Relevant Information on SEK Model Calibration
INTRODUCTION
The water quality model calibration and assessment together with the hydrodynamic, water quality and pollution loading surveys were carried out as a part of the pre-EIA works under the Kai Tak Planning Review (KTPR) (1). The primarily objective was to obtain reliable hydrodynamic and water quality data for the model calibration in order to assess the water quality impacts from the Kai Tak Development (KTD).
This Appendix present relevant information on the set up and calibration of the hydrodynamic and water quality model namely “SEK model” extracted from the Final Model Calibration Report prepared under the KTPR. The SEK model has been used as the basis for modelling of hydrodynamic and water quality conditions in the Study Area under this EIA Study.
DESCRIPTION OF THE MODEL
General
The modelling exercise utilized the Delft3D suite of models as the modelling platform. The Deflt3D-FLOW module and the Delft3D-WAQ module were used for hydrodynamic simulations and water quality simulations respectively.
The development of the SEK model was based on the model setup of a regional Update model. The Update model is a fully calibrated and verified model developed under the EPD Update Study (2). The Update model covers the whole HKSAR waters and the adjacent Mainland waters. Figure 2.1 of this Appendix shows the grid layout of the Update model.
Grid Layout
The SEK model is a cut out from the Update model. The model grid of the SEK Model was refined in Kai Tak Approach Channel (KTAC) and Kowloon Bay to give a better representation of the hydrodynamic and water quality conditions. Figure 2.2a of this Appendix shows the grid layout for the SEK model. The areas covered by the SEK model included the Victoria Harbour water control zone and the adjacent outer waters. A comparison of Update and SEK model grid in the KTAC and Kowloon Bay is shown in Figure 2.2b.
The SEK model consists of 4010 grid cells. The smallest grid cells are located in the KTAC which are less than 70m x 50m. The largest grid size is approximately 2000m x 2000m at the northwest boundary of the model. The existing coastline configuration for year 2005 as shown in Figure 2.2a was used in the model calibration exercise.
Bathymetry Schematization
The bathymetry schematization of the SEK model was based on the depth data from the marine charts (third edition 24 September 2004) produced by the Hydrographic Office, Marine Department, HKSARG. The bathymetry in KTAC, KTTS and Kowloon Bay was also updated with reference to the data collected under the “Comprehensive Feasibility Study for The Revised Scheme of South East Kowloon Development (CFS)”. Figure 2.4a of this Appendix presents graphically the bathymetry schematization of the SEK model. A close up of the bathymetry schematization in the Study Area is shown in Figure 2.4b of this Appendix The Update model bathymetry is also included in Figure 2.4b of this Appendix for comparison. The reference level used in the SEK model was Principal Datum Hong Kong and the depth data were relative to this datum.
(1)
South East Kowloon Comprehensive Planning and Engineering Review Stage 1: Planning Review
(2)
Agreement No. CE 42/97, Update on Cumulative Water Quality and Hydrological Effect of Coastal Developments and Upgrading of Assessment Tool
SPECIFICATIONS OF FIELD SURVEYS
An extensive field surveys were carried out as part of the KTPR in October 2005 and January 2006 for calibration of the SEK model including:
Marine Hydrodynamic Field Survey
Hydrodynamic field survey was conducted at fourteen marine stations including seven stations within the approach channel (AC1-7), one station at the KTTS (KT1), three stations at inner Kowloon Bay (IB1-3), one station at outer Kowloon Bay (OB1) and two stations in the Victoria Harbour adjacent to the Kowloon Bay (VH1-2). Figure 3.2 of this Appendix shows the survey locations.
Current Measurements
Current measurements were conducted at KTTS (KT1), Kowloon Bay (IB1-3), outer Kowloon Bay (OB1) and Victoria Harbour (VH1-2). The parameters measured included:
- Water current (including flow speed and direction),
- Water depth,
- Temperature,
- Salinity.
Water current, temperature and salinity vertical profiles were measured hourly at five vertical depths (namely, 10%, 30%, 50%, 70% and 90% of the water depth from surface) at each of the designated locations.
Dye Tracing
Fluorescent dye was injected at the mouth of Kai Tak Nullah at the first higher high water (HHW) of the marine survey period. Fluorometric dye tracing from KTN was conducted at stations AC1-7, KT1 and VH2. The dye tracing was carried out together with the water current measurements. The dye measurement frequency at these stations was once every 3 hours at different water depths as described in the above section for comparison with the model predictions.
