| 基于BP神经网络渤海湾表层叶绿素浓度反演方法探讨 |
| Discussion on retrieval method of surface chlorophyll concentration of Bohai Bay based on BP neural network |
| 投稿时间：2018-09-07 修订日期：2018-12-03 |
| DOI： |
| 中文关键词: GOCI 叶绿素浓度 BP神经网络 渤海湾 遥感 |
| 英文关键词:GOCI chlorophyll concentration BP neural network Bohai Bay remote sensing |
| 基金项目:天津市自然科学基金项目（14JCYBJC22500）；国家科技支撑项目（2010BAC68B04） |
|
| 摘要点击次数: 81 |
| 全文下载次数: 0 |
| 中文摘要: |
| 由于渤海湾水质光学性质复杂，使用传统线性叶绿素浓度反演方法进行反演时的精度不高，为提高渤海湾叶绿素浓度的反演精度，本文利用2012年4月渤海湾的表层叶绿素浓度实测数据和GOCI同步遥感反射率数据，建立了渤海湾叶绿素浓度反演的BP神经网络模型，并与GOCI的三种叶绿素浓度业务算法（OC2算法、OC3G算法和YOC算法）进行比较。结果表明，在渤海湾GOCI三种业务算法的标准误差均大于1，决定系数均小于20%；BP神经网络算法的标准误差为0.6154，决定系数为89.98%，且反演的叶绿素浓度分布与实测值的分布趋势一致，反演的精度高于GOCI传统业务算法。因此该BP神经网络模型可用于反演春季渤海湾表层叶绿素浓度，且反演精度比GOCI业务反演算法精度高。 |
| 英文摘要: |
| Due to the complexity of the optical properties of the water quality in the Bohai Bay, traditional linear chlorophyll concentration retrieval methods are not accurate enough in Bohai Bay. In order to improve the retrieval accuracy of chlorophyll concentration in the Bohai Bay, a BP neural network model for the chlorophyll concentration retrieval in the Bohai Bay was established based on the in-situ chlorophyll concentration in April 2012 and the synchronous remote sensing reflectance of GOCI. The retrieval accuracy of BP neural network was compared with that of three GOCI operational algorithms (OC2, OC3G and YOC). The study results indicated that the standard error of GOCI operational algorithm is more than 1, and the determination coefficient is less than 20%. The standard error of BP neural network algorithm is 0.6154, the determination coefficient is 89.98%. And the distribution of chlorophyll concentration by BP neural network is consistent with that of the in-site chlorophyll concentration. In general, the BP neural network model could be used to estimate the surface chlorophyll concentration in Bohai Bay in spring and the accuracy of BP neural network model is better than that of GOCI algorithm. |
| 查看/发表评论 下载PDF阅读器 |
