Spatial-Temporal Variations of Turbidity and Ocean Current Velocity of the Ariake Sea Area, Kyushu, Japan Through Regression Analysis with Remote Sensing Satellite Data

Regression analysis based method for turbidity and ocean current velocity estimation with remote sensing satellite data is proposed. Through regressive analysis with MODIS data and measured data of turbidity and ocean current velocity, regressive equation which allows estimation of turbidity and ocean current velocity is obtained. With the regressive equation as well as long term MODIS data, turbidity and ocean current velocity trends in Ariake Sea area are clarified. It is also confirmed that the negative correlation between ocean current velocity and turbidity.


INTRODUCTION
There are strong scientific demands on estimation of environmental situations including water quality of the ocean.Primary productivity, ocean disaster such as red tide, ocean pollution and ocean water quality monitoring is getting much important in conjunction of global environmental issues.This paper deals with water quality, in particular, Suspended Solid: SS related turbidity and ocean current.There are some proposed methods with remote sensing satellite data for SS estimation while not so many method is proposed for ocean current velocity estimation.The method proposed here is based on regressive analysis.Regressive analysis based SS estimation method is conventional method.Also ocean current velocity estimation method proposed here is based regressive analysis.
Recently, the Ariake Sea environment is getting worth.Red tide occurrence is getting much frequent together with scale of the red tide.There are some evidences which indicate increasing of the Ariake Sea transparency.In the meantime, ocean current velocity of the Ariake Sea is decreasing.Most of these evidences are based on the measured data with research vessels.Namely, these data are obtained along with research vessels' routes.On the other hands, remote sensing satellite data derived turbidity, ocean current velocity is essentially acquired with area based measurements.Therefore, it is possible to cover almost whole Ariake Sea area at once.This is the greatest advantage of the satellite based measurements.Furthermore, it is also possible to monitor these situations repeatedly for a long time period by using remote sensing satellite data.Thus seasonal changes and long term trend of turbidity or transparency and ocean current velocity of the Ariake Sea can be clarified.
The following section describes the method for estimation of turbidity and ocean current velocity followed by some experimental data.Then conclusion is described with some discussions.

A. Regression Based Method
Regression based approach is, generally, expressed as follows, (1) where Turb' denotes turbidity while L w (667) N denotes water leaving radiance at 667nm of observation center wavelength of the spectral band.Atmospheric contribution is dominant for the acquired water leaving radiance so that the atmospheric corrected L w (667) N is used for the experiments.There are MODIS Ocean Products including atmospheric corrected water leaving radiance which are provided by NASA/GSFC.On the other hands, α and β are regressive coefficients.These coefficients are obtained through regressive analysis which minimizing the following,

B. Remote Sensing Satellite Data Used
As aforementioned, MODIS onboard Terra and Aqua satellites is the most appropriate mission instrument.The MODIS swath width is wide enough for the Ariake Sea together with Instantaneous Field of View: IFOV.Also MODIS has 36 spectral channels which are suitable for estimation of turbidity and ocean current velocity.Table 1 shows the spectral channels of MODIS.IFOV are different by spectral channels, 250m for channels 1 and 2, 500m for channels 3 to 7 and 1000m for the rest of channels.Therefore, all the ocean products are defined as 1000m IFOV of products.On the other hands, the size of the Ariake Sea is 30km by 100km so that 1km IFOV of ocean products would be enough.

C. Trith Data Used
Match-up data between MODIS and truth data of turbidity which are obtained at the Saga University Observation Tower which is situated at the (33.1001147North, 130.273915 East) are collected for three months during from June to August 2010.Figure 1 shows the Google map of the Ariake Sea which is situated in Kyushu, Japan.Figure 1 also shows the location of Saga University Observation Tower.Turbidity is measured with electro-magnetic wave scattering radiance at the sea surface at the observation tower.Through a comparison with Holmagen standard liquid, the measured turbidity is calibrated.The unit of turbidity is FTU.
Table 2 shows the list of the data of the dates of turbidity measurements (Turb), the water leaving radiance in unit of mW/cm 2 /str/um at the four different channels which a\has the center wavelength, 412, 443, 531, and 667 nm.

A. Results from Regressive Analysis
The Pearson's correlation coefficient is defined as equation (5).Using the aforementioned match-up data, correlation coefficients are evaluated for the different spectral channels of MODIS.Table 3 shows the results.The correlation coefficient www.ijacsa.thesai.org of 667nm wavelength channel is greatest followed by 531, 443, and 412nm of wavelength channels as shown in Figure 2.
(5) Relations between turbidity and water leaving radiance derived from MODIS

TABLE III. CORRELATION COEFFICIENT FOR EACH WAVELENGTH BAND
Therefore, regressive analysis between turbidity and water leaving radiance at 667nm is conducted based on equation ( 1).The result is shown in equation ( 6).

