归一化植被指数(normalized difference vegetation index,NDVI)时间序列已广泛应用于植被信息提取研究,然而目前NDVI时间序列的研究主要集中于中低分辨率遥感影像,从而影响了植被信息提取的精度。随着中国高分专项首颗卫星高分一号(GF...归一化植被指数(normalized difference vegetation index,NDVI)时间序列已广泛应用于植被信息提取研究,然而目前NDVI时间序列的研究主要集中于中低分辨率遥感影像,从而影响了植被信息提取的精度。随着中国高分专项首颗卫星高分一号(GF-1)的发射,为高分辨率NDVI时间序列的构建提供了可能。该文尝试利用GF-1卫星16 m宽覆盖(wide field of view,WFV)影像,构建16 m分辨率NDVI时间序列,以河北省唐山市南部区域为研究区,开展作物分类研究。该文采用覆盖作物完整生长期的GF-1数据构建NDVI时间序列,避免了利用自然年(1-12月)数据构建NDVI时间序列的不足,有助于作物信息的提取。通过分析样地的NDVI时序曲线,发现GF-1/WFV NDVI时间序列能够清晰地区分不同作物的物候差异,捕捉作物特有的生长特性,而且能够识别研究区当年的作物种植模式。该文分别采用最大似然法、马氏距离、最小距离、神经网络分类、支持向量机(support vector machine,SVM)等分类方法,基于GF-1/WFV NDVI时间序列对研究区作物进行分类,研究结果表明SVM分类方法总体精度最高,达到96.33%。同时该文还采用时间序列谐波分析法(harmonic analysis of time series,HANTS)对NDVI时间序列进行了平滑处理,结果表明处理后的NDVI时间序列能更好地描述作物的物候特性,作物分类精度得到进一步提高。展开更多
以国产高分一号(GF-1)宽幅数据(wide field of view,WFV)为数据源,采用简单生物圈模型2(simple biosphere model2,SiB2)对黑龙江省漠河县森林植被叶面积指数(leaf area index,LAI)进行估算,并与增强植被指数(enhanced vegetation index,...以国产高分一号(GF-1)宽幅数据(wide field of view,WFV)为数据源,采用简单生物圈模型2(simple biosphere model2,SiB2)对黑龙江省漠河县森林植被叶面积指数(leaf area index,LAI)进行估算,并与增强植被指数(enhanced vegetation index,EVI)线性模型的估算结果进行对比,结合地面实测LAI数据分别对这2种模型估算结果进行精度评价。结果表明,采用EVI线性模型估算LAI,决定系数R 2为0.582,均方根误差(root mean square error,RMSE)为0.701;而采用SiB2模型估算LAI,R 2为0.798,RMSE为0.358,均比EVI线性模型有所改善。该研究发现,结合中高空间分辨率的GF-1 WFV数据,SiB2模型更适宜于该研究区森林植被的LAI反演。展开更多
Using simultaneously collected remote sensing data and field measurements, this study firstly assessed the consistency and applicability of China high-resolution earth observation system satellite 1 （GF-1） wide fiel...Using simultaneously collected remote sensing data and field measurements, this study firstly assessed the consistency and applicability of China high-resolution earth observation system satellite 1 （GF-1） wide field of view （WFV） camera, environment and disaster monitoring and forecasting satellite （H J-l） charge coupled device （CCD）, and Landsat-8 opera- tional land imager （OLI） data for estimating the leaf area index （LAI） of winter wheat via reflectance and vegetation indices （VIs）. The accuracies of these LAI estimates were then assessed through comparison with an empirical model and the PROSAIL radiative transfer model. The effects of radiation calibration, spectral response functions, and spatial resolution on discrepancies in the LAI estimates between the different sensors were also analyzed. The results yielded the following observations： （1） The correlation between reflectance from different sensors is relative good, with the adjusted coefficients of determination （R2） between 0.375 to 0.818. The differences in reflectance are ranging from 0.002 to 0.054. The correlation between VIs from different sensors is high with the R2 between 0.729 and 0.933. The differences in the VIs are ranging from 0.07 to 0.156. These results show the three sensors＇ images can all be used for cross calibration of the reflectance and VIs. （2） The four VIs from the three sensors are all demonstrated to be highly correlated with LAI （R2 between 0.703 and 0.849）. The linear models associated with the 2-band enhanced vegetation index （EVI2）, which feature the highest R2 （higher than 0.746） and the lowest root mean square errors （RMSE） （less than 0.21）, were selected to estimate the winter wheat LAI. The accuracy of the estimated LAI from Landsat-8 was the highest, with the relative errors （RE） of 2.18% and an RMSE of 0.13, while the H J-1 was the lowest, with the RE of 2.43% and the RMSE of 0.15. （3） The inversion errors in the different sensors＇ LAI estimates using the PROSAIL model are small. The accuracy of the GF-1 is the highest with the RE of 3.44%, and the RMSE of 0.22, whereas that of the H J-1 is the lowest with the RE of 4.95%, and the RMSE of 0.26. （4） The effects of the spectral response function and radiation calibration for the different sensors are small and can be ignored, but the effects of spatial resolution are significant and must be taken into consideration in practical applications.展开更多
