Nyl River Hydrological Modelling EScience Associates (Pty) Ltd Draft Final 14 January 2014
Nyl River Hydrological Modelling
EScience Associates (Pty) Ltd
Draft Final
14 January 2014
2
Nyl River Hydrological Analysis
Project No : W00/JNB/000029
Report by Allan Bailey and Karim Sami dated 13 January 2014
1. Background
Royal Haskoning DHV have been appointed by EScience Associates (Pty) Ltd to compile a brief technical report for the hydrological analysis of the Nyl River in tertiary catchment A61 in the Limpopo Water Management Area (WMA)
The system in question consists of the quaternary catchments A61A to A61E and a small portion of A61F upstream of streamflow gauge A6H033. The specific mining area is near the Grass Valley chrome mine as shown in Figure 1.
The Water Resources Simulation Model (WRSM2000) also referred to as the Pitman model was used.
The existing WR2005 system is to have more detail added and two streamflow gauges, namely : A6H011 and A6H033. The A6H033 streamflow gauge is very short at 5 years but is located just downstream of the study area which is in quaternary A61E.
Rainfall, observed streamflow and land use is to be brought up to date to September 2010.
A key issue is the groundwater-surface water interaction and Royal Haskoning included the groundwater specialist Mr Karim Sami on their team to fine-tune this part of the model and analyse this aspect in detail.
3
Figure 1: Google Earth image with study area and streamflow gauge A6H033
4
2. WRSM2000 Set-up
The existing Water Resources of South Africa, 2005 study (WR2005) was used to obtain the WRSM2000 set-up for the A61 tertiary catchment. WRSM2000 is based on a monthly time step and includes the Sami groundwater-surface water algorithm, the WQT method of irrigation and the CSIR alien vegetation method. Land use in this system appears to be limited to an abstraction for Modimolle from Donkerpooort Dam, alien vegetation, farm dams and irrigation on a small scale. The All Towns Study for the Department of Water Affairs was used to check on the abstraction for Modimolle and was found to be estimated at 0.93 million m3/a. It was assumed to have a constant monthly demand. There are quite a few tributaries flowing from the Waterberg Mountains into the Nyl River and wetland. The detailed schematic diagram for the system is given in Figures 2, 3 and 4. There are four module types used in this analysis, namely: runoff, reservoir, channel and irrigation. Components (modules) of the WRSM2000 system are as follows :
2.1 Runoff Modules
The hexagonal shapes in Figures 2, 3 and 4 denote runoff modules which determine runoff from rainfall and numerous other meteorological, groundwater parameters and calibration data. The system covers the entire A61 tertiary catchment which is split into 9 quaternary catchments (A61A to A61J). In some cases a quaternary catchment has been split into two or more sub-catchments due to the location of streamflow gauges and irrigation areas. The latest Sami groundwater default parameters have been entered into all the runoff modules. These groundwater parameters describe the groundwater – surface water interaction and have been manipulated by Karim Sami to best suit the catchment conditions.
Alien vegetation is present in all the quaternary catchments and have been included as so-called “child runoff modules”.
Monthly rainfall data (which was available from 1920) was updated to September 2010 for individual rainfall stations. Missing months were patched using the in-house program “Patchmp”. All the modules use catchment based rainfall which is a combination of rainfall stations. Monthly evaporation was used from the WR2005 study.
5
The latest WR2005 project runoff calibration parameters were used. These parameters are listed below :
POW - Determines rate at which subsurface flow (interflow plus groundwater) reduces as soil moisture is depleted
GPOW Controls potential recharge (Equivalent to POW in Pitman model)
SL Soil moisture level below which all subsurface flow ceases. Below this level the only soil moisture depletion is via evapotranspiration
HGSL Determines soil moisture level below which recharge ceases
ST Moisture holding capacity of soil
FT Maximum rate of subsurface flow at soil moisture capacity
GW Splits soil moisture into upper (faster response – see TL) and lower (slower response – see GL) zones. (If GW=0 there is only an upper zone.)
HGGW Controls potential recharge (Similar to FT/GW in Pitman model but not the same)
ZMIN Minimum rainfall intensity required to initiate surface runoff, so below this intensity all rainfall absorbed by soil
ZMAX Determines (in conjunction with ZMIN) the average infiltration to soil moisture
PI Interception storage
TL Lag of surface runoff and subsurface flow from the upper zone (see GW)
GL Lag of subsurface flow in the lower zone (see GW)
R Controls rate at which evaporation reduces as soil moisture is depleted.
Different values were assigned to the various runoff modules during the calibration process.
