基于无人机遥感和集成学习的苏州市河流悬浮物浓度反演

余成; 唐毅; 潘杨; 易虹辰; 谷一平; 褚峰; 施骄阳

PDF(3357 KB)

中国环境科学 ›› 2023, Vol. 43 ›› Issue (10) : 5235-5246.

水污染与控制

基于无人机遥感和集成学习的苏州市河流悬浮物浓度反演

余成¹, 唐毅², 潘杨², 易虹辰², 谷一平², 褚峰², 施骄阳²

作者信息 +

Inversion of suspended sediment concentration in rivers of Suzhou based on UAV remote sensing and ensemble learning

YU Cheng¹, TANG Yi², PAN Yang², YI Hong-chen², GU Yi-ping², ZHU Feng², SHI Jiao-yang²

Author information +

文章历史 +

摘要

针对单一模型容易过拟合的缺陷，通过4种不同的集成学习策略实现多种机器学习方法之间的优势互补，从而提升模型的精度和泛化性，并以苏州市为研究区，无人机采集的多光谱遥感影像和实地测量的悬浮物浓度数据为基础建立集成学习反演模型.同时，4种常用的回归分析方法和3种经典的机器学习方法被作为对比方法以探究集成学习策略的有效性.结果表明，集成学习策略能够克服单一模型各自的缺陷，显著提升遥感反演的精度和泛化性，其中表现最优的集成学习反演模型的验证集决定系数达到0.821，相比回归分析方法提升38.21%，相比单一机器学习方法提升16.79%.此外，集成学习方法反演的绝对误差集中于较小值，其绝对误差均值和中值均优于传统方法，以期提高城市悬浮物浓度反演精度，为无人机遥感河流悬浮物浓度反演提供指导.

Abstract

The inversion of suspended sediment concentrations of urban rivers by remote sensing has important practical significance for water environmental management. To address the problem of overfitting in individual models, this study attempts to improve the accuracy and generalizability of the inversion model by realizing the complementary advantages among four different ensemble learning strategies. Ensemble learning inversion models were established based on multispectral remote sensing images collected by unmanned aerial vehicles and field-measured suspended sediment concentrations of Suzhou in this study. Four commonly used regression methods and three classic machine learning methods were used to validate the effectiveness of the ensemble learning strategies. The results demonstrate that the ensemble learning strategies effectively mitigate the limitations of individual models, substantially improving the accuracy and generalizability of the remote sensing inversions. The stacking strategy achieves the best performance with a validation set's coefficient of determination of 0.821, show casing a 38.21% enhancement compared with the regression methods and a 16.79% improvement compared to the individual machine learning methods. The absolute error of the ensemble learning method is concentrated on small values, with its mean and median absolute errors surpassing the traditional methods. This study can improve the accuracy of urban suspended sediment concentration inversion and provide guidance for unmanned aerial vehicle remote sensing of river suspended sediment concentration inversion.

导出引用

余成, 唐毅, 潘杨, 易虹辰, 谷一平, 褚峰, 施骄阳. 基于无人机遥感和集成学习的苏州市河流悬浮物浓度反演[J]. 中国环境科学. 2023, 43(10): 5235-5246

YU Cheng, TANG Yi, PAN Yang, YI Hong-chen, GU Yi-ping, ZHU Feng, SHI Jiao-yang. Inversion of suspended sediment concentration in rivers of Suzhou based on UAV remote sensing and ensemble learning[J]. China Environmental Science. 2023, 43(10): 5235-5246