Survey Period
Field surveys were carried out in October 2005 and January 2006. In each survey, two survey events were conducted and each survey event covered a 26-hour complete tidal cycle as shown in Table 1. The hydrodynamic monitoring period was the same as the period of gathering marine water quality monitoring data as discussed in later sections of this Appendix.
Table 1 Marine Survey Periods
| Tide | Start | Finish | Weather | Average Air Temperature (°C) with range | |
|---|---|---|---|---|---|
| First Survey | Event 1 | 5 October 2005 17:00 | 6 October 2005 18:00 | Fine | 28.2 (26.3 – 30.2) |
| Event 2 | 13 October 2005 4:00 | 14 October 2005 5:00 | Fine | 27.0 (25.4 – 29.3) | |
| Second Survey | Event 3 | 15 January 2006 11:00 | 16 January 2006 12:00 | Fine | 19.7 (18.4 – 21.7) |
| Event 4 | 20 January 2006 13:00 | 21 January 2006 14:00 | Cloudy (with occasionally light rainfall) | 14.2 (11.5 – 18.3) | |
Pollution Loading Survey
Pollution loading surveys were carried out at the outfalls of catchments N, P and Q (refer to Figure 4.1 of this Appendix) in the approach channel during October 2005 and January 2006 under the KTPR. Catchments N and P represents the catchments of Kai Tak Nullah (KTN) and Jordan Valley Box Culverts (JVBC) respectively. The pollution loading survey also covered 8 storm outfalls in Kowloon Bay. Total four loading survey events were carried out, two in October 2005 and two in January 2006. The loading survey periods for KTN and JVBC are shown in Table 2 below. Each survey event for KTN and JVBC covered a continuous period of 3 days. For the remaining storm outfalls (i.e. outfall of catchment Q and 8 outfalls in Kowloon Bay), the survey event covered only the last 24 hours of the survey period shown in Table 2. Only trace rainfall was recorded during the survey periods in October 2005 and no rainfall was recorded during the survey periods in January 2006. Loading compiled from survey data was used for model calibration.
Table 2 Pollution Load and Flow Survey Periods
| Season | Survey Event | Survey Period |
|---|---|---|
| First Survey | 1 | 2 to 5 October 2005 |
| 2 | 9 to 12 October 2005 | |
| Second Survey | 3 | 12 to 15 January 2006 |
| 4 | 17 to 20 January 2006 |
KTN and JVBC
For the loading survey conducted at the outfall of KTN and JVBC, separate water samples were collected for surface and bottom layers, as the outfalls of KTN and JVBC are subject to strong tidal influence. Therefore, flow rates are required to be estimated separately for surface and bottom layers at each measurement interval. For survey event 3 and event 4 (conducted in dry season), vertical flow velocity profiles and vertical salinity profiles were both taken at 0.3 m depth interval. The upper freshwater upstream flow, if any, was distinguished by reviewing the vertical salinity profile as illustrated in the figure below. If no distinct salinity profile could be identified, the boundary of upper and lower water layers was set at the middle depth of water. The loading was then compiled by applying the measured pollutant concentrations in the water samples to the corresponding estimated flow rates.
For survey event 1 and event 2 (conducted in October 2005), salinity data are only available for two water depth only, namely 0.3 m below the water surface and 0.3 m above the seabed, whilst vertical flow velocity profiles were taken at 0.3 m depth interval. It was therefore proposed to use the following approach for layer division: In case when velocities include positive values in upper water layer and negative values in lower water layer (as in figure below), surface layer and bottom layer were divided at the boundary of positive velocity and negative velocity. In any other cases, boundary was set at the middle depth of water.
The approach adopted for event 1 and event 2 for layer division was also used to estimate the loading for event 3 and event 4 for comparison. It was found that due to the lack of vertical salinity profiles for layer division, the loading for BOD5 and NH3-N would be underestimated by less than 10% and 20% respectively for both event 3 and event 4, whilst no significant difference was observed for the remaining parameters. It is therefore concluded that the change of method for data manipulation would not cause significant difference in the pollution loading results. Nevertheless, the loading for event 1 and event 2 was factored up by
1.1 (for BOD5) and 1.2 (for NH3-N) to account for the differences. The adjusted loading was adopted for sensitivity model runs. Sensitivity runs indicated that there would not be any significant difference in the model simulation results due to the use of the adjusted loading.
