SOUTH WEST YARRAGADEE NARATIVE

SOUTH WEST YARRAGADEE CONCEPTUAL AND GROUNDWATER MODELS CRITIQUE

 PREAMBLE

The Water Corporation was required to demonstrate that impacts resulting from the pumping of 45 GL from the SWY will be acceptable to the regulatory authorities. Baddock (2005) details the hydrogeological investigations undertaken as part of those commitments, and presents the current understanding of aquifer systems gained through evaluation of the investigation and pre-existing data. The knowledge from this hydrogeological evaluation allowed the detailed conceptualisation of groundwater flow within the sedimentary basin and its representation in a regional numerical groundwater model incorporating the Blackwood, Bunbury and Busselton-Capel groundwater areas (Baddock 2005).

On the other hand  SWAM v.2 model Sun (2005) was designed to:

  • Simulate groundwater flow within and between all hydrogeological units in the Southern Perth Basin that are within the active part of the groundwater flow system.
  • Establish a water budget for the Yarragadee aquifer and other aquifers.

 There is no doubt that too much effort and money was spent in the study of the many aspects of the South West Yarragadee (SWY) and many reports were published in the Water Corporation Site including studies on groundwater dependent ecosystems, the economics of water demand, and a program of public consultation were undertaken, and are reported separately. This critique will concentrate on two reports only; Baddock (2005) report and its appendices as well as Sun (2005).

 Although Baddock (2005) report incorporated several appendices including detailed hydro-chemistry (Rockwater 2004 & 2005) and isotopes (Leaney, 2004 & 2005), however there was no comprehensive synthesis of these appendices within the report. It seems that the author opted to disregard some of the results which might cause conflict with the regulatory requirements and this made his report disjointed and incomplete.

On the other hand, the numerical model as described by Sun (2005) with a blocked evapotranspiration package and concocted recharge might as well be modelling somewhere else and not an aquifer system in the South West, and if we consider the confessed critical flaws in the model, then we could be looking at an aquifer anywhere even in Atlantis.

Although the authors were aware of the shortcomings of their work and enumerated the reports with the problems they faced, they were gallant enough to claim that their models are good representation of the reality.

The critique presented here is solely based on their findings with some minor interpretation:

 

DOWNWARD LEAKAGE MYTH

The major difficulty with the hydro-geological conceptual model as presented in both Baddock (2005) and Sun (2005) is the assumption that the bulk of the recharge is through downward leakage from the Superficial and Leederville aquifers to the top of the Yarragadee aquifer system. This myth of downward leakage is exposed by:

  • The very detailed geological cross sections prepared by Rockwater (2005) that show extensive water tight layers of clays and mudstones above the Yarragadee system.
  • An aquitard of dominantly clay sediments, named the Mowen aquitard by Baddock, is present over most of the Blackwood Plateau, and comprises the Quindalup and Mowen members of the Leederville (Baddock 2005).
  • Very low hydraulic conductivity applied to the top layer in the numerical model (Sun 2005) which makes downward infiltration virtually impossible.
  • The produced water level maps for the Yarragadee system indicate that the aquifer mound is in the outcrop area with flow southward and northward towards the coast (Rockwater 2004, 2005) under the aquitard layer.
  • The Yarragadee aquifer is the largest confined aquifer within the basin present in the Bunbury Trough, reaching up to about 1700 m thick (Baddock 2005)
  • The aquifer in the coastal areas is confined, as is under all the confining layers of the Parmelia and Leederville formations. Only in very few places there is direct hydraulic connectivity with the Superficial and Leederville formations (which receives the majority of the recharge).
  • In all these areas where the aquifer is confined there is an upward head or a slight downward head.
  • The isotope and C14 data indicate that the water entered the aquifer in eons past and there is no additional water entering the aquifer after the outcrop areas.
  • The very distinct types of water in the Yarragadee aquifer which are different from the Leederville and Superficial aquifers (Rockwater 2004).

 

Based on all this, it can be categorically concluded that there is no downward leakage from the Superficial and Leederville aquifers to the Yarragadee and that the only recharge to the aquifer is in the outcrop area. It is very interesting that the authors used C14 ages to calculate hydraulic conductivity; this technique assumes that there is no additional water entering the system between the point of recharge and point of measurement. Yet the authors assumed that the bulk of recharge to the Yarragadee aquifer is from downward leakage.

 

The panel appointed by the Water Corporation which reviewed the conceptual and SWAM2 models considered that the probability of substantial leakage to the producing aquifers from adjacent clayey formations, the contribution from rejected recharge in high watertable regions of the Swan and Scott Coastal Plains, and the increasing recharge from recycled irrigation waters on the Swan and Scott Coastal Plains all serve to produce a degree of conservatism in the model output.

