Microsoft Word - 3--梁建_news_.doc

Size: px

Start display at page:

Download "Microsoft Word - 3--梁建_news_.doc"

侧垒乐
5 years ago
Views:

1 控制点与检查点数量和比例对水深遥感反演精度的影响分析梁建,, 张杰 2 2, 马毅 (., 6026; 2., 26606) 摘要 : 在水深遥感反演半经验模型中, 水深控制点和检查点选取是一项非常重要的工作, 前者用于建立实测水深值与遥感影像灰度值之间的数量关系, 后者用于评价水深反演精度但前人在相关研究中并未就控制点和检查点的选取数量和比例给出有实验依据的建议, 亦未开展其对水深反演精度的影响分析针对这个问题, 作者使用 6020 个实测水深点和 WorldVew-2 GF- WFV Landsat8 OLI 3 种多光谱影像, 基于三波段水深反演模型开展了 8 组实验, 分析比较了不同数量和比例的控制点与检查点对反演结果的影响, 评价过程利用平均绝对误差 (Mean Absolute Error, MAE) 平均相对误差(Mean Relatve Error, MRE) 和反映实测水深与反演水深相关性的决定系数 R 2 等 3 种参数结果表明 : () 当控制点数量达到 3 个时, 水深反演精度即趋于稳定 ; (2) 检查点数量在 30 个时, 其评价指标已可以代表模型反演精度 ; (3) 控制点和检查点的数量比例对反演精度并无影响, 但控制点选取过多则会产生反演精度降低的现象, 采用人工选取检查点时剔除浅水区的异常点即可有效避免这一问题关键词 : 水深遥感反演 ; 控制点 ; 检查点中图分类号 : P237 文献标识码 : A 文章编号 : (205) do: 0.759/hykx20400 [-5],,, 3, :,,,,,,,, [6],,,, WorldVew- 2 GF- WFV Landsat8 OLI, 6020,,, R 2 3,, 数据与方法. 数据及预处理, WorldVew-2 GF- WFV Landsat8 OLI ( ), : FLAASH : ; : : (202BAB6B0-02) : (985-),,,,, E-mal: fors@63.com Marne Scences / Vol. 39, No. 2 / 205 5

表影像参数 Tab. Specfcatons of the test mages (m) (UTC) WorldVew-2 4 2 202-09-20 T03: 26: 8 2 GF- WFV 4 6 204-03-2 T03: : 04 3 Landsat8 OLI 7 30 204-06-25 T02: 53: 25 ( ) ( ) Fg.

2 表影像参数 Tab. Specfcatons of the test mages (m) (UTC) WorldVew T03: 26: 8 2 GF- WFV T03: : 04 3 Landsat8 OLI T02: 53: 25 ( ) ( ) Fg. Images of Dong Island and dstrbuton of control ponts (green) and check ponts (red) WorldVew-2, WorldVew-2,,, DN WorldVew-2, : L ( ) abscalfactor DN () L( ), W m 2 sr μm ; abscalfactor, WorldVew , , ;, μm abscalfactor WorldVew-2 (.IMD) ENVI 5. FLAASH, 4 WorldVew-2 FLAASH WorldVew-2 ( 8 WorldVew-2 ), ,, 2 908,, 0. m 0 m 30 m ,, 3 :, ;, ; 3, 9, 3 2 3,,, 表 2 控制点分组 Tab.2 Groups of control ponts Ct Ct Ct Ct Ct Ct Ct , 2 Ct , 4 Ct , 0 表 3 检查点分组 Tab. 3 Groups of check ponts Ck Ck Ck , 0 6 / 205 / 39 / 2

3 , WorldVew-2 GF-WFV Landsat8 OLI m,.2 方法 [3-6] : z L Ls CRbe (2),, L ; Ls ; C Rb ; ; k ; f ( 2); Z (2): ln( CR b) ln( L Ls) z (3) (3) ln( CR ) b A 0, fk A, ln( L L ) X, (3) : s z A0 AX (4), A 0 A, (2), : z A0 AX A2X2 A3X3 (5) A 0, A A 2 A 3, 2 结果与分析 27, 4~ 6, MAE MRE R 2, 3,, :, ;,,,, ; 表 4 WorldVew-2 影像的反演试验结果 Tab. 4 Inverson results of WorldVew-2 mage MAE(m) MRE(%) R 2 MAE(m) MRE(%) R 2 MAE(m) MRE(%) R 2 Ct Ct Ct Ct Ct Ct Ct Ct Ct 表 5 GF 影像的反演试验结果 Tab. 5 Inverson results of GF mage MAE(m) MRE(%) R 2 MAE(m) MRE(%) R 2 MAE(m) MRE(%) R 2 Ct Ct Ct Ct Ct Ct Ct Ct Ct Marne Scences / Vol. 39, No. 2 / 205 7

