Cross comparison of the Gaofen-5 AHSI and VIMI sensors
-
摘要: 高分五号卫星搭载的对地观测传感器有高光谱传感器(AHSI)和全谱段光谱传感器(VIMI),但这两种传感器之间的光谱信号是否一致迄今仍鲜有报道。因此,本次研究通过敦煌定标场和滕州地区的3对AHSI和VIMI同步影像对,对2种传感器辐亮度数据之间的一致性进行了交互对比,并与同步的Landsat-8 OLI辐亮度数据进行验证。结果发现,2种传感器的辐亮度数据之间存在着一定的偏差,VIMI的辐亮度数据明显小于AHSI,其偏差率接近32%,二者的R2也只有0.817;在各个对应波段之间,红波段的差距最大,其偏差率超过了40%。与Landsat-8 OLI的验证表明,AHSI的辐亮度数据与OLI很接近,偏差率小于5%,但VIMI数据的偏差率则大于20%。鉴于VIMI和AHSI的辐亮度数据存在着差距,在现阶段如需要同时应用二者的数据,建议要对VIMI数据进行转换校正。本次利用敦煌试验区的模型进行的转换结果表明,VIMI辐亮度数据经转换后可大幅缩小与AHSI数据之间的差距,有利于二者的协同使用。Abstract: The Advanced HyperSpectral Imager (AHSI) and the Visual and Infrared Multispectral Imager (VIMI) are the two sensors onboard China's Gaofen-5 remote sensing satellite. These two sensors provide images of different spectral and spatial resolutions. To date, the quantitative relationship between the two sensors' data has not been investigated in detail. To understand the quantitative relationship and the calibration agreement between AHSI and VIMI, date-coincided image pairs of the two sensors from the Dunhuang calibration site as well as the Tengzhou area were used to conduct a cross comparison. The approach was achieved by evaluating the consistency of the at-sensor radiance data between the two sensors. The results were then validated to those of near-simultaneous Landsat-8 OLI sensor. This study finds that the at-sensor radiance data of VIMI is overall lower than that of AHSI, with a mean absolute percentage error (MAPE) of 32% and an R2 of 0.817. Among the corresponding bands, the red band has the greatest difference between the two sensors, with a MAPE of more than 40%. The validation to Landsat-8 OLI shows that AHSI's radiance data is close to that of OLI with a MAPE of less than 5%, while the MAPE of VIMI is more than 20%. Given the differences between VIMI and AHSI radiance data revealed in this study, it is suggested to correct VIMI data if both data need to be used together. The conversion using the model developed in this study based on the Dunhuang site shows that the difference between the VIMI and the AHSI radiance data can be greatly reduced after conversion. A synergistic use of AHSI and calibrated VIMI data can greatly benefit the science community by proving a high-quality observation of the Earth.
