Analysis of Catchment Model Calibration Through Data Comparison

This report provides an in-depth analysis of catchment model calibration, focusing on the comparison of recorded and predicted data. The study utilizes the Soil and Water Assessment Tool (SWAT) to simulate hydrological processes within the Strathfield catchment, covering an area of approximately 67 km2. The methodology involves applying sensitivity analysis to determine the most critical parameters for model calibration, using a 24-hour, 5-minute data interval from April 23, 1989. The report details the SWAT model's components, including water balance calculations and the application of the SCE optimization algorithm. The findings present parameter values and data obtained from the study, including rainfall data, and discuss the application of the Storm Water Management Model (SWMM). The report highlights the importance of accurate data and model sophistication in achieving reliable calibration results. The analysis covers various parameters, including the number of subcatchments, channel connections, and water quality constituents, offering insights into the model's configuration and performance. The data includes detailed information on rainfall intensity and timing, which is crucial for simulating runoff and other hydrological processes. The study's results are presented in a format that facilitates understanding of the model's behavior under different conditions and provides a basis for further investigation into catchment hydrology.

ANALYSIS OF CALIBRATION OF A CATCHMENT MODEL THROUGH
COMPARISON OF RECORDED AND PREDICTED DATA
INTRODUCTION
Need for Modelling
In water management, hydrological models are critical in decision making especially in instances
where availability of discharge observations in space and time is limited (Beven, 2011). Forecast
of discharge in catchments, land use alteration on numerous hydrograph features, extension of
stream flow records or assessment of the consequence of climate change are among the common
applications for modelled discharge (Piniewski, 2014).
Hydrological models are composed of number of storages and fluctuations of water, which
describes the hydrological operation of a catchment. The fluctuations of water and storages are
computed by parameters. These parameters cannot be estimated from data collected in the
research and thus has to be generated by calibration. Calibration is done because of two main
reasons:
i) Parameters are a definite approximation to reality and in most cases do not represent
any physical attribute.
ii) Parameters which possess physical attributes are scaled but signifies a particularly
enormous scale in the model (J. Marsalek, 2012).
In calibration, objective function is applied in estimation of parameter values. The estimation is
achieved by reducing the difference between observed and computer-generated discharge using
objective function.
Model Performance
Initially, model calibration was dependent on discharge data. As technology advanced and more
inventions sprung, model calibration made use of snow water equivalent and ground water
dynamics were included in modelling (Seibert, 2018). Discharge is the prime information in
modelling and the easiest to use because it does not require much effort to obtain and process. It
is also the first water balance proponent in catchments.

Statistical metrics
Volume error, mean squared error and coefficient of determination were primarily applied in
model calibration. In hydrology, one of the widely used statistical metric is the Nash-Sutcliffe
efficiency. It is used to communicate model performance. Nash-Sutcliffe efficiency is a
dimensionless metric. It compares the ratio of the mean squared error among observed and
computer-generated discharge and the variance accompanying the data. Dependence on mean
squared error however results to undervaluation of discharge unpredictability and the variable
significance of discharge capacity in calibration as a function of a catchment’s discharge
unpredictability (Gupta, 2009).
In pure empirical model, the parameter adjustments is enabled using various optimization
techniques which can either be constrained or unconstrained. This plays a large role in avoiding
trial and error resolutions. Hydrologic models are applied to approximate the discharge features
of a watershed depending on the precipitation and the hydrologic features of the catchment.
Features of the catchment is described in a model by parameters and correlation of catchment
features which are applied to approximate the response of the catchment (John Wainwright,
2013).
Advancement in technology has led to improvement in simulation of hydrologic models. The
quality of the model is dependent on the expertise of the modeller, accuracy of the data and how
sophisticated the model is. Accuracy of data is very significant in determining the quality of
calibration. It is important to indicate the accuracy range of the data. This takes into account all
possible errors which may be incurred during data collection (Fricke, 2013). Having data which
covers an extremely large events results to more accurate and precise models.
INVESTIGATION METHODOLOGY
In this report we are going to utilise the Soil and Water Assessment Tool (SWAT). This tool was
invented to forecast the significance of land management practices on water, agricultural
chemical and sediment in large catchments. This model make use of readily available, physically
based and computationally efficient. The model is composed of plant growth, pesticides, soil
temperature, weather, land management, hydrology and nutrients (Dorota Swiatek, 2011).