文摘归一化植被指数(normalized difference vegetation index,NDVI)时间序列已广泛应用于植被信息提取研究,然而目前NDVI时间序列的研究主要集中于中低分辨率遥感影像,从而影响了植被信息提取的精度。随着中国高分专项首颗卫星高分一号(GF-1)的发射,为高分辨率NDVI时间序列的构建提供了可能。该文尝试利用GF-1卫星16 m宽覆盖(wide field of view,WFV)影像,构建16 m分辨率NDVI时间序列,以河北省唐山市南部区域为研究区,开展作物分类研究。该文采用覆盖作物完整生长期的GF-1数据构建NDVI时间序列,避免了利用自然年(1-12月)数据构建NDVI时间序列的不足,有助于作物信息的提取。通过分析样地的NDVI时序曲线,发现GF-1/WFV NDVI时间序列能够清晰地区分不同作物的物候差异,捕捉作物特有的生长特性,而且能够识别研究区当年的作物种植模式。该文分别采用最大似然法、马氏距离、最小距离、神经网络分类、支持向量机(support vector machine,SVM)等分类方法,基于GF-1/WFV NDVI时间序列对研究区作物进行分类,研究结果表明SVM分类方法总体精度最高,达到96.33%。同时该文还采用时间序列谐波分析法(harmonic analysis of time series,HANTS)对NDVI时间序列进行了平滑处理,结果表明处理后的NDVI时间序列能更好地描述作物的物候特性,作物分类精度得到进一步提高。
文摘以国产高分一号(GF-1)宽幅数据(wide field of view,WFV)为数据源,采用简单生物圈模型2(simple biosphere model2,SiB2)对黑龙江省漠河县森林植被叶面积指数(leaf area index,LAI)进行估算,并与增强植被指数(enhanced vegetation index,EVI)线性模型的估算结果进行对比,结合地面实测LAI数据分别对这2种模型估算结果进行精度评价。结果表明,采用EVI线性模型估算LAI,决定系数R 2为0.582,均方根误差(root mean square error,RMSE)为0.701;而采用SiB2模型估算LAI,R 2为0.798,RMSE为0.358,均比EVI线性模型有所改善。该研究发现,结合中高空间分辨率的GF-1 WFV数据,SiB2模型更适宜于该研究区森林植被的LAI反演。
基金supported by the National Natural Science Foundation of China (41371396,41401491 and 41471364)the Introduction of International Advanced Agricultural Science and Technology,Ministry of Agriculture,China (948 Program,2011-G6)the Agricultural Scientific Research Fund of Outstanding Talents and the Open Fund for the Key Laboratory of Agri-informatics,Ministry of Agriculture,China (2013009)
文摘Using simultaneously collected remote sensing data and field measurements, this study firstly assessed the consistency and applicability of China high-resolution earth observation system satellite 1 （GF-1） wide field of view （WFV） camera, environment and disaster monitoring and forecasting satellite （H J-l） charge coupled device （CCD）, and Landsat-8 opera- tional land imager （OLI） data for estimating the leaf area index （LAI） of winter wheat via reflectance and vegetation indices （VIs）. The accuracies of these LAI estimates were then assessed through comparison with an empirical model and the PROSAIL radiative transfer model. The effects of radiation calibration, spectral response functions, and spatial resolution on discrepancies in the LAI estimates between the different sensors were also analyzed. The results yielded the following observations： （1） The correlation between reflectance from different sensors is relative good, with the adjusted coefficients of determination （R2） between 0.375 to 0.818. The differences in reflectance are ranging from 0.002 to 0.054. The correlation between VIs from different sensors is high with the R2 between 0.729 and 0.933. The differences in the VIs are ranging from 0.07 to 0.156. These results show the three sensors＇ images can all be used for cross calibration of the reflectance and VIs. （2） The four VIs from the three sensors are all demonstrated to be highly correlated with LAI （R2 between 0.703 and 0.849）. The linear models associated with the 2-band enhanced vegetation index （EVI2）, which feature the highest R2 （higher than 0.746） and the lowest root mean square errors （RMSE） （less than 0.21）, were selected to estimate the winter wheat LAI. The accuracy of the estimated LAI from Landsat-8 was the highest, with the relative errors （RE） of 2.18% and an RMSE of 0.13, while the H J-1 was the lowest, with the RE of 2.43% and the RMSE of 0.15. （3） The inversion errors in the different sensors＇ LAI estimates using the PROSAIL model are small. The accuracy of the GF-1 is the highest with the RE of 3.44%, and the RMSE of 0.22, whereas that of the H J-1 is the lowest with the RE of 4.95%, and the RMSE of 0.26. （4） The effects of the spectral response function and radiation calibration for the different sensors are small and can be ignored, but the effects of spatial resolution are significant and must be taken into consideration in practical applications.