-
Keywords:
- Gaofen-5 satellite /
- AHSI /
- VIMI /
- At-sensor radiance /
- Cross comparison
-
-
表 1 同步影像对及其参数
Table 1 Date-Coincident Image Pairs and Parameters
影像对 传感器 轨道号 日期 时间 太阳天顶角/(°) 太阳方位角/(°) 试验区特征 敦煌1 Landsat-8 OLI 138/032 2019-11-14 12:26 60.31 162.57 裸土为主 GF-5 AHSI 356/621 14:41 60.61 199.39 GF-5 VIMI 14:41 60.56 199.40 敦煌2 GF-5 AHSI 354/621 2019-07-14 14:38 22.30 214.35 裸土为主,兼有植被和少量建筑物 GF-5 VIMI 14:38 21.45 214.43 滕州 Landsat-8 OLI 122/036 2019-05-22 10:48 23.04 123.05 植被为主,兼有建筑物和少量水体 GF-5 AHSI 310/611 13:13 20.82 179.80 GF-5 VIMI 13:13 20.78 179.80 表 2 GF-5 VIMI、GF-5 AHSI和Landsat-8 OLI对应波段的信息
Table 2 Specifications of Corresponding Bands of GF-5 VIMI, GF-5 AHSI, and Landsat-8 OLI
波段名 GF-5 VIMI GF-5 AHSI Landsat-8 OLI 波段号 波长范围/nm 波段号 波长范围/nm 波段号 波长范围/nm 蓝 B1 440~510 B13~B28 439~508 B2 452~512 绿 B2 510~580 B29~B45 508~581 B3 533~590 红 B3 620~680 B55~B68 619~679 B4 636~673 近红外 B4 760~870 B88~B113 760~871 B5 851~879 中红外1 B5 1 540~1 700 B215~B233 1 541~1 700 B6 1 566~1 651 中红外2 B6 2 060~2 350 B277~B310 2 062~2 349 B7 2 107~2 294 表 3 AHSI和VIMI对应波段的辐亮度数据对比
Table 3 Radiance Data Comparison of Corresponding Bands Between AHSI and VIMI
试验区 波段 GF-5 VIMI GF-5 AHSI R2 ME MAPE/% RMSE 最小值 最大值 均值 标准差 最小值 最大值 均值 标准差 敦煌1 蓝 49.55 71.77 55.37 1.48 58.86 84.62 64.89 1.64 0.751 9.52 17.19 9.55 绿 39.09 61.45 45.03 1.69 52.83 84.40 61.29 2.29 0.807 16.26 36.11 16.29 红 32.12 52.54 38.61 1.96 45.81 78.26 55.77 2.82 0.853 17.16 44.44 17.21 近红外 24.36 42.27 30.23 1.72 29.60 52.69 37.25 2.25 0.900 7.02 23.22 7.06 中红外1 7.28 8.90 8.19 0.34 7.96 13.05 10.65 0.73 0.609 2.46 30.04 2.51 中红外2 2.18 2.82 2.54 0.13 2.53 4.33 3.49 0.27 0.779 0.95 37.40 0.96 敦煌2 蓝 56.27 219.42 98.21 16.17 60.01 239.11 112.74 21.36 0.940 14.53 14.79 15.95 绿 40.01 186.34 81.89 15.61 51.18 192.95 113.05 23.22 0.937 31.16 38.05 32.43 红 24.17 151.71 69.02 16.55 31.15 171.67 103.80 26.90 0.932 34.78 50.39 36.72 近红外 26.55 114.54 66.46 9.86 28.00 159.43 83.94 15.26 0.907 17.48 26.30 20.57 中红外1 10.01 22.11 14.15 1.54 3.35 50.97 19.59 3.39 0.685 5.44 38.45 5.90 中红外2 2.15 10.01 4.10 0.79 0.74 17.30 5.77 1.57 0.762 1.67 40.73 1.92 滕州 蓝 48.56 179.78 67.27 9.50 55.72 211.03 75.16 12.46 0.839 7.89 11.73 9.53 绿 31.19 156.95 51.23 9.32 38.50 191.12 65.22 14.01 0.834 13.99 27.31 15.50 红 17.84 139.57 38.10 11.11 20.53 169.94 50.26 17.64 0.846 12.16 31.92 14.89 近红外 14.56 124.88 65.02 13.21 15.50 167.03 82.64 18.72 0.837 17.62 27.10 19.57 中红外1 5.69 22.12 9.88 1.88 1.19 48.53 13.67 4.02 0.701 3.79 38.36 4.62 中红外2 1.07 11.70 2.36 0.69 0.21 16.78 3.32 1.52 0.787 0.96 40.68 1.36 均值 0.817 11.93 31.90 12.92 注:最小值、最大值、均值、标准差、ME、RMSE的单位均为W·m-2·sr-1·μm-1 表 4 AHSI和VIMI与Landsat-8 OLI的辐亮度数据对比验证
