GISdevelopment.net ---> AARS ---> ACRS 1989 ---> Water Resources

Estimation of water quality using GIS data and Landsat TM data

M. Shizukuishi, O. Imai, and H. Takeuchi
Systems Engineering Center, Pasco corporation
No. 13-5, 2-chome Higashiyama,
Meguro-ku, Tokyo Japan


Abstract
Remote Sensing technology offers a wider latitude of potential application when combined with excels in storage, manipulation and analysis of geographic information and socio-economic data. In the study, it was attempted to divide the study area (waters) I sections by means of GIS in dealing with Landsat TM data in a bid to estimate water quality at actual water quality could be emphasized for analysis by dividing the study area instead of dealing in one whole area to make more realistic estimation imagery possible to be developed, thus confirming the usefulness of GIS data.

Introduction
Combination of GIS (Geographic Information System) and remote sensing technologies is a natural development in response to the technological needs of recent years. For GIS, remote sensing data including Ladnsat data are one of the most important sources of information. They provide the latest information in a form readily adaptable to computers and available for use in time series. So that the integration of remote sensing data in GIS is most meaningful for creation of a smaller map scale GIS while eliminating problems involved in the usually cumbersome process of data input in computers.

On the other hand, GIS has an excellent technology to offer in manipulation and analysis of volume data that can be applied most effectively to remote sensing data. GIS is equipped with data in other fields that complement remote sensing data, such as geographic information based on smaller administrative units like cities, towns and villages, and socio-economic data derived form the census. Theses two technologies when combined effectively provide wider attitude of potential applications than when applied individually.

The integration of these two technologies, however, was not free from problems in the past despite the awareness of its greater potential. There were problems due tot eh ever expanding size of computer systems for GIS and remote sensing data processing and, more importantly, the difference in data storage methods between GIS and remote sensing data. Today as a result of rapid progress in hardware and software, the two technologies are feasible on the one same hardware system.

One such example is a combination of ARC/INFO (a proprietary system of ESI, USA), a leading vector type of GIS software, and Erdas (similarly of ERDAS, USA) that are in operation as PASCO where the author belongs. In this study, applicability or usefulness of GIS data was examined for estimation of water quality as a leading indicator of the aquatic environment.

Data and study area
Tokyo Bay as the study area serves as an entrances to the sea-borne traffic bound for Cosmopolitan Tokyo with direct bearings on the daily life of citizens. The bay itself is enclosed (surrounded by metropolitan Tokyo, Kanagawa and Chiba Prefectures), deep inside, and very narrow at its mouth restricting a free influx of outside sea water while allowing large volumes of pollutants into the bay via rivers, which in turn add to biological productivity to cause what is commonly known as "red water" a phenomenon caused by masses of dead planktons and observed every year.

The water quality of the bay, therefore, is characteristically subject to tidal currents and river waters and reflect the different aquatic environments along the coast sea and the inside sea. The characteristic was further examined in the present study by using Landsat TM data and the aquatic environment database built on ARC/INFO.
  1. TM data
    The data were selected for study in such a manner that they were of the same dates as those of water quality data and from periods when atmospheric conditions ere relatively stable. The selected data were as follows.

    Path-row : 107-035 and 107-036 Dates : 1) August 6, 1986 2) March 2, 1987 Correction : Status corrected

  2. Water quality data
    The water quality of Tokyo bay has been monitored regularly at the 51 survey stations located in the surrounding Tokyo, Kanagwa and Chiba for the past five years. Based on these data as input, chronological and spatial analyses were made with respect to the study area.

Figure 1. Location map of survey stations 1 ~ 51 : Survey stations


Methodology
The study flow is schematically shown in Figure 2
  1. Preprocessing
    Since the study area was covered by two scenes of TM data as purchased, mosaicking was performed in the lien direction, followed by geometric correction to make TM data properly correspond to water quality data on the 1/50,000 topographic map.

  2. Data input
    Water quality data were input in a computer in a uniform format for chronological and spatial analyses that followed. The bathymetric map as digitized in polygons for input.

  3. Inner correlation
    Inner correlation were computer to examine the distinctions between the bands of TM data and between the water quality survey items. Based on the computation results, the bands with lower values of correlation were applied for the analysis.

  4. Analysis of water quality

    1. Chronological analysis :
      As Seasonal changes in the aquatic environment by developing graphs for such changes.

    2. Spatial analysis :
      To identify the spatial characteristics of the study area, cluster classification was applied to the typical water quality data for summertime in an attempt to classify water masses. A similar attempt was made for the August 6, 1986 data.

  5. Data overlay
    To Consider a topographic impact on the aquatic environment, TM data were overlaid on the bathymetric map in the computer by means of the Live Link to connect ARC/INFO and Erdas.

  6. Water quality Estimation using Landsat Data
    In the water quality analysis, multiple regression analysis was applied to determine correlations between TM data and water quality data. Analysis was made in the following terms considering the characteristics of remote sensing data.

