基于物候分段同化的水稻重金属胁迫遥感监测时间尺度优化

发布时间：2018-02-11 11:21

本文关键词： 水稻重金属胁迫物候提取 WOFOST模型分段同化时间尺度优化　出处：《中国地质大学(北京)》2017年硕士论文　论文类型：学位论文

【摘要】：水稻受到重金属胁迫后,生理生态参数会发生变化。利用遥感影像提取水稻关键物候期进行分段同化,研究水稻不同生长阶段对重金属胁迫的不同响应,通过Harris算法对遥感-作物生长模型同化框架进行时间尺度的优化,提取同化最优时间点,在保证模型模拟精度条件下提高模型运算效率。本文以湖南省株洲市重金属污染程度不同的稻田为研究区,利用2013年的16幅HJ-1A/B卫星CCD数据及田间实测数据进行实验研究。使用HJ-1A/B遥感数据构建EVI时间序列数据,利用非对称性高斯函数法对EVI时间序列进行平滑去噪处理,探寻水稻物候期的转变与EVI时间序列的响应关系,从而提取水稻的物候期。在WOFOST模型中加入胁迫因子f,以f来表征重金属胁迫下水稻生理生态参数的响应机制。依据上述研究,对WOFOST模型进行本地化和区域化扩展,结合所提取的水稻关键物候期,根据DVS的取值来判定作物生长发育所处的阶段,并且基于PSO算法,构建区域尺度下遥感-作物生长模型的分段同化模拟框架,输出不同生长阶段的胁迫因子f,驱动优化后的WOFOST模型对研究区内的水稻根重进行逐像元模拟,利用胁迫因子f和模拟根重来评价遥感-作物生长模型的分段同化模型。对分段同化模型的时间尺度进行研究,基于Harris角点算法,对LAI时序曲线进行灰度转换进而提取角点,为了验证提取的4个角点的准确性,对角点进行线性拟合,角点拟合曲线与LAI时序曲线的R2为0.89,角点拟合曲线与LAI比值曲线的R2为0.78,能够反映出原曲线的主要信息。根据4个角点所对应的日序的CCD影像提取LAI,驱动遥感-作物生长模型同化,与分段同化相对比发现,角点所对应的同化框架时间效率大大提高,并且对根重的模拟精度在95%以上。研究表明,基于物候期的分段同化能够提高遥感-作物生长模型的模拟精度,在此基础上进行角点检测,能够在保证遥感-作物生长模型同化精度的情况下提高同化模型的运行速率。
[Abstract]:The physiological and ecological parameters of rice were changed after heavy metal stress. The key phenological stages of rice were extracted from remote sensing images for segmental assimilation to study the different responses of rice at different growth stages to heavy metal stress. The Harris algorithm is used to optimize the time scale of the assimilation frame of the remote sensing crop growth model, and the optimal time point of assimilation is extracted. The model operation efficiency is improved under the condition of ensuring the model simulation accuracy. In this paper, the paddy fields with different pollution degree of heavy metals in Zhuzhou City, Hunan Province, are taken as the research area. Based on the 16 HJ-1A/B satellite CCD data from 2013 and the field measured data, the EVI time series data were constructed by using HJ-1A/B remote sensing data, and the EVI time series was smoothed by using asymmetric Gao Si function method. To explore the response relationship between phenological changes of rice and EVI time series, the phenological period of rice was extracted. The response mechanism of physiological and ecological parameters of rice under heavy metal stress was characterized by adding the stress factor f into the WOFOST model and using f to characterize the physiological and ecological parameters of rice under heavy metal stress. Based on the localization and regionalization of WOFOST model and the key phenological period of extracted rice, the stage of crop growth and development is determined according to the value of DVS, and based on PSO algorithm, A segmented assimilation simulation framework of remote sensing and crop growth model at regional scale was constructed, and the stress factors f in different growth stages were output. The optimized WOFOST model was used to simulate the root weight of rice in the study area by pixel by pixel. The stress factor f and simulated root weight are used to evaluate the segmented assimilation model of remote sensing crop growth model. The time scale of the model is studied. Based on the Harris corner algorithm, the gray-scale transformation of the LAI time series curve is carried out and the corner points are extracted. In order to verify the accuracy of the extracted four corners, the corners are fitted linearly. The R2 of corner fitting curve and LAI time series curve is 0.89, and that of corner fitting curve and LAI ratio curve is 0.78, which can reflect the main information of original curve. Assimilation of crop growth models, Compared with subsection assimilation, it is found that the time efficiency of the assimilation frame corresponding to corner points is greatly improved, and the simulation accuracy of root weight is more than 95%. Phenophase based subsection assimilation can improve the simulation accuracy of the remote sensing crop growth model. On the basis of this, corner detection can improve the operating rate of the remote sensing crop growth model under the condition of ensuring the assimilation accuracy of the remote sensing crop growth model.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X87;X56

【相似文献】