Table 4 Comparison and Validation of Radiance Data of AHSI and VIMI with Landsat-8 OLI
试验区 波段 Landsat-8 OLI GF-5 AHSI GF-5 VIMI 数值范围 均值 数值范围 均值 R2 ME MAPE/% 数值范围 均值 R2 ME MAPE/% 敦煌1 蓝 55.40 ~ 84.21 62.26 58.86 ~ 84.62 64.89 0.70 2.63 4.22 49.55 ~ 71.77 55.38 0.68 -6.88 11.05 绿 49.43 ~ 89.32 59.08 52.83 ~ 84.40 61.29 0.74 2.21 3.74 39.09 ~ 61.45 45.04 0.73 -14.04 23.76 红 45.42 ~ 80.79 56.26 45.81 ~ 78.26 55.77 0.79 -0.49 0.87 32.12 ~ 52.54 38.61 0.77 -17.65 31.37 近红外 29.27 ~ 54.73 37.56 29.60 ~ 52.69 37.25 0.82 -0.31 0.83 24.36 ~ 42.27 30.23 0.83 -7.33 19.52 中红外1 8.46 ~ 15.20 11.64 7.96 ~ 13.05 10.65 0.86 -0.99 8.51 7.28 ~ 8.90 8.19 0.70 -3.45 29.64 中红外2 2.69 ~ 4.94 3.73 2.53 ~ 4.33 3.49 0.88 -0.24 6.43 2.18 ~ 2.82 2.54 0.80 -1.19 31.90 均值 0.80 0.47 4.10 0.75 -8.42 24.54 滕州 蓝 50.49 ~ 217.79 69.59 55.72 ~ 211.03 75.16 0.83 5.57 8.00 48.56 ~ 179.78 67.27 0.88 -2.32 3.33 绿 31.22 ~ 215.24 60.76 38.50 ~ 191.12 65.22 0.83 4.46 7.34 31.19 ~ 156.95 51.23 0.90 -9.53 15.68 红 17.73 ~ 216.22 49.08 20.53 ~ 169.94 50.26 0.84 1.18 2.40 17.84 ~ 139.57 38.10 0.89 -10.98 22.37 近红外 6.92 ~ 181.47 87.47 15.50 ~ 167.03 82.64 0.81 -4.83 5.52 14.56 ~ 124.88 65.02 0.86 -22.45 25.67 中红外1 0.87 ~ 46.82 13.87 1.19 ~ 48.53 13.67 0.85 -0.20 1.44 5.69 ~ 22.12 9.87 0.68 -4.00 28.84 中红外2 0.24 ~ 22.22 3.32 0.21 ~ 16.78 3.32 0.84 0.00 0.00 1.07 ~ 11.70 2.36 0.78 -0.96 28.92 均值 0.83 1.03 4.12 0.83 -8.37 20.80 注:数值范围、均值、ME的单位均为W·m-2·sr-1·μm-1 表 5 VIMI与AHSI辐亮度数据转换方程
Table 5 Equations for the Conversion of VIMI Radiance Data to AHSI Radiance Data
波段 关系方程 蓝 AHSI = 1.243 3VIMI - 9.407 8 绿 AHSI = 1.439 8VIMI - 4.895 6 红 AHSI = 1.564 8VIMI - 4.236 9 近红外 AHSI = 1.465 6VIMI -11.450 0 中红外1 AHSI = 1.827 6VIMI - 6.266 6 中红外2 AHSI = 1.775 4VIMI - 1.517 3 表 6 VIMI辐亮度数据转换前后的精度对比
Table 6 Accuracy Comparison of VIMI Radiance Data Before and After Conversion
波段 滕州试验区 敦煌1试验区 转换前 转换后 转换前 转换后 ME MAPE/% 斜率差 ME MAPE/% 斜率差 ME MAPE/% 斜率差 ME MAPE/% 斜率差 蓝 7.89 11.73 0.129 -0.93 1.25 0.092 9.51 17.18 0.027 5.45 9.17 0.218 绿 13.98 27.29 0.282 3.65 5.30 0.109 16.26 36.10 0.238 1.35 2.25 0.140 红 12.16 31.91 0.353 5.12 9.25 0.136 17.16 44.44 0.343 -0.41 0.73 0.142 近红外 17.62 27.10 0.191 1.20 1.43 0.188 7.01 23.20 0.251 4.39 13.35 0.146 中红外1 3.79 38.37 0.778 -1.89 16.00 0.027 2.46 30.00 0.679 1.95 22.37 0.081 中红外2 0.96 40.93 0.951 -0.66 24.79 0.099 0.94 37.03 0.827 0.49 16.25 0.029 均值 9.40 29.56 0.447 1.08 9.67 0.109 8.89 31.33 0.394 2.20 10.69 0.126 注:ME的单位为W·m-2·sr-1·μm-1;斜率差取图 4回归方程的斜率与1∶1线差值的绝对值 表 7 AHSI和VIMI空间分辨率对辐亮度值的影响
