Assessment of Long-Term Physicochemical Water Quality Variations by PCA Technique in Lake Hwajinpo , South Korea

Physicochemical properties of water were analyzed to assess the long-term water quality variations in Lake Hwajinpo, Korea. Water quality data monitored from 1998 to 2015 was divided in three periods and descriptive statistics, correlation and rotated components were deducted using statistical procedures. Based on the results of analyses, water quality patterns in three periods (1998-2003, 2004-2009 and 2013-2015) were distinguishable from each other. Water parameters, Chlorophyll-a, total phosphorus, total nitrogen, dissolved oxygen and pH, showed the highest mean values in the first period. On the other hand, conductivity and salinity in the second period and temperature and suspended solids in the third period showed the highest mean values. Principal component analysis (PCA) procedure was utilized to deduct the most significant parameters influencing water quality and observed that each period had different pattern of variables. Salinity and conductivity were the two variables highly contributing in first component/factor (F1) explaining 20.77% and 22.93% of the total variance in the first period and second period, respectively. But total phosphorus and chlorophyll-a were the two variables highly loading in F1 of the third period explaining 23.72% of total variance. These results revealed that the water quality of Lake Hwajinpo had different patterns of variations throughout the study period. Thus, PCA results could be valuable to understand the water quality status of water body and take proper steps to protect the water environment.


Introduction
Coastal lagoons are shallow water bodies separated from sea by barrier islands, coral reefs and sand bars.They have unique and rich habitats from sea to inland since they occupy the area between the sea and land.Along with lagoons, there may be habitats of barrier islands, spits, beaches, sand dunes, salt marshes and seagrasses to support biodiversity.The diverse ecosystem within the surrounding of lagoons plays important role to have high species of diversity and creates ecotones-area of transition between major habitats: marine, freshwater and terrestrial.The ecosystems of the lagoons were disturbed due to changes in climate and regional landscape [1] [2] [3].Anthropogenic activities such as agriculture, aquaculture, urbanization, industrialization and recreational activities are major causes of water quality deterioration of lagoons.Eutrophication is the major threat to ecology and biological communities which arises because of high nutrient from point and non-point sources of pollution.Discharges from non-point source pollution such as agricultural and domestic water can cause the episodic eutrophication and algal blooms which can be toxic [4] [5] [6].Lagoons receive sediments from rivers that flow into them and surface runoff.Therefore, coastal lagoons have a tendency to become filled with the sediments and thus disappear, in time, from the coastal landscape.Overuse of land and rapid industrialization are the main reason to the loss of lagoons area [7] [8].The major threat to the existence and health of lagoons water body is human activities.
The lagoons in the Korean Peninsula are mainly distributed in the eastern coastline.Among the 57 natural lakes in this region 48 of them are marine origin natural lakes which were formed by sand dunes due to the ocean water currents [9] [10].Lake Hwajinpo is one of the seven major lagoons in the eastern coast of Korea and the others are Youngrang, Kyoungpo, Sonji, Mae, Cheongcho and Gwangpo.The slope of the surface of the lake is gentle and the water level is stable throughout the year, and the wetland vegetation is well developed.Unlike the freshwater lake, Hwajinpo Lake is characterized by the combination of seawater and fresh water.It has the characteristics of brackish water.This lake has one inlet channel to exchange water with East Sea and can be categorized as the Choked lagoon [11].Lake Hwajinpo was well protected lagoon in this region till 1991 because the public access was limited since it was designated as a military area.But later it was opened as a national recreation site and several infrastructures: museums, lodging and amusement facilities were built to attract visitors.These facilities along with increasing agricultural activities in the watershed causes high nutrient in the lagoon.The surface area of the lagoon has been slightly changed from 2.29 km 2 in 1918 to 2.06 km 2 in 2000 which is 90% of the initial lagoon area [7].
Water quality of coastal lagoons and other water bodies are very important for the ecosystem as well as human health.It is necessary to monitor and manage lagoons to abate any negative impact on its surrounding.A monitoring program is required to provide reliable measure of the quality of water bodies.Through data analysis, one can interpret and understand the nature of pollution [12].The impact of natural and anthropogenic processes on the lagoon ecosystem could be identified by continuous monitoring and analyzing.Temporal and spatial water quality properties are helpful to recognize the changes and trend of lagoon water [13] [14].Lake Hwajinpo and other major eastern coastal lagoons in Korea are regularly monitored since 1997 and ecosystem restoration efforts are made [15].Previous literature of Lake Hwajinpo demonstrated eutrophication [10], limnological properties [16], landscape change [7] [9], water environment variation [17] and phytoplankton and trophic status [18].
The aim of this study is to assess the long-term temporal variations of physicochemical water quality of Lake Hwajinpo from 1998 to 2015.Descriptive water statistics and correlation were deducted and factor analysis (FA) was conducted to identify the important components or factor of water quality variables.The deduction of major parameters and grouping variables by multivariate techniques such as principal component analysis (PCA) enables to identify the sources of pollution [19] [20] [21].Thus the result of this study will be helpful to water managers to understand the water quality status of Lake Hwajinpo and apply for further management.