2.2 Reservoir Modules The triangular shapes in Figures 2,3 and 4 denote reservoirs or collections of farm dams. The largest reservoir is Doorndraai Dam which is downstream of the study area. Other smaller dams are Donkerpoort, Glen Combrink, Haaskloof and Rooiwal dams. Dams that are not named are collections of small farm dams. There are urban abstractions for Modimolle at Donkerpoort Dam and Mokopane at Gert Combrink Dam. At Doorndraai Dam there are both urban and industrial abstractions as well as for irrigation. Wetlands were originally allocated as comprehensive wetlands to channel reach modules with outflow routes 50, 65 and 140 but in analysing the monthly hydrograph at streamflow gauge 140,
6
it was noticed that there was a few months of lag. This is due to the fact that the comprehensive wetland methodology is really intended for off-channel wetlands whereas the Nylsvlei wetland and Nyl River are all one entity. This was resolved by changing the channel reach modules to dummy reservoir modules shown as RV 17, 18 and 19. The maximum depth of the wetland is unknown but it was also found that the average depth that gave the most realistic calibration was 0.75 m. The following wetlands have been set up as dummy reservoirs: Table 1: Wetlands
Dummy Reservoir Area (km2) Volume (million m3)
17 68.50 51.40
18 17.00 12.75
19 5.00 3.75
2.3 Irrigation Blocks/Modules
Areas of irrigation are shown in rectangular blocks in Figures 2, 3 and 4. Some irrigation is fed from farm dams or large dams (abstractions at reservoir modules) and some as run of river (abstractions at channel reach modules).
2.4 Channel Modules
Channel reach modules are shown as circular shapes in Figures 2, 3 and 4 and link other modules and routes together. They also deal with abstractions and return flows from/to the system.
2.5 Routes
Routes connect modules and describe the flow direction from one module to another. Streamflow gauging stations are depicted on the routes as a small bar. At these gauging stations, simulated streamflow produced by WRSM2000 can be compared using graphical plots and statistics against the measured streamflows.
In quaternary catchment A61A, there are two streamflow gauges A6H006 downstream of Donkerpoort Dam on the Little Nyl River and A6H011 on the Great Nyl River. Neither streamflow gauge is that accurate and it was difficult to achieve a good calibration at these gauges. There are two streamflow gauges on tributaries of the Nyl River in A61B and A61C. The most useful streamflow gauge only had a five year record period but is just downstream of the study area in A61E. Considerable effort was undertaken by both Dr Bill Pitman and Allan Bailey on the surface water aspects and Karim Sami on the groundwater aspects to get a good correlation between simulated and observed streamflow at this streamflow gauge. The only
7
other gauges in the system are at Doordraai Dam where the inflow is calculated from the dam balance or Reservoir Record and further downstream at A6H027.
2.6 Groundwater abstraction
Groundwater abstraction alters the water balance of runoff by:
Depleting baseflow;
Inducing transmission losses and
Reducing evapotranspiration from groundwater.
The impacts of groundwater abstraction must therefore be considered when simulating runoff from systems with significant abstraction from aquifers in hydraulic connection with surface water. Groundwater abstraction data was sourced from GRAII for the upstream catchments outside the study area, and was obtained by hydrocensus and crop water modelling in the study area. Abstraction for irrigation was estimated at 6 Mm3/a (Steyn, 2012). An additional 0.5 Mm3/a was included in catchment A61E2 to account for abstraction from the Makopane wellfield. The groundwater abstraction utilised in the model is shown in Table 2.
