摘要
遥感数据是提取水体信息的常用数据,实际应用中,中、高分辨率空间数据能够以较高精度提取水体信息,但由于时间分辨率较低,难以有效监测水体的动态变化过程.以MODIS(Moderate-resolution Imaging Spectroradiometer)为代表的低空间分辨率遥感数据时间分辨率高,在全球大范围快速提取水体及水体动态变化监测中具有重要意义,但是受其空间分辨率较低的限制,提取的水体信息精度不高.针对该问题,该文提出了一种针对MODIS数据的水体超分辨率制图方法.该方法首先利用MODIS数据MOD09Q1产品的1、2波段,通过线性混合像元分解,得到水体丰度图;然后利用MODIS数据MOD09A1产品的2、4波段,计算得到水体归一化指数,对水体丰度图进行后处理以减少混合像元分解误差;最后以修正后的水体丰度图为输入,建立超分辨率制图模型,从而得到最终的高空间分辨率水体分布图.将上述方法应用于丹江口水库,并与利用Landsat-8/OLI数据提取的水体分布图进行对比,结果表明,相对于传统像元尺度的水体制图结果而言,通过超分辨率制图得到的水体制图结果精度更高,不仅能够更好的保留水体空间分布特征,提取的水体面积也更加准确.基于MODIS数据利用超分辨率制图方法对丹江口水库进行水体提取,完成了2014年12个月丹江口水体面积变化图.超分辨率制图可以获得具有Landsat空间分辨率和MODIS时间分辨率的水体制图结果,实现对丹江口水库的高时空分辨率的水域变化监测.
Remote Sensing images have been widely used for water body mapping.High spatial resolution images can map water body with a high accuracy;however,its low temporal resolution makes it unable to monitor the dynamic change of water bodies effectively.High temporal resolution Remote Sensing imagery,such as MODIS(Moderate-resolution Imaging Spectroradiometer),is important for rapid water bodies monitoring in global range.However,it is generally difficult to obtain water body boundaries precisely due to the low spatial resolution.To address this problem,in the current work,a water extraction method is proposed from MODIS imagery based on super-resolution mapping.Firstly,the water fraction image was estimated by linear mixing using the 1st and 2nd bands of MODIS.Then,the fraction noise caused by hill-shadow is reduced through the Normalized Difference Water Index(NDWI),which is produced from the 2nd and 4th bands.Finally,with the revised water fraction image as input,the final high spatial resolution water body map was generated.Taking the Danjiangkou Reservoir area as an example,the proposed method was assessed by comparison with traditional pixel-based mapping methods.The results show that more detailed spatial information and accurate area of water body are obtained in the water body map extracted from the super-resolution mapping algorithm.The water area changing graph of Danjiangkou reservoir in 2014 is completed based on MODIS data with the above method.These indicate that super-resolution mapping is able to generate water mapping results with a spatial resolution of Landsat and time resolution of MODIS and achieve the goal of monitoring changes of Danjiangkou Reservoir in high spatial and temporal resolution.
出处
《华中师范大学学报(自然科学版)》
CAS
北大核心
2016年第4期624-632,共9页
Journal of Central China Normal University:Natural Sciences
基金
国家自然科学基金青年基金项目(41301398)
湖北省杰出青年基金项目(2013CFA031)
中国科学院测量与地球物理研究所方向项目子课题(Y409123012)
关键词
超分辨率制图
水体提取
空间相关性
遥感
丹江口水库
super-resolution
water extraction
spatial correlation
Remote Sensing
Danjiangkou Reservoir