Study Area
Lake Hwajinpo, a lagoon in the eastern coast of Korea, is located in Geojin township of Goseong County, Gangwon Province (Figure 1).The surface area of the lagoon is 2.06 km 2 with average depth 1.

Sampling and Measurements
Water quality was monitored from 1998 to 2009 and 2013-2015 in a monthly basis in 65 episodes from four sites, two in North and two in South, except the months of severe winter.Water temperature (Temp), electric conductivity (EC), salinity (Sal), dissolved oxygen (DO), Secchi disk transparency (SD), pH, suspended solids (SS), total phosphorus (TP), dissolved inorganic phosphorus (DIP), total nitrogen (TN), nitrate nitrogen (NO 3 -N), ammonia nitrogen (NH 3 -N) and Chlorophyll-a (Chl-a) were the physicochemical parameters chosen.Lorenzen's (1967) method was used to calculate chlorophyll-a concentration.Filtered water was used to calculate dissolved nitrogen and phosphorus.TP was determined according to standard methods APHA [22], employing persulfate digestion and ascorbic acid method.TN was determined by the cadmium reduction method after persulfate digestion, using a flow injection auto-analyzer (BRAN-LUEBBE, Auto Analyzer3).Temperature, salinity, EC, pH and DO were measured in-situ with a multi-probe meter (YSI, 6000).

Statistical Analyses
The data was divided in three periods as 1998-2003, 2004-2009 and 2013-2015 to understand the water quality variations in different periods.Box and whiskers plots for five parameters were presented in (Figure 3).Several outlier values were visible in each period.Therefore we chose to perform non-parametric tests.The non-parametric correlation and PCA/FA test were performed for statistical analysis in SPSS 19 software (IBM Corporation, Armonk, NY, USA).PCA helps to reduce dataset and it deducts dominant components which are valuable to have a better interpretation of variables.The details of these tests methods were detailed in these literatures [23] [24] [25].

Relations between Water Quality Parameters
Spearman's rank correlation coefficients matrices of water variables for three periods are presented in Tables 4-6.Overall, correlation coefficients values (p <

Identification of Significantly Important Variables Affecting Water Quality
PCA procedure was applied on each datasets to deduct important variables for each factor.The factors with Eigen value more than 1 (scree plots, Figure 4,

Discussion
Lake Hwajinpo water quality demonstrated in this study showed wide variations in three periods in 1998-2015.The mean temperature decreased in the second period and increased in the third period (2013-205) with the highest (in bold) temperature 20.6˚C.Lagoon surface water temperature is an important factor to  run physicochemical and biological processes in water [27].Similarly, suspended solids mean value was significantly higher in this period than the previous period with the highest value 27 mg/l.The continuous increase in SS indicates that the construction and agricultural activities in the basin are increasing.The monsoon stormwater runoff causes soil erosion from non-point sources can increase the SS concentration in water body [10] [28].The decrease in Chl-a was observed in comparing to the first period (1998)(1999)(2000)(2001)(2002)(2003) in the third period but wa higher than the second period (2004)(2005)(2006)(2007)(2008)(2009).Water transparency was slightly improved throughout the monitoring periods.Salinity of lagoon plays an important role which varies with inflow of freshwater from streams and runoff, and circulation to the sea water.The salinity in Lake Hwajipo was the lowest in the third period.This could be related to the low water circulation between North and South due to the narrow connecting channel [16].The nutrient (N and P) components were slightly lower in the third period of the study as well as SD.
The nutrients in the water bodies are the byproduct of the human activities in the basin which have a large localized effect on water chemistry.Land use plays significant role on the hydrology system of the water body and its water quality The sources of TP and TN can be interpreted as point sources (domestic, industrial wastewater etc.) and non-point sources (forests agricultural land etc.) [21].Sedimentation of nutrients and their release due to suspension and re-suspension also plays important role [29] [30].Internal and external loading of nutrients are the big factors to be considered to achieve the sustainable water environment of the lagoon.Reclamation of water body through dredging and enhancing water circulation could be the long term goals [31] [32] [33].
In this study, the water quality of Lake Hwajinpo in three different time periods was demonstrated by analyzing physicochemical parameters statistically.
And the results showed that the lagoon water quality was distinctly different in each period.The highly significant parameters in the first period (1998)(1999)(2000)(2001)(2002)(2003) and second period (2004-20090 were less significant in the third period (2013)(2014)(2015).Total phosphorus and chlorophyll-a were the two water quality parameters highly significant in the third period.

