基于多源遥感数据的云南省九大高原湖泊水质反演

发布时间：2018-01-09 03:09

本文关键词：基于多源遥感数据的云南省九大高原湖泊水质反演　出处：《西北大学》2017年硕士论文　论文类型：学位论文

【摘要】：云南省九大高原湖泊(以下简称“九湖”)在云南省社会经济发展中具有不可替代的作用。随着九湖流域内工业化进程加快、城市人口增长和居民生活方式的变化,湖泊污染问题日益严重,因此,对九湖的湖泊环境监测越来越受到重视,其中对湖泊的水质最为关注。本研究以2010-2014年地面分析的水质数据为基础,分别以MODIS卫星影像(464景)和HJ-1B影像(856景)为遥感数据源,建立了反演模型,并通过相关系数(R)、均方根误差(RMSE)以及效率系数(NSE)进行模型验证,从而实现了对九湖水质参数(叶绿素a、总氮、透明度以及藻类总细胞密度)的反演。本研究得出的主要结果如下:(1)以MODIS为数据源,共建立了 153组叶绿素a浓度、163组总氮、146组透明度和128组藻类总细胞密度反演模型,利用R2、RMSE、NSE综合优选出33个叶绿素a浓度、34个总氮、35个透明度和34个藻类总细胞密度组合模型;以HJ-1B影像为数据源,共建立了 86组叶绿素a浓度、77组总氮、86组透明度和86组藻类总细胞密度反演模型,优选出9个叶绿素a浓度、7个总氮、9个透明度和9个藻类总细胞密度组合模型。(2)对比MODIS影像和HJ-1B影像的九湖水质反演模型发现,两种数据源的九湖反演模型的最佳波段组合并没有明显规律,但反演效果相差不大,这说明建立的统计模型与光谱信息的关系并不明显。HJ-1B影像虽然空间分辨率高,但其受云量覆盖等因素的影响更大,使得满足反演要求的数据连续性较差,难以对九湖水质参数进行连续反演。因此,本研究对九湖水质参数的长期变化进行分析时以MODIS数据为主要数据源。(3)将九湖的水质参数反演优选模型应用到2010—2014年的MODIS遥感影像中,共获得2070幅叶绿素a浓度反演图、1840幅总氮反演图、1840幅透明度反演图和1840幅藻类总细胞密度反演图。将优选模型应用到HJ-1B遥感影像中,共得出1840幅叶绿素a浓度反演图、1380幅总氮反演图、1840幅透明度反演图和1840幅藻类总细胞密度反演图。(4)挑选与地面采样时间相近时段的九湖水质参数空间分布图进行分析,结果表明:获取的九湖水质参数空间分布比较符合专业人员的判断,结果较为合理。从空间分布规律看,滇池、洱海、程海、杞麓湖、星云湖、异龙湖、抚仙湖和阳宗海的水质参数基本呈现周围差,中部好的规律,泸沽湖则为周围好,中部低。(5)从年内变化上看,地面分析水质资料和遥感反演水质均表现出一致的变化趋势:九湖的水质参数最大值基本都出现在6—7月份之间,最小值在2—3月份之间,并呈现先增加后减少的趋势;从年际变化上看,抚仙湖、泸沽湖、程海水质变化较为平稳,滇池、阳宗海、洱海水质有好转趋势,异龙湖、星云湖水质略有好转,杞麓湖水质有恶化趋势。总之,采用“一湖一策,分时段”分别建立优选模型,能够较好地模拟出水质参数的空间分布和动态变化特征。
[Abstract]:Nine plateau lakes in Yunnan province (hereinafter referred to as the "nine Lake") plays an irreplaceable role in the social and economic development in Yunnan province. With the nine Lake Basin in the accelerated process of industrialization, population growth and changes of city residents, the lake pollution problem is serious, therefore, the lake environment monitoring is becoming more and more of the nine lake the water quality of Lake attention, most attention. This research is based on the analysis of ground water quality data in 2010-2014, respectively by MODIS satellite images (464 scenes) and HJ-1B images (856 scenes) for remote sensing data source, established inverse model, and the correlation coefficient (R), root mean square error (RMSE) and the efficiency coefficient (NSE) to validate the model, so as to achieve the nine lake water quality parameters (chlorophyll a, total nitrogen, total transparency and algae cell density inversion). The main results obtained in this study are as follows: (1) with MODIS as the data source, and The 153 group of chlorophyll a concentration, 163 groups of total nitrogen, 146 group and 128 group transparency of algal cell density inversion model, using R2, RMSE, NSE selected 33 chlorophyll a concentration, total nitrogen 34, 35 and 34 of total transparency of algae cell density combined with HJ-1B images as data model; the source, there are 86 groups of chlorophyll a concentration, 77 groups of total nitrogen, 86 group and 86 group transparency of algal cell density inversion model, selected 9 chlorophyll a concentration, total nitrogen 7, 9 and 9 of total transparency algal cell density combination model. (2) found that nine of the lake water quality inversion model image and HJ-1B image contrast MODIS, optimum band combination nine Lake inversion model of the two kinds of data sources and the rule is not obvious, but the inversion effect is similar, which illustrate the relationship between statistical model and spectral information of.HJ-1B image is not obvious while high spatial resolution, but the cloud cover The influence factors are more, so as to satisfy the requirements of the inversion data continuity is poor, difficult to carry out continuous inversion of water quality parameters of Lake nine. Therefore, the long term variation of water quality parameters of nine lakes are analyzed in this paper using MODIS data as the main data source. (3) the water quality of Lake nine parameters inversion optimization model application the MODIS remote sensing images from 2010 to 2014, a total of 2070 pieces of chlorophyll a concentration inversion map, 1840 total nitrogen inversion map, 1840 pieces and 1840 pieces of transparency inversion map of algal cell density inversion map. Apply optimization model to HJ-1B remote sensing images, 1840 images of chlorophyll a concentration inversion map were obtained. 1380 total nitrogen inversion map, 1840 pieces and 1840 pieces of transparency inversion map of algal cell density inversion map. (4) selection and sampling time periods similar to ground water quality parameters of nine Lake spatial distribution were analyzed, the results showed that: nine the water quality of Lake ginseng Compared with the spatial distribution of the number of professional judgment, results are reasonable. From the spatial distribution, Dianchi, Erhai, Chenghai, Qilu Lake, Xingyun Lake, from Longhu, Fuxian Lake and Yangzonghai water quality parameters showed around the poor, Central good law, Lugu Lake is around good, middle and low. (5) from the annual change on the ground, analysis of water quality data and water quality remote sensing inversion showed the same trend: the maximum value of nine water quality parameters of Lake basically appear in 6 and July, the minimum value in 2 and March, and showed a decreasing trend after the first increase; from the annual variation at, Fuxian Lake, Lugu Lake, sea water quality change is relatively stable, Dianchi, Yangzonghai, Erhai water quality has improved trend, from Longhu, nebula water quality improved slightly, Qilu water quality has deteriorated. In short, the "a lake a policy optimization models were built in time, can It is good enough to simulate the spatial distribution and dynamic change characteristics of the water quality parameters.

【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X87;X524

【相似文献】