It appears that the panel overlooked the fact that the Yarragadee aquifers in the coastal areas are confined, and the water is under upward pressure. How are they expecting that the water from the superficial aquifers to leak downward to the Yarragadee?

 

RECHARGE QUANDARY

Recharge rates were estimated using four different models and techniques including WAVES and WEC-C recharge models, and WATBAL a water-balance model and chloride method, net recharge was found to range from 282 to 340 GL/a.

The various methods agree that approximately 80% of the water balance from rainfall is lost through evapotranspiration, while the remaining 20% is potential recharge. Potential recharge is that portion of rainfall remaining after evapotranspiration, and is available for groundwater recharge or is rejected as stream-flow. With stream-flow 10% to 15% of rainfall, it implies net groundwater recharge for the region is 265 GL/a to 530 GL/a, and probably close to 400 GL in an average year.

Review: There is no doubt that recharge is taking place to the Superficial and the different permeable layers of the Leederville Formation, but due to the presence of thick layers of clay and mudstone as described in nearly all the cored wells, it is expected that most of what is assumed to be recharge is deep infiltration which will be diverted to surface runoff, stream flow, replenishing wetlands and discharged to the Oceans and the balance lost by evaporation and evapotranspiration.

The authors claim that the differences between the recharge rates estimated by the different methods can cover all the current and future water requirements by the different users, but are these values of any importance? Definitely not, the initial recharge which varies from one year to another is of no importance whatsoever in the estimation of the possible sustainable yield, any additional amount of water to be used must be balanced by the CAPTURE of additional recharge to cover this amount; otherwise the system will be disturbed as all initial recharge is already in equilibrium with the environment.

 

 

PROBLEMS WITH THE MODELS USED TO ESTIMATE RECHARGE:

The following critique of the models used to estimate recharge are from Baddock (2005) and Sun (2005), there is no need for any further comments on the futility of using these models to estimate recharge, at the same time their critique highlighted the other deficiencies in several other areas i.e. rainfall, landuse, LAI, soil characteristics etc..etc..:

 

LUCICAT model

  • The initial application of the LUCICAT model suggests that the quality of the rainfall data may be inaccurate.
  • A quality check needs to be performed on the records used to generate the time series.
  • Land use history, particularly logging, burning, die back and replanting pines in some areas of the catchment has not been incorporated into the modelling.
  • At present the model uses constant LAI for the whole period of study. The ‘land use history’ file has to be redeveloped.
  • A site visit to the catchment will give a better understanding of the channel characteristics, in particular with respect to the channel width, depth and roughness coefficient.
  • Soil characteristic changes with depth need to be incorporated into the model.
  • The calibration of the model needs to be refined once the modifications are made to the model.
  • The incorporation of the above mentioned modifications to the model will yield a better representation of what is occurring out in the field.

WAVES

  • Assumption that the depth to water table for each cell is fixed and did not account for a site gaining or losing water laterally by groundwater flow.
  • Where the watertable was within the rooting depth of vegetation, there was potential for more water to be used by plants than was available from rainfall. In this case the model could generate ‘negative recharge’.
  • The model could generate run-off if the soil became saturated to the surface, but without lateral groundwater inflow this did not happen in the model due to the relatively high permeability for the surface soils.
  • The recharge determined by the WAVES is approximately the excess of water after evapotranspiration and is available for runoff and recharge into the aquifer.
  • In the area with low permeability, e.g, Mowen Member, most of the excessive water will contribute to runoff whilst in the area where the underlying aquifer is more permeable, most of the excessive water will become diffused recharge to the aquifer.

Both WAVES and LUCICAT estimates of recharge did not take account of the inability of the aquifers to accept recharge in their discharge zones

CHLORIDE BALANCE METHOD:

The chloride balance method is a very simplistic method which cannot be used in large sedimentary basins where many geochemical interactions takes place. It is of no use whatsoever in semi-confined and confined aquifers or in the presence of thick clay layers in the unsaturated zone. 

PROBLEMS OF APPLICATION OF RECHARGE IN THE MODEL:

There are many problems related to the application of the recharge in the model:

  • The major error is that the seasonal distribution of recharge was varied in SWAMS 2.0 to better match observed water-level fluctuations. This means that the applied recharge according to Sun (2005) critique is not factual.
  • Recharge was applied to the highest active layer. There have been some problems where there are inactive cells: the underlying active cell is confined, and so recharge applied to a cell where a low storage coefficient applies caused very high hydraulic heads.
  • Net recharge is based on the average annual gross recharge during the modelling period for each aquifer zone receiving recharge, minus any EVT. EVT is only active on the coastal plains. Therefore, for all other areas of the model, recharge is net to the aquifer or zone.