Catchments O, R, S, T, U, V, W, X, ZG, Y and Z
Pollution loading inventory for catchments O, R, S, T, U, V, W, X, ZG, Y and Z (see Figure
4.1 of this Appendix) within or adjacent to KTAC and KTTS was compiled from desktop calculations. Following the same approach adopted in the EPD Update Study, relevant per head flow and load were assigned to residential, transient, commercial and industrial population to obtain the quantity and quality of total untreated wastewater by individual drainage catchments. The latest population and employment statistics provided by Planning Department were used for the estimation. It was assumed that only 10 % of the total loading generated in the catchment would be discharged into the marine waters via storm drains whilst the rest of the flow would be diverted to the sewerage system. These pollution loads were essentially the dry weather loads. The rainfall related pollution loads for each drainage catchment were estimated with reference to the methodology adopted in the EPD Update Study using the rainfall data recorded during year 2004. The pollution loading inventory for KTAC and KTTS has also taken into account the effect of existing dry weather flow interceptors (DWFI) installed at catchments S, U, W and X (refer to Figure 4.1). The effectiveness of the DWFI was derived by analysis of rainfall data recorded in 2004 from the Hong Kong Observatory.
Others
For areas outside KTAC, KTTS and Kowloon Bay, the pollution loads available from the approved Tai Po EIA (3) for the year of 2003 were used for model input. The inventory compiled under the Tai Po EIA was based on the EPD Update Study and is available for the whole HKSAR and adjacent Mainland waters.
Vertical Allocation of Pollution Load
The vertical allocation of pollution loads from storm outfall discharges and sewage outfalls was made consistent with the approach adopted in the EPD Update Study. The pollution loads from storm outfall discharges (including dry weather flows and storm runoff) were specified in the surface layer whilst the pollution loads from sewage outfalls were allocated in the middle layer.
Marine Water Quality Survey
Water quality model simulations were performed using the SEK Model and the water quality model results were compared with the marine survey data collected in October 2005 and January 2006. The water quality survey periods were the same as the hydrodynamic survey periods as shown in Table 1. The water quality survey locations include seven stations within the approach channel (AC1-7), one station at the KTTS (KT1), three stations at inner Kowloon Bay (IB1-3), one station at outer Kowloon Bay (OB1) and two stations in the Victoria Harbour adjacent to the Kowloon Bay (VH1-2) as shown in Figure 3.2 of this Appendix Water quality measurements were conducted at regular intervals within the survey periods at various vertical water depths at the 14 designated survey stations.
In both October and January surveys, two survey events were carried out for typical spring and neap tides respectively. For each survey event, the measurements were carried out at the frequencies specified in Table 3 and Table 4 at all designated monitoring locations to cover a complete tidal cycle (roughly a 26-hour period). Table 3 and Table 4 summarise the survey programme for different water quality parameters. The measurement period of marine water quality was the same as the period of gathering hydrodynamic data.
Table 3 Summary of Baseline Monitoring Programme for Marine Water Quality (Stations KT1, IB1-3, OB1, VH1-2)
| Parameters | Frequency | Analytical Condition |
|---|---|---|
| Suspended solids (SS), ammonia nitrogen (NH3-N), unionized ammonia (NH4), E. coli, nitrite (NO2), nitrate (NO3), total kjeldahl nitrogen (TKN) and chlorophyll a | Every three hours | Laboratory Analysis |
| Secchi disc depth | Every three hours during daylight | In situ |
| Dissolved oxygen (DO), turbidity, pH, temperature, salinity | Every hour | In situ |
Table 4 Summary of Baseline Monitoring Programme for Kai Tak Approach Channel Water Quality (Stations AC1-7)
| Parameters | Frequency | Analytical Condition |
|---|---|---|
| Suspended solids (SS), 5-day biochemical oxygen demand (BOD5), chemical oxygen demand (COD), total organic carbon (TOC ammonia nitrogen (NH3-N), E. coli, nitrite (NO2), nitrate (NO3), total kjeldahl nitrogen (TKN) and chlorophyll a | Every three hours | Laboratory Analysis |
| Secchi disc depth | Every three hours during daylight | In situ |
| Sampling depth, dissolved oxygen (DO), pH, turbidity, temperature, salinity, water depth | Every hour | In situ |
Water samples were taken at four water depths, namely water surface, 1 m below water surface, mid-depth, and 1 m from seabed. In case the water depth was found to be less than 6 m, the mid-depth measurement was omitted. If water depth was found to be less than 3 m, only the mid-depth position was monitored.