B. Correlation Coefficient Between Turbidity and MODIS Based Water Leaving Radiance
Possible reason for this is shown in Figure 3.In accordance with increasing of SS, reflectance at 667nm is decreased.www.ijacsa.thesai.org The variance of the reflectance at 667nm is greatest followed by 531, 443, and 412nm of wavelength channels.

C. Ocean Current Velocity
In the same time as turbidity data is acquired at Saga University Observation Tower, other fundamental observation items are measured including ocean current velocity.In particular, it is known that ocean current velocity is proportional to turbidity.In order to confirm this fact, regressive analysis is made for ocean current velocity.Figure 4 shows the relation between turbidity and logarithmic function of ocean current velocity.

Relation between turbidity and logarithmic function of ocean current velocity
Positive correlation between turbidity and ocean current velocity then is confirmed.From the regressive analysis, the following equation is obtained with Root Mean Square Error: RMSE of 0.1266.
(7) Thus turbidity and ocean current velocity of the Ariake Sea area can be estimated with the atmospheric corrected MODIS data.

D. Examples of Estimated turbidity and Ocean Current Velocity
Using the atmospheric corrected MODIS data of the Ariake Sea area of 2004 and 2006 (which is shown in Table 4), spatiotemporal variations of turbidity and ocean current velocity are estimated.Figure 5 shows the estimated turbidity and ocean current velocity.

E. Trend Analysis of Turbidity and Ocean Current Velocity
Trend analysis of turbidity and ocean current velocity, then can be done with the atmospheric corrected MODIS data.Figure 6 shows the result from the trend analysis for 12 years www.ijacsa.thesai.orgstarting from 2000 to 2011.Meanwhile, annual mean of turbidity is shown in Figure 7 together with its linear approximated trend in 12 years.From the trend, it is found that the linear approximation function is expressed in equation ( 8) with R square value of 0.58.
Result from the trend analysis for 12 years starting from 2000 to 2011 Fig. 7.
Annual mean of turbidity and its linear approximation function

F. Spatial Variation of Turbidity
Bi-annual mean of turbidity of the Ariake Sea during from 2000 to 2010 is shown in Figure 8.In particular, turbidity of the Shiranui Sea areas and further southern portion of areas is getting decreases during from 2000 to 2004, and gradually increased during from 2004 to 2006.
After that, it is getting decreases again during 2006 to 2010 (much severely in 2010 and the after.In other words, transparency of such areas is getting large results in photosynthesis in the Ariake Sea areas is getting much active. In accordance with increasing of photosynthesis activity, phytoplankton is increases together with zoo plankton results in increasing of red tide occurrences.Also it is recognized that the turbidity of the areas just above the Isahaya Sea areas and northern portion of sea areas is getting decreases.Therefore, it can be said that one of the causes of the increasing red tide occurrence in the Shiranui Sea areas and the Ariake Sea areas must be decreasing of ocean current (decreasing of turbidity).IV.CONCLUSION Regression analysis based method for turbidity and ocean current velocity estimation with remote sensing satellite data is proposed.Through regressive analysis with MODIS data and measured data of turbidity and ocean current velocity, regressive equation which allows estimation of turbidity and ocean current velocity is obtained.With the regressive equation as well as long term MODIS data, turbidity and ocean current velocity trends in Ariake Sea area are clarified.It is also confirmed that the negative correlation between ocean current velocity and turbidity Also it is found that phytoplankton is increases together with zoo plankton results in increasing of red tide occurrences in accordance with increasing of photosynthesis activity.Also it is recognized that the turbidity of the areas just above the Isahaya Sea areas and northern portion of sea areas is getting decreases.Therefore, it can be said that one of the causes of the increasing red tide occurrence in the Shiranui Sea areas and the Ariake Sea areas must be decreasing of ocean current (decreasing of turbidity).

( 2 )
where a and b are regressive coefficients x and y are L w (667) N and turbidity, respectively.If the following equation can be assumed, (3) then the coefficients are determined as follows, (4) www.ijacsa.thesai.orgwhere denote means of L w (667) N and turbidity, respectively.

Fig. 1 .
Fig.1.Google map of the Ariake Sea and the location of Saga University Observation Tower TABLE II.MATCH-UP DATA BETWEEN MODIS AND TRUTH DATA OBTANIND AT THE SAGA UNIVERSITY TOWER FOR JUST THREE MONTHS FROM JUNE TO AUGUST 2010.
Fig.2.Relations between turbidity and water leaving radiance derived from MODIS

Fig. 3 .
Fig.3.Illustrative view of spectral reflectance for the different suspended solid concentrations.

Fig. 4 .
Fig.4.Relation between turbidity and logarithmic function of ocean current velocity

Fig. 5 .
Fig.5.Example of the estimated turbidity and ocean current velocity of the Ariake Sea area in 2004 and 2006

Fig. 8 .
Fig.8.Bi-annual mean of turbidity of the Ariake Sea during from 2000 to 2010

TABLE I .
SPECTRAL CHANNELS OF MODIS

TABLE IV .
MODIS DATA USED FOR ESTIMATION OF TURBIDITY AND OCEAN CURRENT VELOCITY