In SWAT model, water balance is demonstrated by the following storage volumes; deep aquifer,
snow, shallow aquifer and soil profile. The soil profile is composed of evaporation, lateral flow,
infiltration and plant uptake. Other mechanisms include pumping withdrawals and seepage
(Sivakumar Bellie, 2010).
The optimization algorithm
The SCE system falls in the family of genetic algorithms. It starts by distributing points
stochastically over the space provided by parameters which fall within the lower and upper
bounds. Each point in this case is a representative of a part of the population. Each member is
described by a unique genetic information. By altering the parameter values (genetic
information), the population adjust towards an optimum objective function which creates a
bigger picture of the relationship between the data from the model and the measured data
(Gunter Bloschl, 2013).
Modelling a catchment using SWAT results to subdivision of the sample into subunits. The
subunits are then processed by use of constraints to produce the desired results.
The Catchment
Strathfield catchment was the point of focus. The catchment covers an area of approximately 67
km2. It is at an altitude of 231 m to 327 m above sea level. The slope is estimated at 11%.
RESULTS AND DISCUSSION
The model was calibrated for a 24-hour of 5-minute data interval collected on 23/04/1989. By
applying sensitivity analysis, the value of parameters was ascertained. The parameters were to be
given high priority during modelling.
The parameter values obtained from the model are as follows:
* Parameter Values on the Tapes Common Block *
***************************************************
Number of Subcatchments in the Runoff Block (NW)...... 1000
Number of Channel/Pipes in the Runoff Block (NG)...... 1000
Number of Connections to Runoff Channels/Inlets (NCP). 6
Number of Runoff Water Quality Constituents (NRQ)..... 10

Number of Runoff Land Uses per Subcatchment (NLU)..... 10
Number of Groundwater Subcatchments in Runoff (NGW)... 100
Number of Interface Locations for all Blocks (NIE).... 1000
Number of Elements in the Transport Block (NET)....... 300
Number of Storage Junctions in Transport (NTSE)....... 100
Number of Input Hydrographs in Transport (NTH)........ 80
Number of Tabular Flow Splitters in Transport (NTSP).. 50
Number of Elements in the Extran Block (NEE).......... 1400
Number of Pumps in Extran (NEP)....................... 75
Number of Orifices in Extran (NEO).................... 200
Number of Tide Gates/Free Outfalls in Extran (NTG).... 200
Number of Extran Weirs (NEW).......................... 60
Number of Extran Printout Locations (NPO)............. 30
Number of Tide Elements in Extran (NTE)............... 20
Number of Natural Channels (NNC)...................... 200
Number of Storage Junctions in Extran (NVSE).......... 300
Number of Time History Data Points in Extran (NTVAL).. 500
Number of Data Points for Variable Storage Elements
in the Extran Block (NVST).......................... 25
Number of Input Hydrographs in Extran (NEH)........... 400
Number of Allowable Channel Connections to
Junctions in the Extran Block (NCHN)................ 15
Number Rain Gages in Rain and Runoff (MAXRG).......... 200
Number PRATE/VRATE Points for Extran Pump
Input (MAXPRA)...................................... 10
Number of Variable Orifices in Extran (NVORF)......... 50
Number of Variable Orifice Data Points (NVOTIM)....... 50
Number of Allowable Precip. Values/yr in Rain (LIMRN). 5000
Number of Storm Events for Rain Analysis (LSTORM)..... 5000
Number of Plugs for Plug-flow in S/T (NPLUG).......... 3000

Number Conduits for Extran Results to ASCII
File (MXFLOW)....................................... 150
APPLICATION OF STORM WATER MANAGEMENT MODEL (SWMM)
POWELLS CREEK SIMULATION
Snowmelt parameter - ISNOW....................... 0
Number of rain gages - NRGAG..................... 1
Horton infiltration equation used - INFILM....... 0
Quality is not simulated - KWALTY................ 0
Default evaporation rate used - IVAP............. 0
Hour of day at start of storm - NHR.............. 3
Minute of hour at start of storm - NMN........... 0
Time TZERO at start of storm (hours)............. 3.000
Use Metric units for I/O - METRIC................ 1
===> Ft-sec units used in all internal computations
Runoff input print control... 0
Runoff graph plot control.... 0
Runoff output print control.. 2
Limit number of groundwater convergence messages to 10000 (if simulated)
Month, day, year of start of storm is: 4/23/89
Wet time step length (seconds)....... 15.
Dry time step length (seconds)....... 240.
Wet/Dry time step length (seconds)... 120.
Simulation length is...... 1200.0 Minutes
Percent of impervious area with zero detention depth 25.0
Horton infiltration model being used
Rate for regeneration of infiltration = REGEN * DECAY
DECAY is read in for each subcatchment
REGEN = ........................................ 0.01000
1
Rainfall from E3 Data Group