In addition to the water quality parameters, other relevant data were also measured, including monitoring location / position, time, weather condition, tidal stage (where appropriate), direction of current and any special phenomena in the vicinity of the study area. Meteorological data of the 8-day period, from 2 days before the commencement of the pollution load survey to 1 day after the end of the marine survey (see Table 1), including hourly air temperature, rainfall, water temperature, tidal elevation, air pressure, wind velocity, cloud cover and solar radiation, were obtained from the Hong Kong Observatory for model calibration.
HYDRODYNAMIC MODELLING
Introduction
Performance of the SEK hydrodynamic model was reviewed through the comparison of the model results predicted by the SEK model with those of the Update model. The field survey data collected at KTAC, Kwun Tong Typhoon Shelter (KTTS) and Kowloon Bay as part of this Study in October 2005 and January 2006 were used to verify the performance of the SEK Model.
Vertical Layers
The hydrodynamic model of the SEK model is 3 dimensional with a total of 10 vertical water layers. The thickness of each water layer is defined in the model as a percentage of the water depth where the total sum of all the vertical layers should be 100%. All the vertical layers of the SEK model were assigned to have the same vertical contribution. Thus, each of the vertical layers in the SEK model contributes 10% of the total water depth.
Boundary and Initial Conditions
The SEK Model was linked to the regional Update model. The hydrodynamic model of the Update Model was fully calibrated and verified and was used to provide open boundary conditions to drive the SEK model through a nesting process. The Update Model covers the outer regions of Pearl River Estuary, Macau, Lamma Channel and Deep Bay. All major influences on hydrodynamics in the outer regions were therefore incorporated into the SEK model. In particular, the Update model covers the major Pearl River estuary discharges including the flow from Humen, Jiaomen, Hongqili, Hengmen, Muodaomen, Jitimen, Hutiaomen and Aimen. The influence on hydrodynamics due to the Pearl River estuary discharges would be transferred to the SEK model through the nesting process at the open boundaries.
The Update model also provided initial hydrodynamic conditions to the SEK model through a mapping process. Therefore, both models started with the same initial conditions.
Model Consistency Check
The regional Update model is able to provide reliable prediction of hydrodynamic conditions in HKSAR waters as demonstrated in the EPD Update Study. The SEK hydrodynamic model results were checked against the results from the approved Update model.
Simulation Periods
The hydrodynamic model covered both dry season and wet season simulations. Each simulation period covered a 15-day full spring-neap cycles, preceded by a spin-up period. The simulation periods adopted in this modelling exercise are shown below:
Dry season
Actual Simulation period (excluding the spin-up period): 9 January 2006 to 24 January 2006
Wet season
Actual Simulation period (excluding the spin-up period): 1 October 2005 to 10 October 2005
An adequate spin-up period of 23 days was adopted before the actual model simulations.
Wind
For the purpose of model consistency check, both the Update model and the SEK model adopted an average wind speed of 5 m/s for both the dry and wet season model simulations. The wind direction was from north-east for the dry season case whilst the wind direction was set to be from south-west for the wet season case.
Discharges
The discharges from the storm and sewage outfalls within the HKSAR waters would not change the overall hydrodynamic regime in the HKSAR waters and therefore not included in the hydrodynamic model for the purpose of model performance check. Averaged fresh water inflows from rivers within the Pearl River estuary for the dry season and wet season were applied separately in the dry and wet season simulations of the Update model. The Update model was subsequently used to provide open boundary conditions to the SEK model.
Model Set-up for Performance Verification
Model Simulation Periods
The simulation periods for model verification were the same as those for model consistency check. These simulation periods were designed to cover the field survey periods as shown in Table 1 and Table 2.
