Effective visualization for the spatiotemporal trend analysis of the water quality in the Nakdong River of Korea

Abstract

Spatial and temporal trend analyses were performed to obtain more meaningful water quality information in table and three-dimensional graph forms. Using the statistical approaches of the Seasonal Mann–Kendall (SMK) and LOcally WEighted Scatter plot Smoother (LOWESS) methods, the trends of three water quality parameters, including Biochemical Oxygen Demand (BOD), Total Nitrogen (TN), and Total Phosphorus (TP) measured along the Nakdong River of Korea between 1992 and 2002 were analyzed. The trends of the slopes were calculated using the SMK method for two consecutive stations and years. These values are provided in the trend tables which indicate the extreme upward and downward trends. Also, three-dimensional graphs of the water quality in the Nakdong River were generated with respect to the distance from upstream of the river and time of month. From this study, it was concluded that these tables and three-dimensional maps could be used as a useful tool to provide the spatiotemporal trend information such as the hot spots/moments of improvement and deterioration in the water quality of the Nakdong River, with the present web-based information system.

Introduction

Water quality management policy is often associated with judgments not only by present situations, but also by past records. From the monitoring data of water quality, it can be determined whether the previous policy was good enough or further regulations will be needed. These types of judgments are usually obtained from the statistical trend analysis of water quality data.

Statistical trend analysis makes water quality data to be more comprehensible such that it can provide the scientific guideline to policy decision makers (Helsel and Hirsch, 1992, Paul and Linfield, 1997). Several statistical methods, such as the Seasonal Mann–Kendall test (SMK) and LOcally WEighted Scatter plot Smoother (LOWESS) methods, have been developed and widely used in water quality management (Cleveland and Devlin, 1988, Lettenmaier et al., 1991, Walker, 1991, Zipper et al., 2002, Passell et al., 2004, Boeder and Chang, 2008, Chang, 2008). Recently, the trend analysis and other statistics are getting more attention due to two emerging issues in water quality management. First, the information techniques for water quality data, such as world wide web (www), Geographic Information System (GIS), database, and highly sophisticated computer models and graphics, are now available in most industrialized countries, for a representative example; http://waterdata.usgs.gov/nwis (Norman et al., 2000, Liang and Frank, 2001, Chang, 2008). Second, an active participation of local watershed residents, as a part of an integrated watershed management, becomes more important for water quality improvement (Cobourn, 1999, Walesh, 1999). For effective water quality management, therefore, the statistically summarized data in numeric and visual formats is needed to provide the quantitative and qualitative information of water quality condition for local residents as well as policy decision makers through the web-based information system.

The water quality of major river systems in Korea has been monitored since late 1970 and the water quality information system based on the web (http://water.nier.go.kr) was developed in 1998 by the National Institute of Environmental Research (NIER) under the supervision of the Ministry of Environment of Korea. The system is web-based and includes GIS, database of pollution source, water quality models, and computer graphics. Due to the lack of statistical tools, the water quality data are currently provided with the raw data, where only limited information can be obtained from the system. Some statistical values and simple graphs, including means, standard deviations, and bar and line graphs, can be obtained along with measured raw data in the present system. Pollution loads from point and non-point sources, sewer systems, land use types, geography of watershed, and some other information for water quality management, are also available. The system continues to be updated with recent monitored water quality data and information periodically.

In previous researches, various visualization techniques with statistical methods were attempted for trend analysis of water quality (Garbrecht and Fernandez, 1994, Zhang, 1998, Miller et al., 2001, Yang et al., 2002). Boyer et al. (2000) proposed various visualization approaches, such as a 1 dimensional box and whisker plots at a single station, 1 dimensional time series line graphs at a single or group station, 2 dimensional contour maps (snapshot in time), 2 dimensional time series animation of contours, and 3 dimensional iso-surface slicing animation. Each technique of visualization showed its own characteristics in demonstrating a trend of water quality (Cleveland and Devlin, 1988, Lettenmaier et al., 1991, Walker, 1991, Helsel and Hirsch, 1992, Paul and Linfield, 1997, Zipper et al., 2002, Bekele and McFarland, 2004, Passell et al., 2004, Boeder and Chang, 2008, Chang, 2008). The snapshot indicates the spatial or temporal variation of specific station or period separately. Although the computer-aid animation could display temporal and spatial changes of water quality, its applicability is limited within the printed snapshot. The limitation of the snapshot could be overcome by the three-dimensional graph, which supplies an overall trend temporally and spatially.

In this study, the statistical and visual tools were proposed in order to upgrade the present water quality information system, which can be more useful and fit well to the recent water quality issues. This paper focused on the spatial and temporal trend analyses of water quality in a large river system, where the results were presented in visual graphs and numeric tables. These quantitative and qualitative analyses are useful tools to analyze and display the long-term trend of water quality.