Table 2 : Groundwater abstraction in Mm3/a
Quaternary Area (km2) 1970 1986 2010
2 89.0 0 1.20 1.20
A61A2 79.0 0 0.60 0.60
A61A3 40.0 0 0 0
A61A5 100.0 0 0.50 0.50
A61A4 73.0 0 0 0
A61B1 211.8 0 0 0
A61B2 150.6 0 0 0
A61C1 63.4 0 0 0
A61C2 298.6 0 2.10 2.10
A61D1 141.5 0 0.08 0.08
A61D2 314.4 0 2.70 2.70
A61E1 211.5 0 0 0
A61E2 335.5 0 6.50 6.50
A61F 789.0 0 1.00 1.00
Note: Only 10% of runoff from A61F contributes discharge to A6H033
8
2
1
RU1
(A61A_1)
(89 km2)
RR2
RR6
4
RU2
(A61B1)
A6H011.OBS
CR4
CR7
CR10
CR3 CR2
CR6 CR5RU4
(A61C1)
RU6
(A61D1)CR8CR9RR14
RV4
RV6
RU3
(A61B2)
RU5
(A61C2)
RR12
RR11
RR5
RR7
RR3
RU7
(A61D2)
99
18
13
14
11 9
10 8
25
23
24 22
2127
33
28
35
20
38
40 3937
36
41
4248
43
49
50
6
A6H012.OBS
A6H010.OBS
A61A.ABS
(Modimolle)
DONKERPOORT DAM
RU14
(A61A_4)
100
CR21
98
106
109
CR22
110
111
16
RU16
(A61A_1)
Al VEG
127
RU15
(A61A_2)
Al VEG
128
RU17
(A61B1)
Al VEG129
RU19
(A61C1)
Al VEG
131
RU18
(A61B2)
Al VEG
130
RU20
(A61C2)
Al VEG
132
RU22
(A61D2)
Al VEG
134RU21
(A61D1)
Al VEG
133
CR28CR303
RU27
(A61A_2)
79 km2
5
CR31A6H006.OBS
RU28
(A61A_3)
40 km2
141
Great Nyl River
Littlet Nyl River
142
Figure 2 : WRSM2000 Network diagram for A61A to D
9
65
50
57 RU8
(A61E1)CR13
RV8
RR20
RR15
72
CR14
CR15
RR16
RR1864
79
58
63
54
56
CR11CR12
RU9
(A61E2)
RU10
(A61F)
RU11
(A61G)
RV11
RV10
RR19
RR21
RR22
RR17
RV9
51
55
53
52
59
6160
62
71
66 (36 %)
69
67 (55%)
6870
78 73
74
75
7776
CR20
103
A61D.ABS
(Mokopane)
101
111
112113
114
115
GERT COMBRINK DAM
RU23
(A61E1)
Al VEG 135
RU24
(A61E2)
Al VEG
RU25
(A61F)
Al VEG
136
137
RU26
(A61G)
Al VEG138
139 (9%)
CR29
140
A6H033.OBS
Figure 3 : WRSM2000 Network diagram for A61E to G
10
94
83
A6H027.OBS
RU12
(A61H) RV13
RR23
97
88
DOORNDRAAI
DAM
CR17
81
80
96
79
CR18
CR19
CR16
RU13
(A61J)
RV14
RV12
RR24
RR26
92
90
89
102
85
93
84
82
86
91
A6R001.OBS
RR2795
Outflow to A62B
(A61RQ97.ANS)
A6R001_IRR.ABS
105
A6R001_IND.ABS
(Industry and town)
115
118
CR26
116
117
CR20
A6R001_RELEASES.TXT
121
RV15
123
HAASKLOOF DAM
122
RV
16
ROOIWAL DAM
125
126
120
Figure 4 : WRSM2000 Network diagram for A61H and J) :
11
3.0 WRSM2000 Analysis and Calibration
Following the setting up and updating of the model, the simulation was run from 1920/21 to 2009/2010 (hydrological years). For the A61A catchment, the annual hydrographs at A6H006 and A6H011 are given in Figures 5 and 6 below.
Figure 5 : Annual hydrograph at streamflow gauge A6H006
12
Figure 6 : Annual hydrograph at streamflow gauge A6H011
For the A61B and A61C upper catchments, the annual hydrographs at A6H012 and A6H010 are given in Figures 7 and 8 below.
Figure 7 : Annual hydrograph at streamflow gauge A6H012
13
Figure 8 : Annual hydrograph at streamflow gauge A6H010
For A6H033, graphs have been shown in Figures 9 to 11 for the monthly hydrograph (seeing as the record period is so short) as well as the mean monthly flows and cumulative frequency of flows. Figure 9 shows that there is only streamflow at A6H033 when there is a flood event otherwise there is no streamflow at all.
14
MONTHLY HYDROGRAPHS
HYDROLOGICAL YEAR
MONTHLY FLOW - Mm³
2005. 2006. 2006. 2007. 2007. 2008. 2008. 2009. 2009. 2010. 2010.
0
5
10
15
20
25
30 Observed Simulated
ROUTE NO. 140 (Vlei outlet)
WRSM 20002013/12/19 (13:11) Record Period: 2005 - 2009
Figure 9 : Monthly hydrograph at streamflow gauge A6H033
MEAN MONTHLY FLOWS
MONTH (Oct - Sep)
MEAN FLOW - Mm³
0 1 2 3 4 5 6 7 8 9 10 11 12
0
1
2
3
4
5
6 Observed Simulated
ROUTE NO. 140 (Vlei outlet)
WRSM 20002013/12/19 (13:12) Record Period: 2005 - 2009
Figure 10 : Mean monthly flows at streamflow gauge A6H033
15
Figure 11 : Cumulative frequency at streamflow gauge A6H033.