KTYPE - Rainfall input type.............. 0
NHISTO - Total number of rainfall values.. 330
KINC - Rainfall values (pairs) per line. 10
KPRINT - Print rainfall (0-Yes,1-No)...... 0
KTIME - Precipitation time units
0 --> Minutes 1 --> Hours............. 0
KPREP - Precipitation unit type
0 --> Intensity 1 --> Volume.......... 0
KTHIS - Variable rainfall intervals
0 --> No, > 1 --> Yes.................. 0
THISTO - Rainfall time interval........... 2.00
TZRAIN - Starting time (KTIME units)...... 180.00
The data obtained from the study was as follows:
213304 23/04/1989 00:00
0.058
213304 23/04/1989 04:10
0.363
213304 23/04/1989 08:20
0.153
213304 23/04/1989 00:05
0.057
213304 23/04/1989 04:15
0.350
213304 23/04/1989 08:25
0.131
213304 23/04/1989 00:10
0.055
213304 23/04/1989 04:20
0.235
213304 23/04/1989 08:30
0.195
213304 23/04/1989 00:15
0.055
213304 23/04/1989 04:25
0.172
213304 23/04/1989 08:35
0.341
213304 23/04/1989 00:20
0.054
213304 23/04/1989 04:30
0.137
213304 23/04/1989 08:40
0.280
213304 23/04/1989 00:25
0.054
213304 23/04/1989 04:35
0.122
213304 23/04/1989 08:45
0.196
213304 23/04/1989 00:30
0.053
213304 23/04/1989 04:40
0.105
213304 23/04/1989 08:50
0.162
213304 23/04/1989 00:35
0.052
213304 23/04/1989 04:45
0.095
213304 23/04/1989 08:55
0.140
213304 23/04/1989 00:40
0.051
213304 23/04/1989 04:50
0.090
213304 23/04/1989 09:00
0.212
213304 23/04/1989 00:45
0.051
213304 23/04/1989 04:55
0.082
213304 23/04/1989 09:05
0.248
213304 23/04/1989 00:50
0.049
213304 23/04/1989 05:00
0.076
213304 23/04/1989 09:10
0.245
213304 23/04/1989 00:55
0.049
213304 23/04/1989 05:05
0.070
213304 23/04/1989 09:15
0.234

