Applications of Remote Sensing for Sargassum on Da Ya Bay

Applications of Remote Sensing for Sargassum on Da Ya Bay

Li Tiefang, Yi Jianchun
Center for Remote, Zhongshan University, P.R.C

Liu Huai, Fang Hongda
Shouth China Sea Marine Environmental Monitoring Center, SOA

Dong Yuguo
Guangzhou Institute of New Techniques in Geologym

Academic Sinica,Chen Xuelian
Scientific Research Institute , Pearl River Water Resources
Commission, P.R.C.

Abstract
There is a lot of sargassum growing in the sea area of DaYa Bay. Sargassumk is a type of big algae with body length about 1~2 meters and the longest about 5~6 meters. The sargassum begins to grow in fall of every year, and breaks in April of May of the next year and floats away with sea currents. It is the major clogged material to the cooling system pipe of DaYa Bay Nuclear Power Station. To provide the design of the filter lack and drum with reliable data, it's necessary to investigate its culture regularity, distribution, output and floating quantity toward cooling system pipe on the Bay. As the Bay areas are very wide, it's quite difficult to determine the distributional range and the total production on the basis of conventional oceanic research, but remote sensing investigation is a very effective method. The paper discusses the methods of using LANDSAT TM data with the aid of on-the-spot research to recognize the sargassum distributional range and estimate its output.

The acquisition of remotely sensed sargassum information

The Sargassum Distributional Characteristics

Sargassum is an algae of fixing life, grows on the low water line down to gravel 5 meters deep under water, and distributes in the states of piece or bar. The water body is usually cleaner when sargassum grows.
Sargassum Wavebands Characteristics

In the spectrum ranges of 0.4 ~ 0.7 um, the sargassum has its own obvious absorption and reflection bands, which are important marks of recognizing sargassum. The average brightness curves on TM images are shown as Fig. 1, which shows that TM1 wavebands have deeper

Penetration and more information contents than that of TM2 wavebands. Although the noise on TM1 image caused by atmosphere is much greater than that on TM2, the noise only affects the shift of the brightness distributional curve, not on its form and its message recognization.

The Acquisition of Sargassum Messages

Based on the sargassum distributional characteristics and waveband features, TM1 images in two periods, one of which is before sargassum grows (October), the other , after the sargassum grows up (January), are chosen to be processed differentially. The image brightness on gravel beach, on which there is no sargassum growing , is the highest. Otherwise, its image brightness will be much lower on the on the TM image when sargassum lives. The difference of brightness between two periods is very small in the area where no sargassum grows . As a matter of fact, the greater difference will occur only in the case of the current in various current speeds in this area from the above analysis, the sargassum massage could be picked up. The red algae and sea current have the same differential value of spectrum. Several differential values of typical bottom materials, current and algaes are shown below:

Type	Muddy bed material	Sandy bed Material	Current	Sand beach	Sargassum	Red algae
	material	Material		Beach
D.V.	0~5	0~3	>10	0~3	8~10	8~10
D.V --- Differential Values of Brghtness

It can be seen from the above table that most of the no-sargassum messages could be removed after the different processing. The possible confusing messages will occur in red algae and currents. But when the states of differential spectrum, original spectrum of sargassum, red algae, current, terrain and topography are applied to the message processing , they could be obviously told apart. As red algae is a type of small algae growing on the water bed of mud with lower current speed, its imagery brightness value is much darker than that of sargassum living on the gravel beach. Basides , the current speed in the seas of red algae growing is slower than of that of sargassum growing , and its imagery texture smoother and colourgrade, darker. So, the sargassum distributional range can be decided and different growing densities can be distinguished. The comprehensive analysis model is given below:

Us(s) = Utm1(D)AUg(G)AUtm1(TM)AUt(T) ----------------(1)

Where:
Us(S) -- The sargassum discriminant function.
Utm1(D) - The differential value of images before and after sargassum growing.
Ug(G) -- The membership function of terrain and topography (gravel beach) for sargassum living.
Utm1(tm1)-The brightness function of TM1 image (original spectrum).
Ut(T) -- Characteristic function of imagery texture for sargassum growing (consist of the textures of piece or bar spottedly).
A --" AND" operator, means the artificial intelligence processing of compuper recognization for differential value processing.

The sargassum distributional range and growing densities determined by the above methods are as Fig 2.

