基于BP ANN和B算法的黄河口水深遥感比较研究  

Comparative study on water depth Remote Sensing in Yellow River Estuary based on BP ANN method and Bottom Albedo-independent Bathymetry Algorithm

作　　者：王正[1,2,3] 毛志华[3] 曾群[4] 田礼乔[5] 樊勇[6,7] 

机构地区：[1]南京大学地理与海洋科学学院,南京210023 [2]中国南海研究协同创新中心,南京210093 [3]国家海洋局第二海洋研究所卫星海洋环境动力学国家重点实验室,杭州310012 [4]华中师范大学学报编辑部,武汉430079 [5]武汉大学测绘遥感与信息工程国家重点实验室,武汉430079 [6]信阳师范学院城市与环境科学学院,河南信阳464000 [7]武汉大学资源与环境科学学院,武汉430079

出　　处：《华中师范大学学报（自然科学版）》2016年第1期112-119,共8页Journal of Central China Normal University：Natural Sciences

基　　金：国家海洋公益项目(201005030);国家自然科学基金项目(41476156;41321004);教育部人文社科青年基金项目(14YJCZH028)

摘　　要：基于Landsat8 OLI传感器数据,用BP人工神经网络模拟水深法和底部反照率独立水深测量算法(bottom albedo-independent bathymetry algorithm,简称B算法)来反演黄河口水深,并与实测水深数据进行比较;然后,基于两种水深反演方法的结果进行了对比分析和适用性评价.发现,用BP神经网络模拟方法提取的水深在近岸水深小于15m的区域和水深大于15m的区域,反演结果与实测水深误差较大,变化趋势不一致,实测值与模拟水深值相关系数低;以B算法提取的水深与实测水深误差较小,在趋势上一致,相关系数为0.899;并基于该算法的表现,通过水深遥感制图进一步在实际应用中验证了该水深反演方法.结果表明,底部反照率独立测深算法精度高、效果好,比较适用于黄河口水深探测.Back propagation Artificial Neural Network（BP-ANN）method and Bottom Albedo-independent Bathymetry Algorithm（B algorithm）were used to derive the water depth in Yellow River Estuary based on the Landsat 8OLI sensor＇s data in this paper.Besides,the derived data and in-situ measurement data were compared and analyzed,from which we find that：there have error-large and change trend-inconsistencies problems during the BP-ANN derive and in-situ measurement water depth in the areas the water depth within and above 10m;while B algorithm has a good performance on the error and change trends aspects,especially in the area which the water depth within and above 10 meters.In addition,the correlation coefficients between B algorithm derive and in-situ measurement data was 0.899.Mapping the water depth is a further proof of B algorithm based on the good performance of which.The results show that B algorithm which have high accuracy and better effect is suitable approach for the water depth derive in the Yellow River Estuary.

关 键 词：人工神经网络 底部反照率独立水深测量算法 黄河口 水深遥感 Landsat8 OLI 

分 类 号：P237[天文地球—摄影测量与遥感] X87[天文地球—测绘科学与技术]