Verifying (testing) a stormwater biofiltration hydrologic model Anna Lintern, E. Daly, A. Deletic Monash Water for Liveability, CRC for Water Sensitive Cities Monash University
Verifying (testing) a stormwater biofiltration hydrologic model
Anna Lintern, E. Daly, A. Deletic
Monash Water for Liveability,
CRC for Water Sensitive Cities
Monash University
Background: • Management of stormwater is a problem
• One solution = biofiltration systems
Background Hydrologic
model Methods Results Conclusion
Background:
• Restores hydrology of
pre-urbanised catchment
• Models are invaluable
Piped outflow Exfiltration
Background Hydrologic
model Methods Results Conclusion
• Incorporated in software
Objective of project
To test a stormwater biofilter hydrologic model using an operating field system.
Background Hydrologic
model Methods Results Conclusion
Infiltration
Overflow Rainfall
Stormwater
inflow
ET Infiltration
Capillary Rise
Exfiltration
How the model works
Background Hydrologic
model Methods Results Conclusion
Ponding
Zone
Filter
media
Submerged zone
Pipe
Outflow Drainage
layer
Untreated
overflow
Rain Stormwater
inflow
Ponding Zone
Infiltration
We
ir h
eig
ht
Hydraulic conductivity of
the filter media, k
Ponding Zone
Filter media
Zone
Evapotranspiration
- Hydraulic conductivity of the filter media, k
- Soil moisture
Infiltration
Hydraulic conductivity of
the filter media, k
Ponding Zone
Filter media
Zone
Evapotranspiration
Drainage layer
Zone
Infiltration Capillary rise
Exfiltration to surrounding soil
Infiltration
Piped outflow
Methodology
Monash University Carpark biofilter
Background Hydrologic
model Methods Results Conclusion
Cell 2 Cell 3 Cell 1
Sandy Loam with Compost
and Mulch Filter media
Methodology
• Available data:
• Continuous inflow and outflows at 1 minute timesteps
• 8 months
• Model calibration
• Calibration parameter = hydraulic conductivity
• Calibrated for flow rates
• Objective functions:
• Nash and Sutcliffe coefficient, E:
• Unbiased function, φ:
• Model performance assessed in terms of:
• Outflow flow rates
• Outflow volumes
Background Hydrologic
model Methods Results Conclusion
Results:
Background Hydrologic
model Methods Results Conclusion
• All results use corrected input data
Results: Hydraulic Conductivity
Background Hydrologic
model Methods Results Conclusion
Measured k Coefficient of efficiency Unbiased function
k E k φ
Cell 1 123 mm/hr 190 mm/hr 0.59 120 mm/hr 4.08
Cell 2 144 mm/hr 180 mm/hr 0.91 170 mm/hr 0.24
Cell 3 77 mm/hr 50 mm/hr 0.68 60 mm/hr 0.4
Results: Flow Rates
Background Hydrologic
model Methods Results Conclusion
3. Event timing/duration predicted successfully
1. Flows are sometimes over or underestimated
2. Model performance affected by objective function
Results: Flow Rates
Background Hydrologic
model Methods Results Conclusion
Results: Volumes
Background Hydrologic
model Methods Results Conclusion
1. Total volumes are generally well modelled
Cu
mu
lati
ve
Ou
tflo
w V
olu
me (
m3)
(φ)
Results: Volumes
Background Hydrologic
model Methods Results Conclusion
Cu
mu
lati
ve
Ou
tflo
w V
olu
me (
m3) 18%
(φ)
Results: Volumes
Background Hydrologic
model Methods Results Conclusion
Cu
mu
lati
ve
Ou
tflo
w V
olu
me (
m3)
9% 12%
(φ)
Although the model works quite well, there are some limitations:
1. Model limitations: a. Outflows are sometimes over- or under- estimated 2. Assumptions:
a. Biofilter represented as a 1D system b. Homogeneity of hydraulic conductivity
Background Hydrologic
model Methods Results Conclusion
Model limitations and assumptions
Background Hydrologic
model Methods Results Conclusion
Hydraulic conductivity affected by: 1)Macropores 2) Soil porosity
Particle size
Particle shape
Roots
Compaction
Conclusions
1. A hydrologic model of a stormwater biofilter was tested 2. Main conclusions about model performance: a. Model is working well i. Appropriate for conceptual design stage ii. Appropriate for modelling pollutant loads b. Model performance dependent on objective function choice c. Model is not perfect 3. Further work: a. Calibrated model should be verified on an independent data set b. Uncertainty should be assessed
Background Hydrologic
model Methods Results Conclusion
Thank you