213304 23/04/1989 01:00
0.049
213304 23/04/1989 05:10
0.063
213304 23/04/1989 09:20
0.193
213304 23/04/1989 01:05
0.049
213304 23/04/1989 05:15
0.058
213304 23/04/1989 09:25
0.162
213304 23/04/1989 01:10
0.049
213304 23/04/1989 05:20
0.053
213304 23/04/1989 09:30
0.142
213304 23/04/1989 01:15
0.049
213304 23/04/1989 05:25
0.051
213304 23/04/1989 09:35
0.134
213304 23/04/1989 01:20
0.049
213304 23/04/1989 05:30
0.050
213304 23/04/1989 09:40
0.199
213304 23/04/1989 01:25
0.049
213304 23/04/1989 05:35
0.050
213304 23/04/1989 09:45
0.214
213304 23/04/1989 01:30
0.049
213304 23/04/1989 05:40
0.049
213304 23/04/1989 09:50
0.175
213304 23/04/1989 01:35
0.049
213304 23/04/1989 05:45
0.049
213304 23/04/1989 09:55
0.149
213304 23/04/1989 01:40
0.049
213304 23/04/1989 05:50
0.049
213304 23/04/1989 10:00
0.141
213304 23/04/1989 01:45
0.049
213304 23/04/1989 05:55
0.048
213304 23/04/1989 10:05
0.137
213304 23/04/1989 01:50
0.049
213304 23/04/1989 06:00
0.048
213304 23/04/1989 10:10
0.131
213304 23/04/1989 01:55
0.049
213304 23/04/1989 06:05
0.058
213304 23/04/1989 10:15
0.127
213304 23/04/1989 02:00
0.049
213304 23/04/1989 06:10
0.084
213304 23/04/1989 10:20
0.135
213304 23/04/1989 02:05
0.049
213304 23/04/1989 06:15
0.085
213304 23/04/1989 10:25
0.172
213304 23/04/1989 02:10
0.049
213304 23/04/1989 06:20
0.079
213304 23/04/1989 10:30
0.200
213304 23/04/1989 02:15
0.049
213304 23/04/1989 06:25
0.072
213304 23/04/1989 10:35
0.191
213304 23/04/1989 02:20
0.049
213304 23/04/1989 06:30
0.060
213304 23/04/1989 10:40
0.174
213304 23/04/1989 02:25
0.049
213304 23/04/1989 06:35
0.053
213304 23/04/1989 10:45
0.160
213304 23/04/1989 02:30
0.049
213304 23/04/1989 06:40
0.050
213304 23/04/1989 10:50
0.180
213304 23/04/1989 02:35
0.049
213304 23/04/1989 06:45
0.048
213304 23/04/1989 10:55
0.248
213304 23/04/1989 02:40
0.049
213304 23/04/1989 06:50
0.048
213304 23/04/1989 11:00
0.389
213304 23/04/1989 02:45
0.049
213304 23/04/1989 06:55
0.048
213304 23/04/1989 11:05
0.857
213304 23/04/1989 02:50
0.049
213304 23/04/1989 07:00
0.049
213304 23/04/1989 11:10
0.877
213304 23/04/1989 02:55 213304 23/04/1989 07:05 213304 23/04/1989 11:15

0.049 0.049 0.678
213304 23/04/1989 03:00
0.050
213304 23/04/1989 07:10
0.049
213304 23/04/1989 11:20
0.773
213304 23/04/1989 03:05
0.049
213304 23/04/1989 07:15
0.049
213304 23/04/1989 11:25
0.844
213304 23/04/1989 03:10
0.048
213304 23/04/1989 07:20
0.049
213304 23/04/1989 11:30
0.873
213304 23/04/1989 03:15
0.048
213304 23/04/1989 07:25
0.049
213304 23/04/1989 11:35
0.891
213304 23/04/1989 03:20
0.048
213304 23/04/1989 07:30
0.049
213304 23/04/1989 11:40
0.671
213304 23/04/1989 03:25
0.048
213304 23/04/1989 07:35
0.050
213304 23/04/1989 11:45
0.555
213304 23/04/1989 03:30
0.048
213304 23/04/1989 07:40
0.050
213304 23/04/1989 11:50
0.493
213304 23/04/1989 03:35
0.048
213304 23/04/1989 07:45
0.050
213304 23/04/1989 11:55
0.439
213304 23/04/1989 03:40
0.048
213304 23/04/1989 07:50
0.050
213304 23/04/1989 12:00
0.396
213304 23/04/1989 03:45
0.048
213304 23/04/1989 07:55
0.259
213304 23/04/1989 12:05
0.363
213304 23/04/1989 03:50
0.048
213304 23/04/1989 08:00
0.232
213304 23/04/1989 12:10
0.338
213304 23/04/1989 03:55
0.048
213304 23/04/1989 08:05
0.267
213304 23/04/1989 12:15
0.316
213304 23/04/1989 04:00
0.048
213304 23/04/1989 08:10
0.236
213304 23/04/1989 12:20
0.301
213304 23/04/1989 04:05
0.048
213304 23/04/1989 08:15
0.189
213304 23/04/1989 12:25
0.285
213304 23/04/1989 12:30
0.273
213304 23/04/1989 16:40
0.089
213304 23/04/1989 20:50
0.075
213304 23/04/1989 12:35
0.262
213304 23/04/1989 16:45
0.089
213304 23/04/1989 20:55
0.075
213304 23/04/1989 12:40
0.248
213304 23/04/1989 16:50
0.086
213304 23/04/1989 21:00
0.075
213304 23/04/1989 12:45
0.233
213304 23/04/1989 16:55
0.085
213304 23/04/1989 21:05
0.075
213304 23/04/1989 12:50
0.215
213304 23/04/1989 17:00
0.085
213304 23/04/1989 21:10
0.075
213304 23/04/1989 12:55
0.202
213304 23/04/1989 17:05
0.085
213304 23/04/1989 21:15
0.075
213304 23/04/1989 13:00
0.193
213304 23/04/1989 17:10
0.085
213304 23/04/1989 21:20
0.075
213304 23/04/1989 13:05
0.184
213304 23/04/1989 17:15
0.085
213304 23/04/1989 21:25
0.075