    1. Chlorophyll - a
    2. SS (Suspended sediment)
    3. Turbidity
    4. Water temperature

Figure 2. Study Flow


Study results
  1. Chronological analysis :
    To find about the chronological changes in sea water in terms of physical properties, nutritional base density, and turbidity due to organic material the coastal sea and the inside sea were studied respectively for turbidity, salinity, T-N, PO4-P, and COD, whose chronological changes in average values were as shown in figure 3. From the figure, it was found as follows:

    1. Turbidity, Salinity, and COD change in regular yearly patterns at both locations, coastal sea and inside sea.

    2. With respect to the inside sea data, the PO4 - P COD data of Jun 1985 were conspicuously greater than rest of data CHA values for the locations were also found to be substantially high. They were considered to reflect the massive red water that prevailed at the particular time.

    3. PO4 - P in the coastal sea shows a tendency to decrease and Cod to increase.


    Figure 3. Chronological changes in Major water quality by item
    (April 1980 - March 1988)


  2. Correlations between Water Quality Items
    A matrix was developed to determine correlations between water quality items with respect to the coast and the inside sea using all data from all relevant stations. The matrix is shown in table 1.

  3. Spatial Analysis
    In order to define the spatial characteristics of the study area, water quality at each station was summarily represented in a graph using the box plot chart showing ranges of fluctuation of water quality for each station, as shown in figure 4. From the figure, the following characteristics were derived.

    1. Generally, the differences between the station and the ranges of fluction at each station were both larger in the coastal sea.

    2. With respect to nutrition base in the coastal sea density of nitrogen type base at st. 1 and 2, phosphatic at st. 12, and both densities at st. 7 and 9, were by far higher than those at other station. The high densities are possible attributable to the waste water discharged from water processing plant at st. 1, 2 and 12.

    3. In the inside sea, salinity fluctuate widely at st. 15 due to its location at river mouth.

    4. NH4 - N is gradually oxidized in the environment to turn into NO2 - N and NO3 - N. Therefore, NH-N/DIN is lower in the inside sea than in the coastal sea.

Figure 4. Fluctuations in Water quality at respective stations
(April 1980 - March 1988)

Table 1: Water quality item correlation matrix
COASTAL SEA
INSIDE SEA WT TRS DO COD T-N T-P SAL NH4 NO2 NO3 PO4 TOC DOC CHL
WT -0.425 -0.045 0.343 -0.016 0.234 -0.518 -0.074 0.391 0.138 0.190 0.335 0.242 "
TRS -0.461 0.288 -0.614 -0.269 -0.348 0.530 -0.232 -0.168 -0.102 -0.277 -0.573 -0.461 "
DO 0.027 -0.194 0.488 0.034 -0.072 0.122 0.018 -0.057 -0.083 -0.206 0.466 0.194 "
COD -0.549 -0.618 -0.436 0.577 0.425 -0.490 0.524 -0.292 -0.004 0.306 0.859 0.713 "
T-N 0.173 -0.391 -0.107 0.497 0.453 -0.526 0.943 0.301 -0.059 0.472 0.539 0.730 "
T-P 0.370 -0.409 -0.179 0.563 0.787 -0.517 0.386 0.170 -0.261 0.960 0.424 0.451 "
SAL -0.577 0.510 -0.221 -0.426 -0.525 -0.484 -0.434 -0.418 -0.397 -0.535 -0.428 -0.506 "
NH4 -0.124 -0.157 -0.245 0.178 0.796 0.633 -0.250 0.225 -0.147 0.413 0.500 0.705 "
NO2 0.216 -0.157 -0.013 0.217 0.465 0.322 -0.260 0.139 -0.125 0.164 0.235 0.281 "
NO3 0.099 -0.171 -0.282 0.027 0.382 0.185 -0.604 0.157 0.094 0.296 -0.045 -0.020 "
PO4 0.167 -0.168 -0.497 0.115 0.671 0.833 -0.393 0.680 0.308 0.246 0.317 0.416 "
TOC 0.544 -0.581 0.442 0.442 0.457 0.521 -0.373 0.161 0.167 -0.012 0.080 0.854 "
DOC 0.467 -0.540 0.213 0.704 0.553 0.510 -0.427 0.343 0.188 -0.101 0.255 0.839 "
CHL 0.470 -0.447 0.833 0.833 0.260 0.436 -0.236 -0.001 0.177 -0.091 -0.043 0.827 0.539

Discussion
From the analysis results, correlations between water quality and TM data were found to be greater in the following items.
  1. Amount of Chlorophyll-a in areas susceptible to river water / TM data.

  2. Amount of S in sea water / TM data

  3. Temperature / Band 6

  4. Amount of chlorophyll - a in inside sea/TM data chlorophyll -a estimation images as of August 8, 1986 and March 2, 1987 are shown as examples.
Summary
In the present study, water quality estimation images were developed according to observation time and water depths to enhance the correlation coefficients for relations between water quality and TM data. From the results, the following can be said.
  1. It is clearly important to consider the impact of tidal current when Landsat TM data are compared to water quality.

  2. Division of the coastal area and the inside sea definitely helped to define the relations between TM data and water quality.
Reference
  • The Institute of Statistical Mathematics. Studies of sample Surveys for Environmental Measurement, 1989.
  • Dangermond, J. The software toolbox approach to meeting the user's needs workshop 1986.