中图分类号： X87

参考文献

[1] 卢世军.Ⅱ类水体悬浮物遥感研究进展[J]. 现代计算机(专业版), 2016,(32):34-39. Lu S J. Research progress on the retrieval of suspended sediment from II water[J]. Modern Computers, 2016,(32):34-39.
[2] 王波,黄津辉,郭宏伟,等.基于遥感的内陆水体水质监测研究进展[J]. 水资源保护, 2022,38(3):117-124. Wang B, Huang J H, Guo H W, et al. Progress in research on inland water quality monitoring based on remote sensing[J]. Water Resources Protection, 2022,38(3):117-124.
[3] Hong S M, Cho K H, Park S, et al. Estimation of cyanobacteria pigments in the main rivers of South Korea using spatial attention convolutional neural network with hyperspectral imagery[J]. GIScience & Remote Sensing, 2022,59(1):547-567.
[4] 杨振,卢小平,武永斌,等.无人机高光谱遥感的水质参数反演与模型构建[J]. 测绘科学, 2020,45(9):60-64,95. Yang Z, Lu X P, Wu Y B, et al. Retrieval and model construction of water quality parameters for UAV hyperspectral remote sensing[J]. 2020,45(9):60-64,95.
[5] 臧传凯,沈芳,杨正东.基于无人机高光谱遥感的河湖水环境探测[J]. 自然资源遥感, 2021,33(3):45-53. Zang C K, Shen F, Yang Z D. Aquatic environmental monitoring of inland waters based on UAV hyperspectral remote sensing[J]. Remote Sensing for Natural Resources, 2021,33(3):45-53.
[6] 丁波,李伟,胡克.基于同期光学与微波遥感的茅尾海及其入海口水体悬浮物反演[J]. 自然资源遥感, 2022,34(1):10-17. Ding B, Li W, Hu K. Inversion of total suspended matter concentration in Maowei Sea and its estuary, Southwest China using contemporaneous optical data and GF SAR data[J]. Remote Sensing for Natural Resources, 2022,34(1):10-17.
[7] 盖颖颖,王章军,杨雷,等.金沙滩近岸水体叶绿素a和悬浮物遥感反演研究[J]. 国土资源遥感, 2020,32(3):129-135. Gai Y Y, Wang Z J, Yang L, et al. Remote sensing retrieval of chlorophyll -a and suspended matter in coastal waters of Golden Beach[J]. Remote Sensing for Land and Resources, 2020,32(3):129-135.
[8] 禹定峰,周燕,马万栋,等.基于HICO模拟数据的杭州湾水体悬浮物浓度遥感反演[J]. 国土资源遥感, 2018,30(4):171-175. Yu D F, Zhou Y, Ma W D, et al. Retrieval of total suspended matter concentration in Hangzhou Bay based on simulated HICO from in situ hyperspectral data[J]. Remote Sensing for Land and Resources, 2018, 30(4):171-175.
[9] Silveira Kupssinskü L, Thomassim Guimarães T, Menezes de Souza E, et al. A method for chlorophyll-a and suspended solids prediction through remote sensing and machine learning[J]. Sensors, 2020,20(7):2125.
[10] Peterson K T, Sagan V, Sidike P, et al. Suspended sediment concentration estimation from Landsat imagery along the Lower Missouri and Middle Mississippi Rivers using an extreme learning machine[J]. Remote Sensing, 2018,10(10):1503.
[11] Nazeer M, Bilal M, Alsahli M M M, et al. Evaluation of empirical and machine learning algorithms for estimation of coastal water quality parameters[J]. ISPRS International Journal of Geo-Information, 2017,6(11):360.
[12] 王思梦,秦伯强.湖泊水质参数遥感监测研究进展[J]. 环境科学, 2023,44(3):1228-1243. Wang S M, Qin B Q. Research progress on remote sensing monitoring of lake water quality parameters[J]. Environmental Science, 2023,44(3):1228-1243.
[13] 杨宇锋,武暕,王璐,等.基于随机森林模型的辽河高时间分辨率氮、磷浓度模拟与预测[J]. 环境科学学报, 2022,42(12):384-391. Yang Y F, Wu J, Wang L, et al. Simulation and prediction of nitrogen and phosphorus concentrations with high time resolution in Liao River using random forest model[J]. Acta Scientiae Circumstantiae, 2022, 42(12):384-391.
[14] Pahlevan N, Smith B, Alikas K, et al. Simultaneous retrieval of selected optical water quality indicators from Landsat-8, Sentinel-2, and Sentinel-3[J]. Remote Sensing of Environment, 2022,270:112860.
[15] 嵇晓燕,杨凯,陈亚男,等.基于ARIMA和Prophet的水质预测集成学习模型[J]. 水资源保护, 2022,38(6):111-115. Ji X Y, Yang K, Chen Y N, et al. An ensemble learning model for water quality forecast based on ARIMA and Prophet. Water Resources Protection, 2022,38(6):111-115.
