1. Impacts of Climate and Land-Cover Changes on Water Resources: Methodology Review
May 6, 2015
Huidae Cho
Water Resources Engineer, Dewberry Consultants
Part-Time Assistant Professor, Kennesaw State University
Joonghyeok Heo, Jaehyung Yu, John R. Giardino, Huidae Cho

2. Overview Why Is This Study Relevant? What Are The Challenges? SWAT
ISPSO
Model Calibration
Analysis Results
Conclusions
References

3. Why Is This Study Relevant?
Understand the impact of climate change only on water resources.
Most of previous studies
Used highly urbanized areas.
Used one representative land-cover data for a long-term assessment.
This study
Uses a watershed with minimum human activities.
Uses multiple land-cover sources for a long-term assessment.

4. What Are The Challenges?
Limited historical records
No surface run-off
No groundwater discharge
No soil water content
Little or no evapotranspiration
What do we have?
Data: streamflow, weather data, soil type, land-cover
Soil and Water Assessment Tool (SWAT)
Curse of dimensionality
Multi-modality, equifinality
Isolated-Speciation-based Particle Swarm Optimization (ISPSO)

5. What Are The Challenges? (Cont.)
Multi-modality: Global & local solutions
Equifinality: Equally good but substantially different
X Factor
Cost OF

6. SWAT A watershed-scale, semi-distributed continuous hydrologic model
Inputs
Topography, land-cover, soil  Hydrologic Response Units (HRUs)  A “LOT” of parameters
Precipitation, temperature
Simulates
Surface run-off
Soil water content
Groundwater discharge
Evapotranspiration

7. ISPSO Heuristic algorithm using swarm intelligence (Cho et al., 2011)
Finding multi-modal solutions while efficiently exploring the search space
Derivative-free optimization
Successfully applied to
Hydrology & Hydraulics
Stochastic rainfall generation
Storm tracking
Uncertainty estimation

8. Angry Birds in ISPSO!

9. ISPSO: Himmelblau Function
+: True Solutions, X: ISPSO Solutions, o: Particles

10. Model Calibration Study area: 2,221 km2 Three models:
Neches River Basin
Study area: 2,221 km2
Three models:
Period 1:
Period 2:
Period 3:
Objective function
Nash-Sutcliffe coefficient of daily streamflows

11. Model Calibration (Cont.)
Model parameters
12 parameters * 25 subbasins * 1-18 HRUs
300-5,400 parameter values
α-rule  12 α values  Still difficult to solve & Equifinality
Baseflow parameters
Pre-calibrated using a baseflow filter

12. Model Calibration (Cont.)
ISPSO: NS=0.623
AutoCal: NS=0.530
AutoCal built in SWAT did not perform well enough.

13. Model Calibration (Cont.)
NS vs. Model Runs
Choose the most realistic model parameters among different solutions.
Multi-Modality
Equifinality

14. Model Calibration (Cont.)
Model performance
Performance rating (Cho & Olivera, 2009)
NS >= 0.75: Very good
NS >= 0.65: Good
Land-cover data NS
Period 1 ( )
LULC
0.74
Period 2 ( )
NLCD1992
0.66
Period 3 ( )
NLCD2001
0.75

15. Temperature Change 5-year moving average H0: Slope=0 or no change
Ha: Slope≠0 or change
Statistical significance test
Increased by 0.7◦C
Not significant compared to previous studies
Low developed land  Low heat capacity
Mean
P-value
Period 1 (1970–1989)
18.8
0.45
Period 2 (1990–1999)
19.1
0.27
Period 3 (2000–2009)
19.5
0.06
Overall
0.02

16. Precipitation Change 5-year moving average H0: Slope=0 or no change
Ha: Slope≠0 or change
Statistical significance test
Increased by 16.3%
Not much different from previous studies
Not affected by urban development
Mean
P-value
Period 1 (1970–1989)
1333.7
0.67
Period 2 (1990–1999)
1495.3
1.00
Period 3 (2000–2009)
1551.6
0.90
Overall
1422.1
0.04

17. Land-Cover Change
Major change: Vegetation (grass, bush/shrub, forest)  Developed land
Vegetation
Barren
land
Crop
Developed
Water
Period 1
(1970–1989)
94.9%
0.7%
3.1%
1.1%
0.2%
Period 2
(1990–1999)
94.8%
0.6%
3.3%
Period 3
(2000–2009)
89.8%
0.1%
3.5%
6.3%
0.3%
Change
-5.1%
-0.6%
0.4%
5.2%

18. Hydrologic Components
Surface run-off  Precipitation (cf. urban watershed)
Groundwater discharge  Main source of agricultural and municipal water
Soil water content  Decrease in vegetation and increase in developed land
Evapotranspiration  Vegetation dominated
Surface run-off
Groundwater discharge
Soil water content
Evapotranspiration
Period 1 (1970–1989)
221.5
14.1
279.0
690.3
Period 2 (1990–1999)
243.1
13.2
283.8
770.4
Period 3 (2000–2009)
254.8
12.8
286.5
829.1
Change
15.0%
-9.2%
2.7%
20.1%

19. References
Huidae Cho, Dongkyun Kim, Francisco Olivera, Seth D. Guikema, August 16, 2011. Enhanced Speciation in Particle Swarm Optimization for Multi-Modal Problems. European Journal of Operational Research 213 (1),  doi: /j.ejor
Huidae Cho, Francisco Olivera, June Effect of the Spatial Variability of Land Use, Soil Type, and Precipitation on Streamflows in Small Watersheds. Journal of the American Water Resources Association 45 (3), doi: /j x
Joonghyeok Heo, Jaehyung Yu, John R. Giardino, Huidae Cho, Accepted in June 2014. How Important is Climate Change on the Water Resource in a Humid Subtropical Watershed?: A Case Study from East Texas, USA. Water and Environmental Journal. doi: /wej.12096

20. Conclusions
We used a semi-distributed model to evaluate the long-term impact of climate change on water resources components.
The main challenges were lack of historical records and the high dimensionality of the model.
We used a swarm-intelligence based heuristic algorithm to calibrate the model parameters.
Surface runoff was mainly affected by precipitation.
Groundwater discharge was mostly affected by human activities.
Soil water content was more sensitive to land-cover change than to climate change.
High evapotranspiration was caused by vegetation-dominated land-cover.