钱永兰,赵晓凤,张晔萍,张明伟,王圆圆,吴门新,王小平. 2023. FY-3B/D星NDVI与MODIS NDVI对作物长势监测对比. 气象学报,81(4):660-672. DOI: 10.11676/qxxb2023.20230012
引用本文: 钱永兰,赵晓凤,张晔萍,张明伟,王圆圆,吴门新,王小平. 2023. FY-3B/D星NDVI与MODIS NDVI对作物长势监测对比. 气象学报,81(4):660-672. DOI: 10.11676/qxxb2023.20230012
Qian Yonglan, Zhao Xiaofeng, Zhang Yeping, Zhang Mingwei, Wang Yuanyuan, Wu Menxin, Wang Xiaoping. 2023. Comparison of crop growth monitoring based on FY-3B/D NDVI and MODIS NDVI. Acta Meteorologica Sinica, 81(4):660-672. DOI: 10.11676/qxxb2023.20230012
Citation: Qian Yonglan, Zhao Xiaofeng, Zhang Yeping, Zhang Mingwei, Wang Yuanyuan, Wu Menxin, Wang Xiaoping. 2023. Comparison of crop growth monitoring based on FY-3B/D NDVI and MODIS NDVI. Acta Meteorologica Sinica, 81(4):660-672. DOI: 10.11676/qxxb2023.20230012

FY-3B/D星NDVI与MODIS NDVI对作物长势监测对比

Comparison of crop growth monitoring based on FY-3B/D NDVI and MODIS NDVI

  • 摘要: 风云三号卫星(FY-3)NDVI数据是中国对全球用户开放可进行全球尺度作物长势监测的主要卫星遥感数据源,客观评估其进行作物长势监测的真实状况和能力,对业务产品应用推广、产品算法和流程改进以及长时序标准数据生产等具有重要意义。采用国际通用的年际差值比较模型、植被状况指数模型(VCI)以及基于NDVI的作物生长过程评估了中国风云三号B星(FY-3B)和D星(FY-3D)NDVI旬、月合成业务产品与同期MODIS MOD13A1和MOD13A3 NDVI数据对作物长势的监测情况。年际比较模型结果表明,FY-3 NDVI的监测动态范围比MODIS NDVI的窄约0.1—0.15,两者差值在−0.02—0.02,均方根误差在0.08—0.09,在长势分级评估时可忽略,但动态范围偏窄提示有必要在分级评估时使用窄一些的阈值范围。植被状况指数模型结果表明,FY-3B NDVI和MODIS NDVI的监测结果高度一致,因此FY-3B NDVI和MODIS NDVI一样可准确反映作物生长的年际差异。作物生长过程动态监测结果表明,FY-3和MODIS月合成NDVI的年际比较结果总体一致,但在作物快速生长阶段,MODIS用3个16 d数据采用权重法合成月值的方法有一定偏差,导致两者出现差异;FY-3旬合成数据和MODIS 16 d合成数据的监测结果总体一致,前者在作物生长旺季的年际区分度更优,后者的监测曲线更平滑。但是,FY-3 NDVI与MODIS NDVI存在总体偏差,需进一步的数据处理、质量控制和客观标定,以生成与MODIS、AVHRR等一致且连续的长时序FY-3 NDVI标准数据集产品,方可用于作物生长参数定量评估。

     

    Abstract: Objective evaluation of NDVI data, the crop growth monitoring data from Chinese Fengyun 3 (FY-3) that is open to the worldwide users, is helpful for the application and promotion of operational products, algorithm improvement and calibration of long-term time series datasets. In the study, the operational 10-day and monthly composite NDVI data sets of FY-3B and FY-3D are compared with the similar operational MODIS NDVI data sets of MOD13A1 and MOD13A3 and assessed for crop growth monitoring based on the interannual difference comparative model, vegetation condition index model and the NDVI-based crop growing progress. The results of the interannual difference comparative model suggest that the dynamic range of monitoring based on FY-3 NDVI is about 0.1—0.15 narrower than that based on MODIS NDVI, the model output differences are between −0.02 and 0.02, and their standard deviations are between 0.08 and 0.09, which can be ignored in cataloguing evaluation of the crop growth condition. But the narrower dynamic range of FY-3 suggests it is necessary to grade the crop level with a narrower threshold. As for the vegetation condition index model, the differences between FY-3B NDVI and MODIS NDVI are close to zero, which indicates FY-3B NDVI can generate the approximately equal model output to MODIS NDVI. Therefore, FY-3B NDVI, like MODIS NVDI, can make a distinction of different crop growth conditions among different years. Additionally, FY-3 and MODIS monthly composite NDVIs produce approximately the same change trend along with the crop growing season, but during the rapid change period of crop NDVI, MODIS monthly composite NDVI usually yields a different trend, which is resulted from its aggregation method using three 16-day composites with a weighted average approach. FY-3 10-day composite NDVI and MODIS 16-day composite NDVI have the same crop progress monitoring result, but the former provides better interannual differentiation during the exuberant period of crop growth, while the latter has a smoother monitoring curve. However, there is an overall difference between FY-3 NDVI and MODIS NDVI, which needs to be eliminated through further data processing, quality control and objective calibration to produce standardized NDVI products consistent with MODIS NDVI for long-term quantitative assessment of crop growth condition.

     

/

返回文章
返回