SHALLOW BATHYMETRY ESTIMATION BASED ON LANDSAT 8 REMOTELY SENSED DATA AT BOHAI SEA

Bathymetry is a key variable in ocean monitoring and measurement research. It becomes more and more important for development of rapid method to invert shallow sea water depth. In this study, a water depth inversion method based on multi-band model is established to analyze the relationship between different bands of Landsat 8 OLI multi-spectral and measured data. The average absolute error of the model is 1.48m at 10-20m water depth and the average relative error is 13.12%. The water depth inversion accuracy under normal conditions are achieved, indicating that the model will have a promising practical application in the future.


INTRODUCTION
As an important indicator of ocean measurement, water depth is also a topographical factor that cannot be ignored in shallow waters [1][2][3]. It has a non-negligible effect on island coastal zone management and development, maritime shipping and transportation, and marine engineering projects. Water depth measurement is one of the most basic tasks in marine topographic survey, and it is also one of the important contents of docks, ports, anchorage construction and navigation channels.
The estimation methods can be divided into active and passive remote sensing methods [4] [5]. According to the characteristics of water depth inversion, it can be divided into hyper-spectral remote sensing water depth inversion, Lidar sounding and shallow seawater deep fusion detection. The most advantage of active remote sensing methods is the high accuracy of the obtained data. However, it is extremely expensive and labor-waste. Compared with active remote sensing, passive methods can achieve measurements in a variety of sea areas with fast speed and large area. The only drawback is that the accuracy is not so high for those cases where the accuracy requirements are very strict.

2.METHOD
The functional model of the measured water depth data and the reflectance of 11 different bands in the Where Z is the measured water depth data; X i is the reflectivity of the i-band.

Study area
The Bohai Sea is an inland sea. Compared with the Yellow Sea, the East China Sea and the South China Sea, it has many shallow sea areas. Its latitude is 37°07′~41°0′N, and the longitude is 117°35′ ～ 121°10′E (Figure 1). The coast is inclined to the center and the strait, the terrain is monotonous and gentle. And the water depth in the shallow waters is obviously different. Therefore, it is suitable to select the Bohai Sea as a study area for shallow seawater depth inversion research.

Data
The remote sensing images used in this study are

Discussion
The estimation results are influenced by many aspects. These factors should be considered as following: (1) The measured water depth data. The human activities in the shallow seas of the Bohai Sea are relatively frequent. The discharge of human pollutants, the dumping of marine debris, and the discharge of sewage from chemical parks have caused the rapid change for shallow water depth of Bohai Sea.
(2) The optical characteristics of the Bohai Sea itself.
Bohai's own marine optical characteristics are not only related to its internal properties, but also affected by the discharge of pollutants from the chemical parks and ships in the ports around the Bohai Sea. Excess organic matter will cause a large number of phytoplankton to rapidly multiply, produce excessive suspended matter and unknown colored substances, which will cause remote sensing inversion of water depth.
(3) Omissions in the data preprocessing process.
Coupled with the impact of atmospheric suspended matter on the blue-green band, the corrections have not been completely resolved, reducing the practical significance of our calibration process by more than half. As can be seen from the figure, the green part of the coastal zone cannot reflect the actual water depth due to the atmospheric correction influence.

CONCLUSION
In this paper, the Landsat 8 OLI multi-spectral