Table 7 Impact of the Resolution Difference Between AHSI and VIMI on Radiance Data
类型 波段 AHSI 30 m分辨率4个像元的辐亮度均值/(W·m-2·sr-1·μm-1) VIMI 20 m分辨率9个像元的辐亮度均值/(W·m-2·sr-1·μm-1) MAPE/% 裸土 绿 122.09 87.46 39.60 红 110.14 72.86 51.17 近红外 70.84 56.76 24.81 均值 101.02 72.36 39.62 植被 绿 57.24 49.25 16.21 红 33.71 28.60 17.86 近红外 133.16 103.21 29.01 均值 74.70 60.36 23.77 -
[1] Teillet P M, Barker J L, Markham B L, et al. Radiometric Cross-Calibration of the Landsat-7 ETM+ and Landsat-5 TM Sensors Based on Tandem Data Sets[J]. Remote Sensing of Environment, 2001, 78(1/2):39-54
[2] Xu Hanqiu. Retrieval of the Reflectance and Land Surface Temperature of the Newly-Launched Landsat 8 Satellite[J]. Chinese Journal of Geophysics, 2015, 58(3):741-747(徐涵秋. 新型Landsat8卫星影像的反射率和地表温度反演[J]. 地球物理学报, 2015, 58(3):741-747) [3] Mishra N, Haque M, Leigh L, et al. Radiometric Cross Calibration of Landsat 8 Operational Land Imager (OLI) and Landsat 7 Enhanced Thematic Mapper Plus (ETM+)[J]. Remote Sensing, 2014, 6(12):12619-12638
[4] Mancino G, Ferrara A, Padula A, et al. Cross-Comparison between Landsat 8(OLI) and Landsat 7(ETM+) Derived Vegetation Indices in a Mediterranean Environment[J]. Remote Sensing, 2020, 12(2):291
[5] Angal A, Xiong X X, Wu A S, et al. Multitemporal Cross-Calibration of the Terra MODIS and Landsat 7 ETM+ Reflective Solar Bands[J]. IEEE Transactions on Geoscience and Remote Sensing, 2013, 51(4):1870-1882
[6] Zhang Aiyin, Zhang Xiaoli. Land Surface Temperature Retrieved from Landsat-8 and Comparison with MODIS Temperature Product[J]. Journal of Beijing Forestry University, 2019, 41(3):1-13(张爱因, 张晓丽. Landsat-8地表温度反演及其与MODIS温度产品的对比分析[J]. 北京林业大学学报, 2019, 41(3):1-13) [7] Xu H Q, Zhang T J. Assessment of Consistency in Forest-Dominated Vegetation Observations between ASTER and Landsat ETM+ Images in Subtropical Coastal Areas of Southeastern China[J]. Agricultural and Forest Meteorology, 2013, 168:1-9
[8] Xu Hanqiu, Zhang Tiejun, Li Chunhua. Cross Comparison of Thermal Infrared Data between ASTER and Landsat ETM+ Sensors[J]. Geomatics and Information Science of Wuhan University, 2011, 36(8):936-940(徐涵秋, 张铁军, 李春华. ASTER与Landsat ETM~+热红外传感器数据的比较研究[J]. 武汉大学学报·信息科学版, 2011, 36(8):936-940) [9] Chander G, Coan M J, Scaramuzza P L. Evaluation and Comparison of the IRS-P6 and the Landsat Sensors[J]. IEEE Transactions on Geoscience and Remote Sensing, 2008, 46(1):209-221
