热带地理2025，Vol.45Issue(8)：1417-1428,12.DOI:10.13284/j.cnki.rddl.20240743
基于多源遥感数据的广西北部湾水产养殖活动监测及其环境效应
Monitoring of Aquaculture Activities in the Beibu Gulf of Guangxi Based on Multi-Source Remote Sensing Data and Its Environmental Effects
摘要
Abstract
Offshore aquaculture is an important marine economic activity;however,unreasonable aquaculture practices have led to the deterioration of the marine ecological environment.Dissolved inorganic nitrogen(DIN)and active phosphate(PO4)are important indicators for characterizing the health status of aquaculture areas,and chlorophyll a(Chl-a)is a key indicator of the biomass and eutrophication status of phytoplankton.In this study,the coastal zone of the Beibu Gulf in Guangxi Province was selected as the study area,and Sentinel-2 and Landsat-8 satellite images and water quality data were collected from 2014 to 2022.The spatial distribution characteristics of the aquaculture area were extracted using the kernel density method.A water quality remote sensing inversion model based on random forest was constructed,which had good prediction accuracy in the offshore low concentration area.The R² value of the model was 0.739-0.882,Approximately 70%of the water quality inversion results achieved R² values exceeding 0.8,and the fitting degree was high.The prediction performances for DIN,PO4,and Chl-a were excellent,and the R² values of the validation set were 0.858,0.882,and 0.872,respectively,indicating the stability and reliability of the model on different datasets.The prediction results in the scatter plot are close to the y=x line,indicating that the model better reflects the overall trend between the measured and predicted values.Both the RMSE and MAE values were low overall,ranging from 0.003 to 1.722,which is appropriate for remote sensing inversion of nutrient concentration and chlorophyll a concentration in the coastal waters of the Beibu Gulf,and better fits the nonlinear relationship between nutrient concentration and remote sensing reflectance data.This not only verifies the feasibility of machine-learning methods in remote sensing monitoring of water quality,but also provides an efficient and low-cost technical means for future water environment monitoring.The results showed that the intensity of aquaculture activities in Beibu Gulf has been increasing in recent years.Through the analysis of seasonal changes in water quality of the coastal waters of Beibu Gulf in 2022,it was found that there were significant temporal and spatial changes in the impact of aquaculture activities on water quality in different seasons and regions,and the concentrations of inorganic nitrogen and active phosphate in the high-density aquaculture area during summer were significantly higher than those in other regions.Based on the data analysis of the coastal zone of Beibu Gulf from 2014 to 2022,changes in water quality indicators(such as PO4)in the high-density aquaculture area showed an upward fluctuating trend and the water quality of that area showed a potential degradation trend.The results of this study provide a scientific basis for the prevention of pollution and environmental management in coastal aquaculture.关键词
水质反演/水产养殖/环境效应/Google Earth Engine/北部湾Key words
water quality inversion/aquaculture/environmental effects/Google Earth Engine/Beibu Gulf分类
资源环境引用本文复制引用
史兴峰,邓琰,蓝文陆,周文能,胡泓达,荆文龙,张灵波,杨骥,孙嘉,尹小玲,彭小燕..基于多源遥感数据的广西北部湾水产养殖活动监测及其环境效应[J].热带地理,2025,45(8):1417-1428,12.基金项目
科技基础资源调查专项(2023FY100805) （2023FY100805）
广东省基础与应用基础研究基金(2022A1515240041) （2022A1515240041）
广东省科技计划项目(2024B1212050010) （2024B1212050010）