Conclusions
Long-term water quality data was analyzed to understand how the physico-

Figure 1 .
Figure 1.Map of the Lake Hwajinpo showing location in Korea and monitoring stations.The Geographic co-ordinates are Latitude 38˚45-48'N and Longitude 128˚42-44'.

Figure 2 .
Figure 2. Annual rainfall in Goseong County, the basin of Lake Hwajinpo.
for the period of 1998-2003, 2004-2009 and 2013-2015, respectively.The results showed that the highest mean values of temperature, SD transparency and SS were in period of 2013-2015 and conductivity, salinity and NO 3 -N were in the period of 2004-2009.However, the highest mean values of DO, pH, Chl-a and nutrient components (TP, DIP and TN) were in the period of 1998-2003.There was not much variation of mean values of NH 3 -N.The highest mean value of temperature was 20.6˚C in the latter part of study period and the minimum was 17.51˚C in 2004-2209.The highest mean values of SD and SS were 0.80 m and 27 mg/L, respectively.Salinity, conductivity and NO 3 -N had the highest values as 25 psu, 2625 µs/cm and 0.08 mg/L in 2004-2009, respectively.In the period of 1998-2003, the highest observed mean values of DO and pH were 10.56 mg/L and 9.3, respectively.Meanwhile, these two parameters had the lowest mean values in 2013-2015.The nutrient components had the maximum mean values in the earlier period of the study.And the lowest mean values were

0
.05, two-tailed, in bold) were not very high; however, salinity and conductivity demonstrated highly significant positive correlation for all three periods.DO showed highly significant negative correlation with temperature with the highest coefficient 0.72 in 2013-2015.DO showed significant positive correlation with pH with coefficient values nearly 0.50 in the first two periods.SS correlation with TP and TN were positive and Chl-a and SS were negative.These correlation coefficients were in the range of 0.30 -0.50.TP was positively correlated with TN with all three periods with coefficients approximately 0.50.TP showed significant correlations with SS in the first two periods but no correlation in 2013-2015.Chl-a demonstrated significant positive correlation with TP and the highest coefficient values was 0.649 in 2013-2015.Similarly, Chl-a showed positive correlation with TN and the values were approximately 0.32.The positive correlations of SS with nutrient components and Chl-a might be related to the discharges of pollutants in lagoon from natural and anthropogenic sources[23] [26].

Figure 5 andFigure 4 .
Figure 5 and Figure 6) were retracted PCA produced six components/factors (Figure 4) of the first period (1998-2003) which explained 74.07% of total variance.The first three factors explained nearly 50% of the total variance (20.77%, 16.96% and 11.01% for F1, F2 and F3, respectively).F1 showed the highest positive correlation with Cond and salinity as 0.983 and 0.981, respectively.F2 was

[ 2 ]
. DO in water depends on water circulation and photosynthesis activities by phytoplankton in lagoon.Principal component analysis in this study helped to identify which components/factors were influencing the water quality in different periods.In the first two periods, 1998-2003 and 2004-2005, F1 was mainly contributed by Cond and salinity.These two parameters are related to the physical source variabily.TN and Chl-a in first period and TP, Temp and Chl-a in the second period were the F2 contributing variables.But in the third period TP and Chl-a showed the highest correlation with first factor and contributed 23.72% of the total variance.

Table 1 .
Descriptive statistics of surface water quality of Lake Hwajinpo from 1998-2003.

Table 2 .
Descriptive statistics of surface water quality of Lake Hwajinpo from 2004-2009.

Table 3 .
Descriptive statistics of surface water quality of Lake Hwajinpo from 2013-2015.

Table 4 .
Spearman's rank correlation and p-value of water quality variables in Hwajinpo from1998-2003.

Table 5 .
Spearman's rank correlation and p-value of water quality variables in Hwajinpo from 2004-2009.

Table 6 .
Spearman's rank correlation and p-value of water quality variables in Hwajinpo from 2013-2015.
214 Journal of Environmental Protection
Extraction method: Principal component analysis.Rotation method: Varimax with Kaiser Normalization.a: Rotation converged in 5 iterations.