基于CNN-SVR网络的黄渤海近岸海域叶绿素a浓度预测-海洋预报

基于CNN-SVR网络的黄渤海近岸海域叶绿素a浓度预测

作者：王晓霞¹ 2 3 汪健平¹ 3 王佳莹¹ 3 孙珊¹ 3 苏博¹ 3 姜会超¹ 3 朱明明¹ 3

单位：1. 山东省海洋资源与环境研究院, 山东烟台 264006;
2. 自然资源部空间海洋遥感与应用重点实验室, 北京 100081;
3. 山东省海洋生态修复重点实验室, 山东烟台 264006

关键词：卷积神经网络结合支持向量回归模型叶绿素a浓度预测单因子敏感性分析海洋卫星海洋生态水质因子

分类号：X145

出版年·卷·期（页码）：2024·41·第四期（77-87）

摘要：

利用海洋卫星观测数据和黄渤海近岸海域实测生态水质数据，建立了一种基于卷积神经网络结合支持向量回归（Convolutional Neural Network-Support Vector Regression，CNN-SVR）的深度学习网络模型的叶绿素a浓度预测方法。采用皮尔逊方法对叶绿素a与环境动力因子和生态水质因子作相关分析，发现营养盐因子大多与叶绿素 a有显著相关性，水质因子如 pH、溶解氧、盐度等与叶绿素a的相关性不大；将黄渤海近岸海域划分为渤海南部与黄海北部、黄海中部，进行春夏、秋冬两个时期 1×1和 2×2两种卷积核大小的 CNN-SVR网络模型实验以及单因子敏感性分析试验。结果显示：卷积核大小为2×2时，CNN-SVR网络模型对训练数据的学习和对测试样本的预测检验效果都更优；渤海南部与黄海北部近岸海域模型预测效果更好。营养盐因子对模型预测能力的影响更显著，悬浮物等水质因子的影响相对较弱。单变量对模型预测的敏感性较弱，多变量整合具有互补性，改善了模型的预测效果。

A chlorophyll-a (Chl-a) concentration prediction method based on the Convolutional Neural NetworkSupport Vector Regression (CNN-SVR) model is developed using satellite observations and in-situ ecological water quality measurements in near-shore waters of the Yellow and Bohai Seas. Firstly, we use Pearson method to establish correlation between Chl-a concentration and factors of environmental dynamics and ecological water quality. It is found that Chl-a concentration correlates significantly with nutrient salt factors, while poorly with water quality factors such as pH, dissolved oxygen, salinity. Then, we divide two regions, one is nearshore waters of the southern Bohai Sea and northern Yellow Sea, and the other one is nearshore waters of the central Yellow Sea. We also divide two periods: spring — summer and autumn — winter. We perform the CNN-SVR model experiments with two convolutional kernel sizes, 1×1 and 2×2, as well as the single factor sensitivity analysis experiment. The results show that the CNN-SVR network model has better learning of the training data and better prediction of the test samples when the convolution kernel size is 2×2. The CNN-SVR network model performs better in nearshore areas of the southern Bohai Sea and northern Yellow Sea. Compared to water quality factors, the nutrient salt factors have larger impacts on the model's prediction ability. The sensitivity of single factor to model's prediction ability is weak, while multiple variables exhibit complementary feature which improves the model's prediction ability.

