基于物候特征的盐渍化信息数据挖掘研究

发布时间：2018-04-01 09:43

本文选题：盐渍化　切入点：物候信息　出处：《生态学报》2017年09期

【摘要】：盐渍化是影响植被和作物长势的重要因素,精确反演盐渍化的时空分布信息至关重要。基于MOD13A1-NDVI数据反演生长季开始日期(SOS)、生长季结束日期(EOS)、生长季长度(LEN)等物候参数和计算出能高精度反演盐渍化空间分布的多种植被指数、盐分指数、地形指数、干旱指数等参数后作为BP-ANN人工神经网络的输入因子来反演盐渍化信息,同时按照植被类型和地貌类型进行分区来反演盐渍化信息,以探讨盐渍化受植被和地貌类型的影响。主要结论如下:(1)盐渍化的形成受多种因素的影响,与物候参数大多呈非线性关系,不能单纯的以某拟合公式来进行表达,需要借助人工神经网络超强的非线性拟合能力来反演盐渍化信息。(2)通过深入挖掘植被物候信息,在融入物候参数后的反演精度显著提高。可决系数R2从0.68(非物候参数)增加到0.79(包括物候参数),但是需要加入地形、影像数据和土壤水分等方面的信息来更加精确的反演盐渍化信息。生物累积量指标LSI(Large seasonal integral)和SSI(Small seasonal integral)能够很好的表征盐渍化的信息。(3)划分植被类型后的盐渍化提取精度进一步提高,可决系数R~2达到了0.88。(4)以地貌特征作为类型分区后,反演结果的R~2达到了0.85,精度较高,比以植被类型作为分区的精度略小。高程较低区域的盐渍化现象普遍较重,盐渍化程度受到地形和地貌因素的影响显著。(5)农用地区域多为非盐渍化和轻度盐渍化地,稀疏植被区多为重盐渍化地。研究区的非盐渍化和轻盐渍化地、中盐渍化地和重度盐渍化地比例分别为53.42%,13.71%,32.87%。以上的研究结果提出了一种融合物候信息和非物候参数来反演盐渍化信息的方法,进行深入的协同植被物候监测盐渍化信息方面的数据挖掘,在融入了物候参数后,盐渍化的预测精度显著提高。
[Abstract]:Salinization is an important factor affecting vegetation and crop growth, the accurate inversion of salinization distribution information is very important. MOD13A1-NDVI data inversion of the growing season start date based on (SOS), end of growing season (EOS), Ji Changdu (LEN) and other phenological parameters and calculated to grow high precision space by inversion of the distribution of salinization index. Salt index, topographic index, drought index and other parameters as the input factor of BP-ANN artificial neural network to retrieve the information at the same time according to the vegetation types of salinization, and landform zoning to retrieve the salinization information, to investigate the salinization influenced by vegetation and landscape types. The main conclusions are as follows: (1) is affected by many factors of salinization mostly, a non-linear relation with the phenological parameters, not simply a fitting formula to express, need the help of non line super artificial neural network Fitting ability to retrieve the salinization information. (2) through in-depth excavation of vegetation information significantly improve retrieval accuracy with phenological parameters after. The coefficient of determination of R2 from 0.68 (non phenological parameters (including) increased to 0.79, but the phenology parameters) to image data and terrain, soil moisture and other aspects of the information information retrieval salinization more accurately. The biomass accumulation index of LSI (Large seasonal integral) and SSI (Small seasonal integral) to characterize the salinization information very well. (3) classification of salinization extraction accuracy of vegetation types after further improvement, the coefficient of determination R~2 reached 0.88. (4) with geomorphic features as the type partition the inversion results, R~2 reached 0.85, higher accuracy, than to vegetation types as smaller precision partition. A low elevation area of salinization is generally heavier, salinization by terrain and landform. The effect is significant. (5) for the non agricultural area and mild salinity, sparse vegetation area is heavily salinized land. The study area of salinization and light saline soil, salt salinization and severe salinity respectively 53.42%, 13.71%, 32.87%. of the above results and put forward a method to retrieve the salinization information fusion phenological information and non phenological parameters, the collaborative monitoring of salinization information of vegetation phenology data on the depth of excavation, in the integration of the phenological parameters, the prediction accuracy of salinization increased significantly.

【作者单位】：新疆大学资源与环境科学学院;新疆大学绿洲生态教育部重点实验室;
【基金】：新疆维吾尔自治区重点实验室专项基金(2016D03001,2014KL005) 新疆维吾尔自治区科技支疆项目(201591101) 2014级新疆大学博士生科技创新项目(XJUBSCX-2014013) 国家自然科学基金项目(U1303381,41261090,41161063) 教育部促进与美大地区科研合作与高层次人才培养项目
【分类号】：S156.41

【相似文献】