[10] Thenkabail P S. Inter-Sensor Relationships between IKONOS and Landsat-7 ETM+ NDVI Data in Three Ecoregions of Africa[J]. International Journal of Remote Sensing, 2004, 25(2):389-408
[11] Chastain R, Housman I, Goldstein J, et al. Empirical Cross Sensor Comparison of Sentinel-2A and 2B MSI, Landsat-8 OLI, and Landsat-7 ETM+ Top of Atmosphere Spectral Characteristics over the Conterminous United States[J]. Remote Sensing of Environment, 2019, 221:274-285
[12] Barsi J A, Alhammoud B, Czapla-Myers J, et al. Sentinel-2A MSI and Landsat-8 OLI Radiometric Cross Comparison over Desert Sites[J]. European Journal of Remote Sensing, 2018, 51(1):822-837
[13] Quan W T. A Multiplatform Approach Using MODIS Sensors to Cross-Calibrate the HJ-1A/CCD1 Sensors over Aquatic Environments[J]. Journal of the Indian Society of Remote Sensing, 2015, 43(4):687-695
[14] Zhao Kai, Xu Janbo, Zhao Zhizhong, et al. Cross Comparison of HJ-1 A/B CCD and Landsat TM/ETM+ Multispectral Measurements for NDVI, SAVI and EVI Vegetation Index[J]. Remote Sensing Technology and Application, 2013, 28(4):674-680(赵凯, 徐剑波, 赵之重, 等. HJ-1 A/B CCD与Landsat TM/ETM+植被指数的交互比较[J]. 遥感技术与应用, 2013, 28(4):674-680) [15] Zhang Xuewen, Fu Qiaoyan, Han Qijin, et al. The Field Radiometric Calibration and Validation of ZY-3 Multispectral Sensor[J]. Spectroscopy and Spectral Analysis, 2014, 34(9):2476-2480(张学文, 傅俏燕, 韩启金, 等. 资源三号多光谱传感器场地辐射定标与验证[J]. 光谱学与光谱分析, 2014, 34(9):2476-2480) [16] Xu Hanqiu, Liu Zhicai, Guo Yanbin. Comparison of NDVI Data between GF-1 PMS1 and ZY-3 MUX Sensors[J]. Transactions of the Chinese Society of Agricultural Engineering, 2016, 32(8):148-154(徐涵秋, 刘智才, 郭燕滨. GF-1 PMS1与ZY-3 MUX传感器NDVI数据的对比分析[J]. 农业工程学报, 2016, 32(8):148-154) [17] Li J, Feng L, Pang X P, et al. Radiometric Cross Calibration of Gaofen-1 WFV Cameras Using Landsat-8 OLI Images:A Simple Image-Based Method[J]. Remote Sensing, 2016, 8(5):411
[18] Wu Xiaoping, Xu Hanqiu, Jiang Qiaoling. Cross-Comparison of GF-1, GF-2 and Landsat-8 OLI Sensor Data[J]. Geomatics and Information Science of Wuhan University, 2020, 45(1):150-158(吴晓萍, 徐涵秋, 蒋乔灵. GF-1、GF-2与Landsat-8卫星多光谱数据的交互对比[J]. 武汉大学学报·信息科学版, 2020, 45(1):150-158) [19] Chen Y P, Sun K M, Li D R, et al. Radiometric Cross-Calibration of GF-4 PMS Sensor Based on Assimilation of Landsat-8 OLI Images[J]. Remote Sensing, 2017, 9(8):811
[20] Gao C X, Liu Y K, Qiu S, et al. Radiometric Cross-Calibration of GF-4/VNIR Sensor with Landsat8/OLI, Sentinel-2/MSI, and Terra/MODIS for Monitoring Its Degradation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13:2337-2350