The flow statistics of the simulation are shown in Figure 12. The flow generated upstream of A61E is shown in Figure 13. There is a significant reduction in flow in catchment A61E due to the effect of the wetland. This can be attributed to evapotranspiration in the wetland, which attenuates flood events. A comparison of Figures 9 and 14 illustrates that flood peaks downstream of A61E are smaller than upstream of A61E
16
Figure 12 : Flow statistics at streamflow gauge A6H033
Figure 13 : Flow statistics upstream of A61E
17
MONTHLY HYDROGRAPHS
HYDROLOGICAL YEAR
MONTHLY FLOW - Mm³
2005. 2006. 2006. 2007. 2007. 2008. 2008. 2009. 2009. 2010. 2010.
0
5
10
15
20
25
30
35
Simulated
ROUTE NO. 103
WRSM 20002013/12/19 (13:21) Record Period: 2005 - 2009
Figure 14 : Monthly hydrograph upstream of A61E
Figure 15 shows a histogram of monthly flows and shows that the frequency of zero flow, at just over 80% has been adequately simulated.
MONTHLY HISTOGRAM
MONTHLY FLOW - Mm³
FREQUENCY - % total
0 .10 .20 .30 .40 .50 .60 .70 .80
0
10
20
30
40
50
60
70
80 Observed Simulated
ROUTE NO. 140 (Vlei outlet)
WRSM 20002013/12/19 (15:06) Record Period: 2005 - 2009
Figure 15 : Histogram of monthly flows at A6H0033
18
4.0 Groundwater – surface water interface
The WRSM2000 model simulates the following surface water and groundwater interactions:
BASEFLOW
Interflow occurring from the unsaturated zone contributing to hydrograph recession following a large storm event, or discharge from perched water tables via temporary or perennial springs located above low permeability layers, which may cause prolonged baseflow following rain events, even when the regional water table is below the stream channel and
Groundwater baseflow discharged from the regional aquifer to surface water as baseflow to river channels, either to perennial effluent or intermittent streams.
RIVER LOSSES
Transmission losses of surface water when river stage is above the groundwater table in phreatic aquifers with a water table in contact with the river and
Groundwater baseflow reduction and induced recharge caused by pumping of aquifer systems in the vicinity of rivers causing a flow reversal.
EVAPOTRANSPIRATION
Depletion of aquifer storage by evapotranspiration from riverine areas and areas of shallow water table as a function of aquifer storage.
The processes simulated and a description of the algorithms are given in (Sami, 2013).
Simulation of interactions is relevant under conditions where groundwater abstraction takes place. The decline of water levels around pumping boreholes near surface water bodies creates gradients that capture some of the ambient groundwater that would have discharged as groundwater baseflow. At sufficiently high pumping rates this water level decline also induces flow out of the surface water body, a process known as induced recharge, which results in transmission losses from the channel. Both these processes lead to streamflow depletion, which can significantly impact the ecology and yield of dams.
4.1 Groundwater Balance To investigate the long term impact of irrigation and mining, the model was run for A61E2 with constant abstraction from 1920-2010 to compare the ground water balance under virgin conditions and under irrigation scenarios. The objective was to provide water balance figures
19
against which the MODFLOW model could be calibrated in terms of river leakage and baseflow, and evapotranspiration.
Abstractions of 6.5 and 8.5 Mm3/a were considered in A61E2, which is the catchment affected by the mine. The results are shown in Table 3. The impact of irrigation is to decrease the MAR from 16.06 mm to of the order of 12 to 13 mm. This is caused by an increase in transmission losses from 1.23 mm to 4.3 mm. In addition, evapotranspiration from groundwater is reduced from 20.5 mm to 1 mm, and groundwater outflow through the aquifer out of the catchment is reduced from 0.23 and inflow of over 4 mm, implying groundwater is drawn in from the adjacent catchment(s).
The interflow represents baseflow generated from the high lying areas, which comes off the mountains. With abstraction, flow is zero over 80% of the time (Figure 11), hence this interflow is lost as transmission losses in permeable alluvial fans as it comes off the mountains.
