地理学报 ›› 2019, Vol. 74 ›› Issue (3): 490-503.doi: 10.11821/dlxb201903007

• 气候变化与地表过程 • 上一篇    下一篇

基于蒙特卡洛生存分析探究东北森林物候的影响因素

卓莉(), 张子彦, 雷小雨, 李秋萍, 陶海燕()   

  1. 中山大学地理科学与规划学院 广东省城市化与地理环境空间模拟重点实验室/综合地理信息研究中心,广州 510275
  • 收稿日期:2017-08-02 修回日期:2018-12-23 出版日期:2019-03-25 发布日期:2019-03-19
  • 作者简介:

    卓莉(1973-), 女, 教授, 主要从事资源环境遥感、城市信息、灾害风险分析。E-mail: zhuoli@mail.sysu.edu.cn

  • 基金资助:
    国家自然科学基金项目(41371499)

Monte Carlo survival analysis on the influencing factors of forest phenology in Northeast China

Li ZHUO(), Ziyan ZHANG, Xiaoyu LEI, Qiuping LI, Haiyan TAO()   

  1. Guangdong Provincial Key Laboratory of Urbanization and Geo-simulation, Center of Integrated Geographic Information Analysis, School of Geography and Planning, Sun Yat-sen University, Guangzhou 510275, China
  • Received:2017-08-02 Revised:2018-12-23 Online:2019-03-25 Published:2019-03-19
  • Supported by:
    National Science Foundation of China, No.41371499

摘要:

植被是生态环境变化的指示器,分析植被物候的影响因素不仅有助于气候变化分析,提高区域气候模式的模拟精度,而且对于准确评估植被生长趋势、生产力以及全球碳收支均具有重要意义。基于遥感的植物物候监测已取得了长足的发展和进步,但当前利用大范围、长时间序列的遥感数据分析植被物候影响因素的研究尚不多,采用线性回归模型对非线性的植被物候影响因素进行分析可能存在偏误。因此,本文提出一种基于蒙特卡洛模拟的生存分析方法,对东北森林物候的影响因素进行量化分析。首先利用东北森林地区1982-2009年间AVHRR GIMMS NDVI数据,应用双Logistic曲线拟合方法对植被春季返青期(SOS)、秋季落叶期(EOS)及植被生长期(GSL)进行提取;然后基于蒙特卡洛模拟和生存分析构建植被物候影响因素分析模型;最后运用所构建模型探讨了东北森林区春季返青期、秋季落叶期的可能影响因素。结果发现:温度、降水和风力对中国东北森林关键物候期有一定影响,其中温度是春季返青期和秋季落叶期的最主要驱动因素,长期平均温度比短期内的温度突变对物候影响更显著,落叶期前的风速增加有可能使落叶时间提前;除了环境因素,春季返青早的年间秋季落叶倾向于更晚。研究表明,结合蒙特卡洛方法的生存分析可以较好地对物候期的影响因素进行定量分析,可为物候现象的归因分析提供一种新的方法。

关键词: 生存分析, 蒙特卡洛模拟, 植被物候, 东北森林区, 响应

Abstract:

Quantitative analysis of the influencing factors of vegetation phenology is not only helpful to accurate assessment of impacts of climate change on vegetation but also has great importance in the improvement of regional climate models, as well as accurate estimation of vegetation net primary productivity and carbon balance. Vegetation phenology monitoring based on remote sensing data has made great progress, however, few studies have focused on analyses of the influencing factors of vegetation phenology based on large-scale and time series remote sensing data. The use of the linear regression model in some existing studies has certain limitations due to the nonlinearity of vegetation phenology. In this paper, we propose a Monte Carlo based survival analysis method, which was applied to the forest regions of Northeast China. Start of season (SOS), end of season (EOS) and growing season length (GSL) were firstly extracted from time series AVHRR GIMMS NDVI data of the study area in the period of 1982-2009, using the double logistic curve fitting method. And then the survival analysis model of vegetation phenological influencing factors based on Monte Carlo estimation was constructed. Finally, the proposed method was applied to the forest regions in Northeast China to investigate possible influencing factors of vegetation phenology in the rejuvenation period and deciduous period. Results show that temperature, precipitation, and wind can influence phenology of the forest in the region, with temperature being the primary influencing factor for both start and end of seasons. Long-term changes of average temperature have more significant impacts on the forest phenology, compared with short-term temperature variations. The increase of wind speed before the EOS may lead to an early EOS. In addition to environmental factors, EOS tends to be later if SOS is early. The results also prove that the proposed survival analysis method can provide a good scheme to quantitatively analyze the influencing factors of the phenological periods.

Key words: survival analysis, Monte Carlo, phenology, Northeast Forest Region, Response