Sargassum products estimation

Sargassum Output Estimation Model

The different growing densities of sargassum have difference in brightness and differential value. The higher of the sargassum growing grades are, the darker the brightness, and the larger their differential vlaues. On the other hand, the lower sargassum growing grades are, the brightener the and the smaller the differential value. In order to estimate the sargassum output, the on-the-spot sampling and weighing are carried based upon different densities shown on the image to decide the quantities of the density grades on the image. The sargassum output Gs follows:

Gs = (C1. L1. + C2.L2 + C3.L3) . Pa. gs -----------------------(2)

Where:
C1,C2,C3-- The pixels colour grade (Red, Green and Blue).
L1,L2,L3-- The rate of sargassum length at the sampling area.
pa -- The spatial -resolution of remotely sensed image (m2)
gs -- The average output in unit area at sampling seas.

The above sargassum output estimating model is the accompaniment of remote sensing technique with actual field sampling and surveying . L1, L2, and L3, could be regarded as the weight coefficients on the sargassum desnities Ci, which coame from the remotely sensed images , also as the calibration value. From the equation (2), the sargassum output in all of DaYa Bay is about 1700T
The Predication of Broken Sargassum Drifting Route.

The sargassum, which breaks in April or May every year, would float away with tidal currents in side the Bay. The drifting route and hold-up time are the key problem for the clogging of cooling system pipe of the power station. The drifting route has much to do with tidal currents and ti usually mapped out thought the analysis of very few points at actual station surveying. Because the ordinary current analysis method couldn't reflect at the current state and it's details inside the Bay length, it's quite difficult to foretell the broken sargassum drifting routes at every mouth. The TM image at tidal flood and ebb are chosen to pick up surface current message to analyze the broken sargessum's floating routes. The trace and state of surface current message to analyze the broken sargassum's floating routes. The trace and state of surface currents, which are shown in great details on remotely sensed images, are the dynamic factors of sargassum drifting. This is because the surface current field (speed and direction) has an association with the brightness and texture on remotely sensed images. [1], [2]

As a matter of fact, the higher the current sped is, the much rougher the surface state and the brighter the imagery brightness. In the contrary, the slower the current speed is, the darker the imagery brightness. Based upon the statistic results of TM imagery at the same tidal and meteorological condition with the surveying of current speed and direction on-the-sopt, the corrlelation coefficient of pixle's brightness with the current speed is about 0.7 ~ 0.8. Also, where water flows, the surface roughness, suspended sediment, water colour and so on, would be the marks of current trace shown on the image , and the state of the interaction of different water bodies or curren5 system, such as circumfluent, eddy, shearing, and so on, would be depicted on TM imagery explicitly. As a result, the surface current field of flood and ebb current inside the Bay could be mapped out on Fig. 3. By the analysis of sargassum floating direction and holdup condition at every bay mouth, the sargassum, which is dangerous for the cooling system pipe of the power station , is estimated at about 85% of the total output on DaYa Bay. They are mainly from the west of Central Islets.

Fig. 3 Current field map of Da Ya Bay's tidal flood and ebb current

Results and Discussion
So far, remote sensing for sargassum has been applied only in 2 or 3 countries in the world. The Characteristics of the method discussed in the paper are as following:

The scientific combination of oceanic remote sensing and biological remote sensing with under-water topography.
The sargassum output estimating model established by the proper combination of remote sensing method with the ordinary oceanographic research.
The recognization and analysis combining water surface remote sensing massage with under water remote sensing information.

Thanks to the combination of multi-massages with different processing methods, the remote sensing of sargassum on DaYa Bay has good results and it would provide the engineering design of the filter lock and drum with reliable data.

Acknowledgments
The original TM images were provided by the Remote Sensing Satellite Ground Station of Chinese Academy of Science. This research was also partly supported by Mr. Ou Huamin, Mr. Lin Zuheng, Ms. Li Jianrong, Mr. Jiang Yuejin, of South China Sea Environmental Monitoring Center of SOA and Mr. Li Yinxi, Mr. Ma Yachuan, of Remote Sensing Satellite Ground Station of Chinese Academy of science.

References

Robert H. Stewart, Methods of Satellite Oceanography, Published under the auspices of the California, p108-113.
A.P. Crecknel, Remote Sensing In Meteorology, Oceanography and Hydrology, first published by Ellis Horwoods Limited in 1981, p 178-204.
Ifrmer, seaweeds Northe Brittary Harvesting Area, France Department Environment Littoral Service Applications data Teledetection 1987, p1-6.
Pan Youlian, Chlorophy1 and Primary Productivity, Marine Science, No.1, 1987 (In Chinese).