Ähnlich wie Study on Discharge Characteristics of Pollutant Load at Gyoungahn River with GIS and Water Quality Modelling for Total Maximum Loads Management
Ähnlich wie Study on Discharge Characteristics of Pollutant Load at Gyoungahn River with GIS and Water Quality Modelling for Total Maximum Loads Management (20)
Slide deck for the IPCC Briefing to Latvian Parliamentarians
Study on Discharge Characteristics of Pollutant Load at Gyoungahn River with GIS and Water Quality Modelling for Total Maximum Loads Management
1. 오염총량관리에 따른
GIS와 수질모델링에 의한
경안천 유역의
오염부하량 배출특성 연구
Study on Discharge Characteristics of Pollutant Load
at Gyoungahn River with GIS and Water Quality Modelling
for Total Maximum Loads Management
서울시립대학교
물 환 경 연 구 실 Lee, Kwan-Woo
3. 1. Introduction
Background1
Industrialization, Urbanization
increase pollution loading on water supply source
As Management of Point Pollution Source enhanced
Contribution rate of Non-point Pollution Source increased
Impose of Total Maximum Loads Management (‘99)
Ratio of Non-point Pollution Source in Discharge : about 42% (‘05)
3
5. 1. Introduction
Background1
5
strengthen Effluent Standard, more Installation of Sewerage
Point Source Loading decreased
Ratio of Discharge Loading by Non-point Pollution Source
increased
current Pollution Management reaches the limit
related Laws revised ;
Government demands Prediction and Countermeasure to
Developer
6. 1. Introduction
Purpose1
6
Estimating Pollutant Load based on Land Use Plan, Precipitation
and Pollution Rate
Hard to predict on Runoff Loading of actual Sub-watersheds
by Non-point Pollution Source
Non-point Pollution Source run off at rainfall
Runoff flow vary on each Season,
therefore hard to predict, quantify
for more efficient Total Maximum Loads Management
require exact Pollution State
and Geographic Information System using Digital Elevation Model,
to analyze corresponding Water Basins
7. 7
Purpose1
Input DATA
- GIS
DEM, Soils, Land Use
- Climate
Precipitation etc.
- Point Source
Sewage Treatment Plant
Wastewater Treatment Plant
SWAT Model
Output for each stream
- Flow
- Constituent Yields
Additional DATA
- Hydraulic, Hydrologic Coeff.
of Sub-watershed
- Slope Length
1. Introduction
8. 1. Introduction
Purpose1
8
Research on Land Use Plan, Soil, Climate, Precipitation
Input SWAT Model =
estimating Pollution Loads by Non-point Pollution Source
Research on Characteristic Data about Streams / River
Input SWAT Model =
Slope Length Patch, Hydraulic / Hydrologic Coefficient
Analyzing Discharge Characteristics of Pollutant Load
Apply to Policy, considering Priority
9. 1. Introduction
Water Quality Modelling2
9
Using Water Quality Modelling for Total Maximum Loads
Management
- to decrease Conflict between local Governments
- to seek balanced Development between local Governments
Therefore,
Modelling required reasonable, fair, scientific Process
10. 1. Introduction10
1) Procedure
- Setting up Purpose
- Understanding current State
- Setting up Scope
- Making Scenario
- Analyzing Prediction Results
2) Problem and Limit
- lack of fundamental Data or unsuitable
- not enough considering local Characteristics
- Uncertainty of Prediction
Water Quality Modelling2
11. 1. Introduction
SWAT3
11
SWAT(Soil and Water Assessment Tool) is
River basin, or Watershed Scale Model
developed by USDA Agricultural Research Service)
to predict the impact of land management practices on water,
sediment and agricultural chemical yields
SWAT’s benefits are
Watersheds with no monitoring data can be modeled,
The relative impact of alternative input data on water quality
or other variables of interest can be quantified
SWAT is continuous time Model
13. Preparation
1. Introduction
SWAT3
13
Schematic of SWAT Model
Input
- GIS
DEM, Soils, Land Use
- Climate
Temp., Relative Humidity, Precipitation etc.
- Point Source
Sewage Treatment Plant
Wastewater Treatment Plant
Effluent Water Quality Data
SWAT Model Analyze Output
Calibration
& Validation
21. 2. Materials and Methods
Study Watershed1
21
Location of Gyoungahn River
22. Watershed Overview
2. Materials and Methods
Study Watershed1
22
Basin Length (km) Area (㎢) Sub-Watershed Area (㎢)
Gyoungahn
River
22.50 / 49.30 575.32
Gyoungahn A 198.4
Gyoungahn B 248.9
Location of Monitoring Site
23. 2. Materials and Methods
Study Watershed1
23
Aspect Analysis Altitude Analysis
24. 2. Materials and Methods
Study Watershed1
24
Slope Analysis Soil Analysis
28. 2. Materials and Methods28
Streams within Basin Land Use and etc.