参考文献：

[1] 刘建萍, 张玉超, 钱新, 等. 太湖叶绿素a浓度预测模型初探[J]. 环境保护科学, 2009, 35(4): 46-49. LIU J P, ZHANG Y C, QIAN X, et al. Preliminary studies on the prediction model of chlorophyll-a in Taihu Lake[J]. Environmental Protection Science, 2009, 35(4): 46-49.
[2] AHN O, PETRELL R J, HARRISON P J. Ammonium and nitrate uptake by Laminaria saccharina and Nereocystis luetkeana originating from a salmon sea cage farm[J]. Journal of Applied Phycology, 1998, 10(4): 333-340.
[3] KIM J H, KANG E J, PARK M G, et al. Effects of temperature and irradiance on photosynthesis and growth of a green-tide-forming species (Ulva linza) in the Yellow Sea[J]. Journal of Applied Phycology, 2011, 23(3): 421-432.
[4] 苑克磊, 侯查伟, 丁一, 等. 我国黄海浒苔绿潮暴发海域环境因子季节性变化及年际变化分析[J]. 广西科学院学报, 2018, 34(3): 204- 209. YUAN K L, HOU C W, DING Y, et al. Analysis of the seasonal and inter-annual changes of environmental factors in Enteromorpha prolifera green tide outbreak in the Yellow Sea[J]. Journal of Guangxi Academy of Sciences, 2018, 34(3): 204-209.
[5] 夏斌, 马绍赛, 崔毅, 等. 黄海绿潮(浒苔)暴发区温盐、溶解氧和营养盐的分布特征及其与绿潮发生的关系[J]. 渔业科学进展,2009, 30(5): 94-101. XIA B, MA S S, CUI Y, et al. Distribution of temperature, salinity, dissolved oxygen, nutrients and their relationships with green tide in Enteromorpha prolifera outbreak area of the Yellow Sea[J]. Progress in Fishery Sciences, 2009, 30(5): 94-101.
[6] 庞秋婷, 李凤, 刘湘庆, 等. 围隔实验中浒苔在不同营养盐条件下的生长比较[J]. 环境科学, 2013, 34(9): 3398-3404. PANG Q T, LI F, LIU X Q, et al. Compare the growth of Enteromorpha prolifera under different nutrient conditions[J]. Environmental Science, 2013, 34(9): 3398-3404.
[7] 吴婷, 田小亭, 刘召辉, 等. 浒苔对不同氮源吸收特性的初步研究[J]. 海洋科学, 2011, 35(4): 51-54. WU T, TIAN X T, LIU Z H, et al. Absorption of different nitrogen species by Enteromorpha prolifera[J]. Marine Sciences, 2011, 35(4): 51-54.
[8] 孙晓. 黄海绿潮和叶绿素a浓度的遥感时空分异及两者响应机制[D]. 烟台: 鲁东大学, 2018. SUN X. Spatial and temporal variations and response mechanism of green tide and chlorophyll a concentrations based on remote sensing in southern Yellow Sea[D]. Yantai: Ludong University, 2018.
[9] 汪益嫔, 张维砚, 徐春燕, 等. 淀山湖浮游植物初级生产力及其影响因子[J]. 环境科学, 2011, 32(5): 1249-1256. WANG Y P, ZHANG W Y, XU C Y, et al. Phytoplankton productivity and its influencing factors in Dianshan Lake[J]. Environmental Science, 2011, 32(5): 1249-1256.
[10] 金相灿, 刘鸿亮, 屠清瑛, 等. 中国湖泊富营养化[M]. 北京: 中国环境科学出版社, 1990. JIN X C, LIU H L, TU Q Y, et al. Eutrophication in lake of China [M]. Beijing: China Environmental Science Press, 1990.
[11] NGUYEN H Q, HA N T, NGUYEN-NGOC L, et al. Comparing the performance of machine learning algorithms for remote and in situ estimations of chlorophyll-a content: a case study in the Tri An Reservoir, Vietnam[J]. Water Environment Research, 2021, 93(12): 2941-2957.
[12] MAIER P M, KELLER S. Estimating chlorophyll a concentrations of several inland waters with hyperspectral data and machine learning models[J]. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2019, IV-2-W5: 609-614.
[13] ZHANG T. Estimation of chlorophyll-a concentration of lakes based on SVM algorithm and Landsat 8 OLI images[J]. Environmental Science and Pollution Research, 2020, 27(13): 14977-14990.
[14] JIN D, LEE E, KWON K, et al. A deep learning model using satellite ocean color and hydrodynamic model to estimate chlorophyll-a concentration[J]. Remote Sensing, 2021, 13(10): 2003.
[15] APTOULA E, ARIMAN S. Chlorophyll-a retrieval from Sentinel- 2 images using convolutional neural network regression[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 6002605.
[16] XIA J J, ZENG J. Environmental factor assisted chlorophyll-a prediction and water quality eutrophication grade classification: a comparative analysis of multiple hybrid models based on a SVM [J]. Environmental Science: Water Research & Technology, 2021, 7(6): 1040-1049.
[17] WANG S Q, HUAN Y, QIU Z F, et al. Remote sensing of particle cross-sectional area in the Bohai Sea and yellow sea: algorithm development and application implications[J]. Remote Sensing, 2016, 8(10): 841.
[18] ZHANG M W, TANG J W, SONG Q J, et al. Backscattering ratio variation and its implications for studying particle composition: a case study in Yellow and East China seas[J]. Journal of Geophysical Research: Oceans, 2010, 115(C12): C12014.
[19] 姜玲玲, 王龙霄, 王林, 等. 基于Sentinel-3 OLCI影像的渤海透明度遥感反演研究[J]. 光谱学与光谱分析, 2022, 42(4): 1209-1216. JIANG L L, WANG L X, WANG L, et al. Research on remote sensing retrieval of Bohai sea transparency based on sentinel-3 OLCI image[J]. Spectroscopy and Spectral Analysis, 2022, 42(4): 1209-1216.
[20] 何恩业, 王丹, 黄莉, 等. 西太平洋副热带高压的变动对我国赤潮发生的影响分析[J]. 海洋预报, 2015, 32(4): 83-89. HE E Y, WANG D, HUANG L, et al. Impact of the west pacific subtropical high on the red tide in the China seas[J]. Marine Forecasts, 2015, 32(4): 83-89.
[21] 黄岁樑, 臧常娟, 杜胜蓝, 等. pH、溶解氧、叶绿素a之间相关性研究Ⅰ: 养殖水体[J]. 环境工程学报, 2011, 5(6): 1201-1208. HUANG S L, ZANG C J, DU S L, et al. Study on the relationships among pH, dissolved oxygen and chlorophyll a I: aquaculture water[J]. Chinese Journal of Environmental Engineering, 2011, 5(6): 1201-1208.
[22] 张建乐, 张心宇, 张永丰, 等. 北戴河海域夏季叶绿素a分布特征及其与pH、溶解氧的关系[J]. 海洋湖沼通报, 2015(4): 155-160. ZHANG J L, ZHANG X Y, ZHANG Y F, et al. Distribution of chlorophyll a of the coastal water of beidaihe and its relationship with pH and dissolved oxygen in summer[J]. Transactions of Oceanology and Limnology, 2015(4): 155-160.
[23] 吴浠. 基于卷积神经网络的推荐算法[D]. 广州: 华南理工大学, 2016. WU X. Recommendation algorithm based on convolutional neural network[D]. Guangzhou: South China University of Technology, 2016.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】