The model simulation periods consist of a spin-up period of 23 days plus an actual simulation period of 15 days (total 38 days). Model test runs were conducted to check whether the spin-up period of 23 days are sufficient by continuing the model run for one more spring-neap cycle. The salinity results of this additional spring-neap cycle were checked against those of the 15-day actual simulation period at Stations AC1, AC3 and AC5 within the KTAC. The salinity results for these two consecutive model runs were found to be consistent with each other which implied that a spin-up period of 23 days are sufficient to produce acceptable modelling results.
Wind
Actual hourly wind data recorded at the Kai Tak weather station during the simulation periods were used for the model input.
WATER QUALITY MODELLING
Simulation Period
The simulation periods for the dry and wet seasons water quality modelling are listed as follows:
Dry season Actual simulation period: 9 January 2006 to 24 January 2006
Wet season Actual simulation period: 1 October 2005 to 10 October 2005
The actual simulation period covered 15 days as shown above. A spin up period of 45 days was provided before actual model simulation. Thus, the model runs were performed for a total of 60 days. The spin-up period has been checked and is considered to be sufficient.
Meteorological Forcing
Meteorological forcing including solar surface radiation and water temperature was defined in the water quality model. The solar radiation and water temperature were assumed to be constant over the entire modelling area of the model.
The hourly solar radiation data recorded at King’s Park station by Hong Kong Observatory during the simulation periods were used for model input.
Flow Aggregation
A 2 x 2 spatial aggregation of the model grid was performed to optimize the computational run time and data storage. No vertical aggregation was performed.
Model Parameters
The SEK model simulated the transport of substances and associated water quality processes. Based on the field survey data, the assumed value of pH for calculating UIA is 7.72 and 7.61 for dry and wet seasons respectively. For temperature, the assumed value is 17.5 and 28.1 °C for dry and wet seasons respectively. Table 5 presents the state variables and process coefficients used as the base case for verification of the SEK model. Model verification was carried out by comparing the model results with the field data collected in October 2005 and January 2006 under this Study. Detailed description of each of the variables is presented in the Delft3D-WAQ manual.
Table 5 State Variables and Process Coefficients for Water Quality Modelling
| Variable | Description | Value | Unit |
|---|---|---|---|
| fSOD | Autonomous SOD | 1 | g/m 2/d |
| RcMrtEColi | First-order mortality rate E.coli | 4 | 1/d |
| SpMrtEColi | Specific mortality with respect to chloride E.coli | 0.11E-4 | m 3/g/d |
| Variable | Description | Value | Unit |
|---|---|---|---|
| VsedIM1 | Settling velocity of suspended inorganic matter fraction 1 (IM1) | 1 | m/d |
| VsedIM2 | Settling velocity of suspended inorganic matter fraction 2 (IM2) | 15 | m/d |
| VsedDetC | Settling velocity of Detritus Carbon (DetC) | 1 | m/d |
| VsedDiat | Settling velocity of Diatoms (Diat) | 1 | m/d |
| VsedGreen | Settling velocity of Non-diatoms (Greens) | 1 | m/d |
| VsedBOD5 | Settling velocity of Biological Oxygen Demand (BOD5) | 1 | m/d |
| TauCSIM1 | Critical shear sedimentation of IM1 | 0.075 | Pa |
| TauCSIM2 | Critical shear sedimentation of IM2 | 1.0 | Pa |
| TauCSdetC | Critical shear sedimentation of DetC | 0.075 | Pa |
| TauCSDiat | Critical shear sedimentation of Diat | 0.075 | Pa |
| TauCSGreen | Critical shear sedimentation of Green | 0.075 | Pa |
| TauCSBOD | Critical shear sedimentation of BOD | 0.075 | Pa |
| PpmaxGreen | Potential maximum production rate of Greens | 2.3 | 1/d |
| TcGroGreen | Temperature coefficient for growth processes green algae | 1.07 | - |
| MrespGreen | Maintenance respiration for green algae | 0.036 | 1/d |
| GrespGreen | Growth respiration factor for green algae | 0.11 | - |
| Mort0Green | Mortality rate of green algae | 0.32 | 1/d |
| MortSGreen | Mortality rate of green algae at high salinity | 2 | 1/d |
| SalM1Green | Lower salinity limit for mortality of green algae | 5 | ppt |