SWAT Input2
29. 29
ArcView GIS Patch3
SWAT model calculates Average Slope using DEM,
simulates with Average Field Slope Length
existing SWAT model is developed that
use Field Slope Length as 0.05m
in the topography of average slope ≥ 25%
existing SWAT model is suitable for U.S. topography, in generally
gradual Slope
these condition is hard to apply to Korean topography, sharp
Slope
2. Materials and Methods
30. 30
Applying SWAT ArcView GIS Patch II
correct Field Slope Length to 10m
ArcView GIS Patch3
2. Materials and Methods
35. 35
Main Channel4
Ch_side_slope
Fd_side_slope
existing SWAT model is suitable for the river of wide channel and
gradual side slope
Korean River : relatively narrow channel and sharp side slope
Therefore, classify by Sub-watershed, modify manually and simulate
2. Materials and Methods
Fd_width
Ch_width
Ch_depth
1
1
47. 47
0
100
200
300
400
5000
4
8
12
16
20
2009-01-15 2009-08-03 2010-02-19 2010-09-07 2011-03-26 2011-10-12
Precipitation(mm)
Observed&SimulatedBOD(mg/L)
Comparison of Observed and Simulated BOD
Precipitation
Observed BOD
Simulated BOD
SWAT simulate ; Patch II O,
Main Channel X
3. Results
Output2
48. 48
Num. of Data : 47
Nash-Sutcliffe Efficiency (NSE) : 0.8510
Coefficient of Determination (R2) : 0.8793
Absolute Percent Bias (APB, %) : 47.4026
Sum of Square Error (SSE) : 27062.4067
Root Mean Square Error (RMSE) : 23.9957
Mean Absolute Error (MAE) : 11.6771
Index of Aggrement (d) : 0.9514
Num. of Data : 47
Nash-Sutcliffe Efficiency (NSE) : -2.4622
Coefficient of Determination (R2) : 0.0167
Absolute Percent Bias (APB, %) : 207.1894
Sum of Square Error (SSE) : 157834.4200
Root Mean Square Error (RMSE) : 57.9498
Mean Absolute Error (MAE) : 35.6983
Index of Aggrement (d) : 0.2930
SWAT simulate ; Patch II O,
Main Channel X
3. Results
Output2
Validation of Output of SWAT by NSE (Flow, SS)
NSE : Nash-Sutcliffe Efficiency
53. 53
0
100
200
300
400
5000
5
10
15
20
25
2009-01-15 2009-08-03 2010-02-19 2010-09-07 2011-03-26 2011-10-12
Precipitation(mm)
Observed&SimulatedBOD(mg/L)
Comparison of Observed and Simulated BOD
Precipitation
Observed BOD
Simulated BOD
SWAT simulate ; Patch II O,
Main Channel O
3. Results
Output3
54. 54
Num. of Data : 47
Nash-Sutcliffe Efficiency (NSE) : 0.8012
Coefficient of Determination (R2) : 0.8486
Absolute Percent Bias (APB, %) : 51.6699
Sum of Square Error (SSE) : 36109.8475
Root Mean Square Error (RMSE) : 27.7181
Mean Absolute Error (MAE) : 12.7284
Index of Aggrement (d) : 0.9299
Num. of Data : 47
Nash-Sutcliffe Efficiency (NSE) : -3.5058
Coefficient of Determination (R2) : 0.0070
Absolute Percent Bias (APB, %) : 232.1376
Sum of Square Error (SSE) : 205414.7885
Root Mean Square Error (RMSE) : 66.1100
Mean Absolute Error (MAE) : 39.9968
Index of Aggrement (d) : 0.2304
SWAT simulate ; Patch II O,
Main Channel O
Output3
Validation of Output of SWAT by NSE (Flow, SS)
NSE : Nash-Sutcliffe Efficiency
3. Results
55. 55
Calibration and Validation4
① Mean Error (ME)
② Mean Absolute Deviation (MAD)
③ Mean Absolute Error (MAE)
④ Root Mean Square Error (RMSE)
⑤ Nash-Sutcliffe Efficiency
Poor Fair Good Very Good
NSE for
daily Simulation
< 0.60
0.60 ~
0.70
0.70 ~
0.80
0.80 <
Criteria for evaluating model performance (Donigian and Love, 2003)
3. Results