213304 23/04/1989 13:10
0.174
213304 23/04/1989 17:20
0.085
213304 23/04/1989 21:30
0.075
213304 23/04/1989 13:15
0.165
213304 23/04/1989 17:25
0.085
213304 23/04/1989 21:35
0.075
213304 23/04/1989 13:20
0.157
213304 23/04/1989 17:30
0.085
213304 23/04/1989 21:40
0.075
213304 23/04/1989 13:25
0.155
213304 23/04/1989 17:35
0.085
213304 23/04/1989 21:45
0.075
213304 23/04/1989 13:30
0.149
213304 23/04/1989 17:40
0.085
213304 23/04/1989 21:50
0.075
213304 23/04/1989 13:35
0.144
213304 23/04/1989 17:45
0.085
213304 23/04/1989 21:55
0.075
213304 23/04/1989 13:40
0.137
213304 23/04/1989 17:50
0.085
213304 23/04/1989 22:00
0.075
213304 23/04/1989 13:45
0.133
213304 23/04/1989 17:55
0.085
213304 23/04/1989 22:05
0.075
213304 23/04/1989 13:50
0.129
213304 23/04/1989 18:00
0.077
213304 23/04/1989 22:10
0.075
213304 23/04/1989 13:55
0.127
213304 23/04/1989 18:05
0.077
213304 23/04/1989 22:15
0.075
213304 23/04/1989 14:00
0.127
213304 23/04/1989 18:10
0.077
213304 23/04/1989 22:20
0.075
213304 23/04/1989 14:05
0.121
213304 23/04/1989 18:15
0.077
213304 23/04/1989 22:25
0.075
213304 23/04/1989 14:10
0.116
213304 23/04/1989 18:20
0.077
213304 23/04/1989 22:30
0.075
213304 23/04/1989 14:15
0.112
213304 23/04/1989 18:25
0.077
213304 23/04/1989 22:35
0.075
213304 23/04/1989 14:20
0.110
213304 23/04/1989 18:30
0.076
213304 23/04/1989 22:40
0.075
213304 23/04/1989 14:25
0.108
213304 23/04/1989 18:35
0.076
213304 23/04/1989 22:45
0.075
213304 23/04/1989 14:30
0.106
213304 23/04/1989 18:40
0.076
213304 23/04/1989 22:50
0.074
213304 23/04/1989 14:35
0.104
213304 23/04/1989 18:45
0.076
213304 23/04/1989 22:55
0.074
213304 23/04/1989 14:40
0.103
213304 23/04/1989 18:50
0.076
213304 23/04/1989 23:00
0.074
213304 23/04/1989 14:45
0.101
213304 23/04/1989 18:55
0.076
213304 23/04/1989 23:05
0.074
213304 23/04/1989 14:50
0.100
213304 23/04/1989 19:00
0.076
213304 23/04/1989 23:10
0.074
213304 23/04/1989 14:55
0.099
213304 23/04/1989 19:05
0.076
213304 23/04/1989 23:15
0.074
213304 23/04/1989 15:00
0.116
213304 23/04/1989 19:10
0.076
213304 23/04/1989 23:20
0.074
213304 23/04/1989 15:05 213304 23/04/1989 19:15 213304 23/04/1989 23:25