[21] Liu Yinnian, Sun Dexin, Cao Kaiqin, et al. Evaluation of GF-5 AHSI On-Orbit Instrument Radiometric Performance[J]. Journal of Remote Sensing, 2020, 24(4):352-359(刘银年, 孙德新, 曹开钦, 等. 高分五号可见短波红外高光谱相机在轨辐射性能评估[J]. 遥感学报, 2020, 24(4):352-359) [22] Chander G, Haque M O, Micijevic E, et al. A Procedure for Radiometric Recalibration of Landsat 5 TM Reflective-Band Data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2010, 48(1):556-574
[23] Choi T, Xiong X, Chander G, et al. Assessment of the Short-term Radiometric Stability between Terra MODIS and Landsat 7 ETM+ Sensors[C]. 2009 IEEE International Geoscience and Remote Sensing Symposium, Cape Town, South Africa, 2009
[24] Yan Lei, Gou Zhiyang, Zhao Hongying, et al. In-Flight Spectral Calibration of UAV Hyperspectral Imager Based on Radiance Matching[J]. Journal of Infrared and Millimeter Waves, 2012, 31(6):517-522(晏磊, 勾志阳, 赵红颖, 等. 基于辐亮度匹配的无人机载成像光谱仪外场光谱定标研究[J]. 红外与毫米波学报, 2012, 31(6):517-522) [25] Chander G, Meyer D J, Helder D L. Cross Calibration of the Landsat-7 ETM+ and EO-1 ALI Sensor[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(12):2821-2831
[26] Chen W, Zhou G. Comparison of Satellite Measured Temperature Using Terra ASTER and Landsat ETM+ Data[J]. IEEE International Geoscience and Remote Sensing Symposium, 2004, 3:1723-1726
[27] Meyer D J, Chander G. The Effect of Variations in Relative Spectral Response on the Retrieval of Land Surface Parameters from Multiple Sources of Remotely Sensed Imagery[C]. 2007 IEEE International Geoscience and Remote Sensing Symposium, Barcelona, Spain, 2007
[28] Malila W A, Meyers T J. Tasseled-cap Transformation of JERS-1 OPS Multispectral Data-an Initial Version[C], Geoscience and Remote Sensing Symposium (IGARSS 1995), Firenze, Italy, 1995
[29] Liu Q Y, Yu T, Gao H L. Radiometric Cross-Calibration of GF-1 PMS Sensor with a New BRDF Model[J]. Remote Sensing, 2019, 11(6):707
[30] Li Juan, Feng Lian, Pang Xiaoping. Comparison of the Cross-Calibration Methods between Image-Based and RTMBRDF for GF-1 Images[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(7):882-890(李娟, 冯炼, 庞小平. 针对GF-1遥感影像的基于影像与基于辐射传输模型的两种交叉定标方法比较[J]. 测绘学报, 2017, 46(7):882-890) [31] Zhang H K, Roy D P, Yan L, et al. Characterization of Sentinel-2A and Landsat-8 Top of Atmosphere, Surface, and Nadir BRDF Adjusted Reflectance and NDVI Differences[J]. Remote Sensing of Environment, 2018, 215:482-494
[32] Zhao Y G, Ma L L, Li C R, et al. Radiometric Cross-Calibration of Landsat-8/OLI and GF-1/PMS Sensors Using an Instrumented Sand Site[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018, 11(10):3822-3829