虑到原始MODIS LST误差在0~2 K之间, 该方法可以获取较为理想的0 cm土壤温度, 为冻土模型提供准确的上边界输入.

关键词: 小波分析, 自适应网络模糊推理系统, MODIS地表温度产品, 青藏高原, 0 cm土壤温度


0-cm soil temperature is the upper boundary condition of many permafrost models. Continuous, high-quality 0-cm soil temperature data are necessary inputs to simulate permafrost distribution. However, owing to the influence of complex underlying surface, remote sensing approaches cannot provide reliable 0-cm soil temperature. In this study, in order to estimate 0-cm soil temperature, adaptive network-fuzzy inference system (ANFIS) combining with the data measured in the Tibetan Plateau is used to establish the relations between remote sensing land surface temperature (LST) and 0-cm soil temperature. In this paper, different parameter combinations of ANFIS are examined, and a Wavelet-ANFIS model established by optimized wavelet functions, wavelet windows and wavelet layers is found able to estimate the 0-cm soil temperature more accurately. A comparison analysis of the estimated results and the 0-cm soil temperatures measured at the meteorological sites shows that the approach can achieve desirable estimation with an absolute error less than 2 K and a correlation coefficient greater than 0.98. In view of the original MODIS LST error range from 0 to 2 K, the proposed method may provide more accurate 0-m soil temperature inputs to permafrost models.

Key words: wavelet analysis, adaptive network-fuzzy inference system (ANFIS), MODIS land surface temperature product, Tibetan Plateau, 0-cm soil temperature


  • P407.8