0.143 0.076 0.074
213304 23/04/1989 15:10
0.135
213304 23/04/1989 19:20
0.076
213304 23/04/1989 23:30
0.074
213304 23/04/1989 15:15
0.126
213304 23/04/1989 19:25
0.076
213304 23/04/1989 23:35
0.074
213304 23/04/1989 15:20
0.117
213304 23/04/1989 19:30
0.076
213304 23/04/1989 23:40
0.074
213304 23/04/1989 15:25
0.111
213304 23/04/1989 19:35
0.076
213304 23/04/1989 23:45
0.074
213304 23/04/1989 15:30
0.106
213304 23/04/1989 19:40
0.076
213304 23/04/1989 23:50
0.074
213304 23/04/1989 15:35
0.103
213304 23/04/1989 19:45
0.076
213304 23/04/1989 23:55
0.074
213304 23/04/1989 15:40
0.099
213304 23/04/1989 19:50
0.076
213304 24/04/1989 00:00
0.074
213304 23/04/1989 15:45
0.096
213304 23/04/1989 19:55
0.076
213304 23/04/1989 15:50
0.095
213304 23/04/1989 20:00
0.076
213304 23/04/1989 15:55
0.094
213304 23/04/1989 20:05
0.076
213304 23/04/1989 16:00
0.093
213304 23/04/1989 20:10
0.076
213304 23/04/1989 16:05
0.093
213304 23/04/1989 20:15
0.076
213304 23/04/1989 16:10
0.092
213304 23/04/1989 20:20
0.076
213304 23/04/1989 16:15
0.091
213304 23/04/1989 20:25
0.076
213304 23/04/1989 16:20
0.090
213304 23/04/1989 20:30
0.076
213304 23/04/1989 16:25
0.089
213304 23/04/1989 20:35
0.076
213304 23/04/1989 16:30
0.089
213304 23/04/1989 20:40
0.075
213304 23/04/1989 16:35
0.089
213304 23/04/1989 20:45
0.075
The data was processed and calculated by the calibrated model and compared to the observed
data. A graph obtained from the tabulated data above was as shown below.

1 15 29 43 57 71 85 99 113 127 141 155 169 183 197 211 225 239 253 267 281
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time in interval of 5 minutes
runoff (mm/d)
The correlation between the two sets of data was calculated and presented as shown in the table
below.
Observed Calculated
Mean stream flow 0.076 0.073
Standard deviation 0.083 0.077
Modal efficiency 0.69
Correlation 0.80
CONCLUSIONS
The model efficiencies obtained are dictated by the variability of the data and thus cannot be
directly compared. The level of efficiency is obtained when the previous findings are compared
to the current data.
From the results, it is clear that distributed hydrologic models that are complex can be
satisfactorily be automatically calibrated.

References
Beven, Kevin. &. Wilson., 2011. On red herrings and real herrings: Disinformation and information in
hydrological inference. In: s.l.:Hydrological Processes, pp. 1600-1720.
Dorota Swiatek, Teddy O., 2011. Modelling of Hydrological Processes in the Narew Catchment. In:
GeoPlanet: Earth and Planetary Sciences. s.l.:Springer Science & Business Media, pp. 33-39.
Fricke, K., 2013. Analysis and Modelling of Water Supply and demand under climate change, Land use
transformation and Socio-Economic Development: The Water Resource Challenge and Adaptation
Measures for Urumqi. In: s.l.:Springer Science & Business Media, pp. 117-129.
Gunter Bloschl, 2013. Runoff Prediction in Ungauged Basins: Synthesis Acros Processes, Places and
Scales. In: s.l.:76-101, p. Cambridge University Press.
Gupta, 2009. Decomposition of the mean squared error and NSE performance criteria: Implications for
improving hydrological modelling. In: s.l.:Journal of Hydrology, pp. 56-67.
J. Marsalek, 2012. Flood Issues in Contemporary Water Management. In: s.l.:Springer Science & Business
Media, pp. 332-376.
John Wainwright, Michael M., 2013. Environmental Modelling: Finding Simplicity in Complexity. In:
s.l.:John wiley & Sons, pp. 222-254.
Piniewski, M., 2014. Scenario-based impact assessment of global and regional change on the semi-
natural flow regime. In: s.l.:Anchor Academic Publishing, pp. 65-69.
Seibert, 2018. Upper and lower benchmarks in hydrological modelling. In: s.l.:Hydrological Processes, pp.
1132-1148.
Sivakumar Bellie, Brian R., 2010. Advances In Data-based Approaches For Hydrologic Modelling And
Forecasting. In: s.l.:World scientific, pp. 324-333.