[16] 王成贺,宋宁,王京禹,等.基于时空演变多重特性建模的近海叶绿素浓度时序预测[J]. 信号处理, 2022,38(6):1232-1239. Wang C H, Song N, Wang J Y, et al. Temporal prediction of chlorophyll concentration in coastal waters based on multi-characteristics modeling of spatio-temporal evolution[J]. Journal of Signal Processing, 2022,38(6):1232-1239.
[17] 李雪清,郑航,刘悦忆,等.基于多源数据机器学习的区域水质预测方法研究[J]. 水利水电技术(中英文), 2021,52(11):152-163. Li X Q, Zheng H, Liu Y Y, et al. Multi-source data machine learning-based study on method for regional water quality prediction[J]. Water Resources and Hydropower Engineering, 2021,52(11):152-163.
[18] 陈点点,陈芸芝,冯险峰,等.基于超参数优化CatBoost算法的河流悬浮物浓度遥感反演[J]. 地球信息科学学报, 2022,24(4):780-791. Chen D D, Chen Y Z, Feng X F, et al. Retrieving suspended matter concentration in rivers based on hyperparameter optimized CatBoost algorithm[J]. Journal of Geo-information Science, 2022,24(4):780-791.
[19] DeLuca N M, Zaitchik B F, Curriero F C. Can multispectral information improve remotely sensed estimates of total suspended solids? A statistical study in Chesapeake Bay[J]. Remote Sensing, 2018,10(9):1393.
[20] Dehkordi A T, Ghasemi H, Zoej M J V. Machine learning-based estimation of suspended sediment concentration along Missouri River using remote sensing imageries in Google Earth Engine[C]//20217th International Conference on Signal Processing and Intelligent Systems (ICSPIS). IEEE, 2021:1-5.
[21] Kwon S, Seo I W, Noh H, et al. Hyperspectral retrievals of suspended sediment using cluster-based machine learning regression in shallow waters[J]. Science of The Total Environment, 2022,833:155168.
[22] Abdi H. Partial least squares regression and projection on latent structure regression (PLS Regression)[J]. WIREs Computational Statistics, 2010,2(1):97-106.
[23] Wold S, Sjöström M, Eriksson L. PLS-regression:A basic tool of chemometrics[J]. Chemometrics and Intelligent Laboratory Systems, 2001,58(2):109-130.
[24] Hoerl A E, Kennard R W. Ridge regression:Biased estimation for nonorthogonal problems[J]. Technometrics, 1970,12(1):55-67.
[25] Hoerl A E, Kennard R W. Ridge regression:applications to nonorthogonal problems[J]. Technometrics, 1970,12(1):69-82.
[26] Tibshirani R. Regression shrinkage and selection via the lasso[J]. Journal of the Royal Statistical Society. Series B (Methodological), 1996,58(1):267-288.
[27] Tibshirani R. The lasso method for variable selection in the Cox model[J]. Statistics in Medicine, 1997,16(4):385-395.
[28] Drucker H, Burges C J C, Kaufman L, et al. Support vector regression machines[C]//Advances in Neural Information Processing Systems. MIT Press, 1996.
[29] Suykens J A K, Vandewalle J, De Moor B. Optimal control by least squares support vector machines[J]. Neural Networks, 2001,14(1):23-35.
[30] Kamiński B, Jakubczyk M, Szufel P. A framework for sensitivity analysis of decision trees[J]. Central European Journal of Operations Research, 2018,26(1):135-159.
[31] Cover T, Hart P. Nearest neighbor pattern classification[J]. IEEE Transactions on Information Theory, 1967,13(1):21-27.
[32] Breiman L. Bagging predictors[J]. Machine Learning, 1996,24:123-140.
[33] Breiman L. Random forests[J]. Machine Learning, 2001,45(1):5-32.
[34] Painsky A, Rosset S. Cross-validated variable selection in tree-based methods improves predictive performance[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,39(11):2142-2153.
[35] Freund Y, Schapire R E. A decision-theoretic generalization of on-line learning and an application to boosting[C]//Vitányi P, ed.//Computational Learning Theory. Berlin, Heidelberg:Springer, 1995:23-37.