| SalM2Green | Upper salinity limit for mortality of green algae | 10 | ppt |
| Ppmaxdiat | Potential maximum production rate of Diat | 2.3 | 1/d |
| MrespDiat | Maintenance respiration of Diat | 0.036 | 1/d |
| GrespDiat | Growth respiration factor of Diat | 0.11 | - |
| Mort0Diat | Mortality rate of Diat | 2 | 1/d |
| MortSDiat | Mortality rate of Diat at high salinity | 0.32 | 1/d |
| SalM1diat | Lower salinity limit for mortality of Diat | 5 | ppt |
| SalM2Diat | Upper salinity limit for mortality of Diat | 10 | ppt |
| RcDetC | First-order mineralisation rate of DetC | 0.1 | 1/d |
| RcDetN | First-order mineralisation rate of Detritus Nitrogen (DetN) | 0.1 | 1/d |
| RcDetP | First-order mineralisation rate of Detritus Phosphorus (DetP) | 0.08 | 1/d |
| RcDetsi | First-order mineralisation rate of Detritus Silicon (DetSi) | 0.01 | 1/d |
| RcBOD | Decay reaction rate of CBOD | 0.1 | 1/d |
| RcNit | First-order nitrification rate | 0.1 | 1/d |
| TaucRS1DM | Critical shear resuspension of dry matter (DM) from sediment layer 1 | 0.1 | Pa |
| KLRear | Reaeration transfer coefficient | 1.5 | 1/d |
| ExtVLBak | Background extinction of visible light | 0.5 | 1/m |
| ExtVL1M1 | Visible light specific extinction coefficient for IM1 | 0.075 | m 2/g |
| ExtVLDetC | Visible light specific extinction coefficient for DetC | 0.47 | m 2/g |
| ExtVLGreen | Visible light specific extinction coefficient for Green | 0.15 | m 2/g |
| ExtVLDiat | Visible light specific extinction coefficient for Diat | 0.15 | m 2/g |
| ExtUVBak | Background extinction of uv-light | 0.5 | 1/m |
| ExtUVIM1 | UV-light specific extinction coefficient for IM1 | 0.075 | m 2/g |
| ExtUVIM2 | UV-light specific extinction coefficient for IM2 | 0.075 | m 2/g |
| ExtUVDetC | UV-light specific extinction coefficient for DetC | 0.47 | m 2/g |
| ExtUVGreen | UV-light specific extinction coefficient for Green | 0.15 | m 2/g |
| ExtUVDiat | UV-light specific extinction coefficient for Diat | 0.15 | m 2/g |
| ZresDM | Zero order resuspension flux of DM | 30 | g/m 2/d |
Initial and Boundary Conditions
A spin-up period of three full spring/neap cycles was adopted before the actual water quality model run. After performing the spin-up, the effect due to the initial conditions would subside and would not affect the final modelling results.
The modelling results from the Update model provided data to define the water quality conditions in terms of concentrations of relevant water quality parameters at the open boundaries of the SEK model.
MODEL CALIBRATION RESULTS
The regional Update model is able to provide reliable prediction of hydrodynamic and water quality conditions in HKSAR waters as demonstrated in the EPD Update Study. The predictions from the SEK model have been checked to be consistent with the Update model predictions. The hydrodynamic predictions from the SEK model were also found to be in order as compared to the field data for salinity, current velocity, dye tracing and water quality collected as part of the KTPR in October 2005 and January 2006. It was suggested from the model calibration exercise conducted under the KTPR that the SEK model can be used for predicting the hydrodynamic and water quality conditions in the modelling area including KTAC, KTTS , Kowloon Bay and the Victoria Harbour. Details of the model calibration results are given in the separate model calibration report prepared under the KTPR.
In order to provide an indication of the model performance for the readers’ reference, the 2005 model results (averaged model output for dry and wet seasons) are compared against the available historical data collected in 2005 at four EPD stations (VM1, VM2, VM4 and VM5) closest to the proposed dredging site in Annex A of this Appendix. It is considered that the key parameters of concern for the EIA study of this dredging project would be SS and DO in the open harbour and therefore only SS and DO are presented for reference. As shown in Annex A, the 2005 model predictions for DO and SS are all compared very well with the 2005 EPD data. The DO and SS levels predicted by SEK model are therefore considered appropriate to be used in this EIA as the background levels under the pre-Project scenario (refer to Section 5.6.82 of the main EIA text).
FIGURES
ANNEX A