4 表 6 Landsat8 OLI 影像的反演试验结果 Tab. 6 Inverson results of Landsat8 OLI mage MAE(m) MRE(%) R 2 MAE(m) MRE(%) R 2 MAE(m) MRE(%) R 2 Ct Ct Ct Ct Ct Ct Ct Ct Ct ,, ;,,,, Ct3, 3, 3,,,,,,,, Hughes [6],,,, ;,, ;,,,,,,, 30,,,,, 3 结论, : () 3, ; (2) 30, ; (3),,, m 3,, 5.8 m ZY-3, 6.0 m SPOT6, 8.0 m GF- PSM 0.0 m SPOT5,, : [] Lyzenga D R. Passve remote Sensng technques for mappng water depth and bottom features[j]. Appled Optcs, 978, 7(3): [2] Lyzenga D R. Remote sensng of bottom reflectance and water attenuaton parameters n shallow water usng arcraft and Landsat data[j]. Internatonal Journal of Remote Sensng, 98, 2(): [3] Polcyn F C, Sattnger I J. Water depth determnaton usng remote sensng technques[c]//proceedngs of the 6th Internatonal Symposum on Remote Sensng of Research Insttute, 969: [4] Polcyn F C, Lyzenga D R. Calculaton of water depth 8 / 205 / 39 / 2

5 from ERTS-MSS data[c]//proceedngs Symposum on Sgnfcant Results Obtaned from ERTS-. Washngton: NASA 973: [5] Tans F J, Hallada W A. Evaluaton of landsat thematc mapper data for shallow water bathymetry[c]//proceedng of 8th Internatonal Symposum on Remote Sensng of Research Insttute, 984: [6] Tans F J, Byrne H J. Optmzaton of multspectral sensors for bathymetry applcatons[c]//proceedng of 9th Internatonal Symposum on Remote Sensng of Research Insttute, 985: [7] Jupp D L B, et al. Remote sensng for plannng and managng the Great Barrer Reef of Australa[J]. Photogrammetra, 985, 40: [8] Clark R K, Fay T H. Thematc mapper band selecton for bathymetrc measurements[j]. Ocean Optcs X, 990, 302: [9] Lafon V, Frodefond J M, Lahet F, et al. SPOT shallow water bathymetry of a moderately turbd tdal nlet based on feld measurements[j]. Remote Sensng of Envronment, 2002, 8: [0]. [J]., 982, 3(3): []. [J]., 998, 26(6): [2],. [J]., 200, 4: [3],. [J]., 2003, 22(3): [4]. [D]. :, [5]. [D]. :, 20. [6],,,. Hughes [J]., 2004, 2(4): Analyss of the nfluence of the amount and proporton of control ponts and check ponts on the accuracy of bathymetry remote sensng nverson LIANG Jan, ZHANG Je, 2, MA Y 2 (. Dalan Martme Unversty, Dalan 6026, Chna; 2. Frst Insttute of Oceanography, State Oceanc Admnstraton, Qngdao 26606, Chna) Receved: Oct.,, 204 Key words: Bathymetry remote sensng nverson; control pnts; check ponts Abstract: Selecton of control ponts and check ponts s an mportant work n sem-emprcal model for remote sensng depth nverson. Control ponts are used to establsh the mathematcal relatonshp between the measured water depth and the gray values of remote sensng mage. Check ponts are used to evaluate the accuracy. However, there are no expermental suggestons on the amount and proporton of control ponts and check ponts gven by prevous studes, nor water depth retreval accuracy analyses were made. In ths paper, 6020 measured depths and WorldVew-2, GF- WFV and Landsat8 OLI multspectral mages are appled to carry out 8 experments based on 3-band water depth nverson model. Mean Absolute Errors (MAE), Mean Relatve Errors (MRE) and R 2 are analyzed, and the results show that: () The nverson accuracy tends to be steady when the amount of control ponts comes to 3. (2) The ndexes can represent the accuracy of nverson when the amount of check ponts comes to 30. (3) The proporton of control ponts and check ponts has no effect on the accuracy of nverson, but the accuracy wll be lowered f too many control ponts were selected. Ths phenomenon can be avoded by elmnatng the exceptonal ponts manually. ( 本文编辑 : 谭雪静 ) Marne Scences / Vol. 39, No. 2 / 205 9

Microsoft Word - 4--赵鹏_new_.doc

Microsoft Word - 4--赵鹏_new_.doc 47 4 Vol.47, No.4 2016 7 OCEANOLOGIA ET LIMNOLOGIA SINICA Jul., 2016 澙 * 1, 2, 3 3 1 1, 2 1, 2 4 (1. 266071; 2. 100049; 3. 300171; 4. 100081) 澙, 澙, 澙, IKONOS, (Band3),,,, 271.23cm, 13.16% 澙 ; ; ; ; IKONOS

Microsoft Word - 3--梁 建_news_.doc

Microsoft Word - 3--梁建_news_.doc