ArcGIS Python-based Hybrid Hydrologic Model (Distributed-Clark)

for Spatially Distributed Rainfall-Runoff Generation and Routing

Younghyun Cho1, 2 and Bernard A. Engel2 | Email: [email protected] Seventh Symposium on Advances in Modeling and Analysis Using Python | Seattle, WA
1Hydrometeorological Cooperation Center, K-water (Korea Water Resources Corporation), Gwacheon, Korea, The Republic of
2Department of Agricultural and Biological Engineering, Purdue University, West Lafayette, IN 47907
Advances in Modeling and Analysis Using Python, Poster 316 | January 23, 2017

Introduction Implementation Application

• A GIS-based hybrid hydrologic model (conceptually lumped and • DistributedClark_10.1 (storm event version) • Watershed, Gauge, Radar Locations, Land use, Soil, and CN map
distributed feature model), Distributed-Clark, was developed for spatially
distributed rainfall-runoff flow prediction (Cho, 2016). In this model, the • Deriving time-area histogram and unit hydrograph using Watershed Pre-
SCS curve number estimated spatially distributed excess rainfall and processing and Unit Hydrograph Toolbox (Python Script Tools)
GIS-derived time-area diagram (isochrones) based on a set of separated - input data: DEM, outlet point, NLCD land use, and NEXRAD grid
unit hydrographs are utilized to calculate a direct runoff hydrograph.
• Development of the Distributed-Clark model includes four main steps:
watershed pre-processing, spatially distributed excess rainfall
estimation, which includes NEXRAD precipitation data processing (Cho
and Engel, 2017), spatially distributed unit hydrograph derivation,
and direct runoff hydrograph convolution.
• For this implementation, ArcGIS 10.1 was used as the GIS platform to
build and execute the Python script tools (DistributedClark_10.1;
Python Toolbox, storm event ver.) for each step.

Model Description [ Muscatatuck River near Deputy, IN ]

• Procedures: Distributed-Clark development & NEXRAD data processing • Storm event (4 cases, 2002 to 2006) simulation results
Time-Area Histogram Map
(Isochrones)
Runoff routing
: Separated unit hydrograph
Spatially distributed
unit hydrograph

Observed Simulated (distributed; NEXRAD data) Simulated (averaged; gauged data) < Parameters: i 2.25 mm/hour, R 7.5 hour >
• NEXRAD precipitation data processing for spatiotemporally varied rainfall 150
MUSCATATUCK (event #2:
#1 Nov. 09-14, 2002)
200
MUSCATATUCK (event #4:
#2 Oct. 18-21, 2004)
350
MUSCATATUCK (event #3
#5: Mar. 27-31, 2005)
200
MUSCATATUCK (event #6:
#4 Oct. 27-30, 2006)

inputs; Runoff can be obtained using Excess Rainfall Toolbox 120 160 280 160

90 120 210 120

- input data: NEXRAD NCEP Stage IV product and soil map for CN 60 80 140 80

30 λ: 0.112 (II) 40 λ: 0.213 (I) 70 λ: 0.068 (II) 40 λ: 0.261 (III)

0 0 0 0
23 47 71 95 119 143 8 32 56 80 12 36 60 84 108 3 27 51 75

DRH: ENS 0.82, R2 0.87, PBIAS -1.87% DRH: ENS 0.97, R2 0.97, PBIAS - % DRH: ENS 0.99, R2 0.99, PBIAS -0.38% DRH: ENS 0.98, R2 0.99, PBIAS -0.15%
(NEXRAD: ENS 0.04↓, R2 0.03↓) (NEXRAD: ENS 0.01↓, R2 0.01↓) (NEXRAD: ENS 0.02↑, R2 0.02↑) (NEXRAD: ENS 0.09↑, R2 0.06↑)

Results
• A model case study of single storm event application for a river basin was
NEXRAD data conducted; the Mucatatuck River near Deputy, IN using NEXRAD
Cumulative precipitation
map for 27-30 Oct 2006 precipitation product demonstrated relatively good fit (direct runoff ENS
0.94, R2 0.96, and PBIAS -0.60%) against observed streamflow as well as
a slightly better fit (direct runoff; ENS of 0.2% and R2 1.0%) in comparison
with the outputs of spatially averaged rainfall data simulations.

References
• Cho, Y., 2016: Development and evaluation of a watershed-scale hybrid
hydrologic model. Ph.D. dissertation, Dept. of Agricultural and Biological
Convolution tool for
Eng., Purdue University, 211 pp.
Direct Runoff Hydrograph
• Cho, Y. and B. A. Engel, 2017: NEXRAD Quantitative Precipitation
Runoff generation
: Spatially distributed excess rainfall
Estimations for Hydrologic Simulation Using a Hybrid Hydrologic Model.
J. Hydrometeor., 18, 25-47.