[36] Piryonesi S M, El-Diraby T E. Data analytics in asset management:Cost-effective prediction of the pavement condition index[J]. Journal of Infrastructure Systems, 2020,26(1):04019036.
[37] Wolpert D H. Stacked generalization[J]. Neural Networks, 1992,5(2):241-259.
[38] Breiman L. Stacked regressions[J]. Machine Learning, 1996,24(1):49-64.
[39] Littlestone N, Warmuth M K. The weighted majority algorithm[J]. Information and Computation, 1994,108(2):212-261.
[40] 周志伟,邓化,施华宏.基于无人机遥感与机器学习的岸滩大型塑料垃圾监测方法[J]. 海洋环境科学, 2023,42(1):141-150. Zhou Z W, Deng H, Shi H H. Monitoring of marine macro-plastic litter in the coastal zone based on UAV remote sensing and machine learning. Marine Environmental Science, 2023,42(1):141-150.
[41] 毛正君,石硕杰,王贵荣,等.基于无人机遥感的区域黄土滑坡识别及发育特征分析[J]. 灾害学, 2023,38(1):63-71. Mao Z J, Shi S J, Wang G R, et al. Identification and development characteristics analysis of loess landslide based on UAV remote sensing[J]. Journal of Catastrophology, 2022,38(1):63-71.
[42] 申华磊,苏歆琪,赵巧丽,等.基于深度学习的无人机遥感小麦倒伏面积提取方法[J]. 农业机械学报, 2022,53(9):252-260,341. Shen H L, Shu X Q, Zhao Q L, et al. Extraction of lodging area of wheat varieties by unman[J]. Transactions of the Chinese Society for Agricultural Machinery, 2022,53(9):252-260,341.
[43] 谢运鸿,荆雪慧,孙钊,等.基于实例分割的高郁闭度林分单木树冠无人机遥感提取[J]. 林业科学研究, 2022,35(5):14-21. Xie Y H, Jing X H, Sun Z, et al. Tree crown extraction of UAV remote sensing high canopy density stand based on instance segmentation. Forest Research, 2022,35(5):14-21.
[44] Dethier E N, Renshaw C E, Magilligan F J. Rapid changes to global river suspended sediment flux by humans[J]. Science, 2022,376(6600):1447-1452.
[45] Overeem I, Hudson B D, Syvitski J P M, et al. Substantial export of suspended sediment to the global oceans from glacial erosion in Greenland[J]. Nature Geoscience, 2017,10(11):859-863.
[46] Park E, Latrubesse E M. Modeling suspended sediment distribution patterns of the Amazon River using MODIS data[J]. Remote Sensing of Environment, 2014,147:232-242.
[47] Umar M, Rhoads B L, Greenberg J A. Use of multispectral satellite remote sensing to assess mixing of suspended sediment downstream of large river confluences[J]. Journal of Hydrology, 2018,556:325-338.
[48] Sabzekar M, Hasheminejad S M H. Robust regression using support vector regressions[J]. Chaos, Solitons & Fractals, 2021,144:110738.
[49] Yu Q, Liu Y, Rao F. Parameter selection of support vector regression machine based on differential evolution algorithm[C]//2009Sixth International Conference on Fuzzy Systems and Knowledge Discovery. Tianjin, China:IEEE, 2009:596-598.
[50] Moses D, Deepa B, Patri T, et al. A review of decision tree algorithms for predictive analysis in data mining[J]. Journal on Software Engineering, 2017,12(1).
[51] Alsagheer R H, Alharan A F, Al-Haboobi A S. Popular decision tree algorithms of data mining techniques:a review[J]. International Journal of Computer Science and Mobile Computing, 2017,6(6):133-142.
[52] Losing V, Hammer B, Wersing H. KNN classifier with self adjusting memory for heterogeneous concept drift[C]//2016IEEE 16th International Conference on Data Mining (ICDM). IEEE, 2016:291-300.
[53] Bagnall A, Lines J, Bostrom A, et al. The great time series classification bake off:a review and experimental evaluation of recent algorithmic advances[J]. Data Mining and Knowledge Discovery, 2017,31(3):606-660.
[54] Zhou Z-H. Ensemble methods:foundations and algorithms[M]. CRC press, 2012.
[55] Romero-Trigueros C, Nortes P A, Alarcón J J, et al. Effects of saline reclaimed waters and deficit irrigation on Citrus physiology assessed by UAV remote sensing[J]. Agricultural Water Management, 2017, 183:60-69.