Table 3 : Water balance of A61E2 under virgin and modified conditions
Virgin Abs = 6.5 Mm3/a Abs = 8.5 Mm3/a
MAP mm 588.12 588.12 588.12 MAR mm 16.06 13.35 12.02
Abstraction (Mm3/a) 0.00 6.50 8.50 Abstraction (mm/a) 0.00 19.37 25.34
Aquifer Recharge mm/a 19.64 19.76 19.76
Baseflow mm/a 2.36 1.98 1.98
G'water B'flow mm/a 0.16 0.00 0.00
Interflow mm/a 2.21 2.21 2.21
Transmission losses (mm) 1.23 2.97 4.30
G'Water Evap mm/a 20.50 4.12 0.99
G'water outflow mm/a 0.23 -1.28 -4.27
Aquifer storage change mm -1.46 -0.87 -0.68
4.2 Groundwater balance during mining
It has been estimated that irrigation will be reduced by 0.75 Mm3/a of irrigation due to suspension crop irrigation to enable the establishment of the North Pit (du Toit, 2014 and Fischer, 2014). In addition, mine pit inflows will increase to 1.70 Mm3/a. This results in a net increase in abstraction of groundwater of 1.0 Mm3/a with irrigation of excess water and without tailings seepage recovery. If tailings seepage is recovered no irrigation is undertaken and excess production is recharged to the Grassvally mine then demand has been estimated be 0.55 Mm3/a.. The water balance is shown in Table 4. The increased removal of groundwater results in a 2.47% decrease in runoff, due to an 11% increase in transmission losses. This can be attributed to a lowering of the saturated area, which allows more runoff emanating from the high lying areas in to infiltrate into the ground. Abstraction at 7.05 Mm3/a
20
(21.01 mm/a) exceeds the aquifer recharge of 19.76 mm/a, hence the aquifer becomes increasingly reliant on transmission losses from interflow and storm events.
Table 4 : Water balance pre mining and under the mining scenario
Table 5 provides mean annual run-off reductions relating to various water abstraction levels.
Table 5 : MAR for various
Additional take A61E Mm3/a
Additional take A61E Total Abstraction
Mm3/a
MAR at gauge A6H033 (mm)
2 8.5 12.02
1 7.5 12.45
0.75 7.25 12.69
0.5 7 12.92
0 6.5 13.35
Q= 7.05 Mm
3/a
With Mining Q= 6.5 Mm
3/a
Current status % Decrease
Potential recharge (mm) 21.96 MAP (mm)
588.12 588.12
MAR (mm)
13.02 13.35 2.47
Abstraction (Mm3/a)
7.05 6.50 -8.46
Abstraction (mm/a)
21.01 19.37 -8.46
Aquifer Recharge (mm/a) 19.76 19.76 0.00
Baseflow (mm/a)
1.98 1.98 0.02
Groundwater Baseflow (mm/a)
0.00 0.00 20.34
Interflow (mm/a)
2.21 2.21 0.00
Transmission losses (mm) 3.30 2.97 -11.09
Groundwater Evaporation ( mm/a)
3.09 4.12 25.06
Groundwater outflow (mm/a) -1.84 -1.28 -42.97
Aquifer storage (mm)
-0.81 -0.87 6.42
21
5.0 Conclusions
Apart from two observed streamflow gauging stations in quaternary catchment A61A, there is only one other gauge on the Nyl River of relevance to the mine situation which is A6H033 just downstream of the wetland. There are two other gauges on tributaries leading to the wetland and gauging at Doorndraai Dam further downstream. Unfortunately the record period at A6H033 is only of the order of 5 years. The observed gauges A6H006. A6H011, A6H012 and A6H010 are not of a very high accuracy. Despite all the abovementioned facts, a reasonable calibration of streamflow was achieved at the relevant gauges which should provide for a reasonably good analysis of surface and groundwater at the site of the mine. Under the existing situation with irrigation, groundwater is recharged by surface water via transmission losses due to the lowering of water levels. Evapotranspiration from groundwater is also significantly reduced. The increase in transmission losses increases the duration of zero flow in the river. Mining will result in a net increase of 11% in transmission losses and a 2.5% decrease in MAR from that sub-area. It is recommended that mine water abstraction scenarios be further analysed through combined further modelling utilising the combined WARMS and MODFLOW models set up to further investigate the effects of tailings seepage and aquifer recharge at Grassvally mine to inform water management decisions and license applications.
22
6.0 References Du Toit, G. (2014): “Volpsruit Geohydrological Report” Fischer, F. of EScience Associates (2014): “Volspruit Mine EIA” Sami, K. (2013): “Water Resources 2012, Sami Groundwater Module, Verification Studies, Default Parameters and Calibration Guide” Steyn, J. M. (2012) : “Estimation of the Irrigation Water Requirements of Crops Produced on the farm Volspruiit”