 

AQUIFER CONNECTIVITY OR AQUIFER DISCONNECTION

Although millions of dollars were spent to define the aquifers extensions, connectivity, properties etc, only one aquifer test was conducted. The results of the single aquifer test showed significant drawdown observed in monitoring bores higher in the Yarragadee aquifer as well as within the lower third of the Leederville aquifer; this was interpreted as due to appreciable downward leakage through the aquifer and that the vertical permeability must be high. The results of this test was ignored in the conceptual design of the models used for the environmental studies by several consultants as well as by SWAM2 where it was assumed that the aquitard layers have a masked effect on the drawdown caused by abstracting from the Yarragadee aquifer system.

 

WATER BALANCE INCONSISTENCY

There are many estimates of the annual water balance including:

  • Average gross recharge to the aquifer system during 1990-2003 was 651.5 GL/year, with a net recharge of 373.6 GL/year. Average abstraction during the calibration period was modelled at 45.4 GL/year starting from 35 GL in 1990 to 62 GL in 2003 (Sun 2005).
  • In another estimate Sun suggested that net recharge vary from 282 to 340 GL/a. With stream-flow 10% to 15% of rainfall, it implies net groundwater recharge for the region is 265 GL/a to 530 GL/a, and probably close to 400 GL in an average year.
  • The water balance for the model indicates that net recharge to the Yarragadee aquifer (excluding discharge to rivers) comprises 120 GL/annum leakage from the overlying Leederville aquifer and 44 GL/annum from direct recharge in areas of outcrop.
  • Baddock (2005) estimates of net recharge in the outcrop ranged from 24.8 to 33.6 GL/a.
  • No return of groundwater to the superficial aquifer from irrigation has been accounted for.
  • Similarly, the importation of water from the Leederville and Yarragadee aquifers into the superficial aquifer, via irrigation is also ignored.
  • The reasons for not simulating irrigation return water are that the irrigation area is relative small and the effort to simulate the distribution of deeper aquifer water to shallow aquifer water is substantial.

 

 CONSTRUCTION, CALIRATION AND APPLICATION OF THE SOUTH WEST YARRAGADEE AQUIFER MODEL V2.0 (SUN, 2005)

 MODFLOW LIMITATIONS

Sun (2005) used MODFLOW for the construction of his groundwater model, although he was aware of all its limitations for modelling multilayered interactive aquifers with very active surface processes of infiltration, evpaotranspiration and surface runoff generation as is the case in the SWY. The other significant problem he faced is that in areas where there is a perched watertable, or where the water table is deep, the model was not representative (Sun 2005). Therefore, in these areas, the modelling results are indicative only and may need to be interpreted using alternative information or a source of data that is not sufficiently represented in the regional model, in other words to delve into geo-fantasy to cover the deficiency.

 THE UNSTABLE STEADY STATE

Sun (2005) failed to construct, simulate and validate the steady state solution of his model. He stated that: “It was found that due to the complexity of the model, the high variability of the topography on the Blackwood Plateau and the large dynamic range of horizontal and vertical hydraulic conductivity, that the resulting steady state solution tended to be either unstable due to the dry cells in the top three layers on the Blackwood Plateau, or unrealistic due to low Kv in controlling layer, restricting water leakage downward. To remove the dry cells, the controlling aquifer property for the dry cells on the Plateau (which is normally the vertical hydraulic conductivity Kv in layer 2), needed to be sufficiently low compared to recharge. However, low Kv restricted infiltration of water to the lower layers causing head build up in the upper layer and head dropping to below that observed in the lower layers.

In addition to the numerical problem of drying elements, problems with rainfall recharge and perching of elements were also encountered in the steady state model.

However, by forcing perched elements to be inactive two artefacts are introduced to the model:

  • recharge is forced into a lower saturated layer, requiring the recalibration of recharge

based on net recharge to the lower layer; and

  • shallow calibration bores may be eliminated from the model, as some measure

perched water levels.

The allowance of perching results in recharge only to the highest saturated layer could

mean that no recharge is possible through the unsaturated zone”.

Review: It is usual for modellers to face problems and try to solve those using different techniques, but it is not acceptable that after such detailed description of failure to achieve steady state solution to continue with this model. In this case the modeller should have abandoned his trial and thought for alternative conceptual model and/or computer model to simulate his problem.

 

BLOCKAGE OF EVAPOTRANSPIRATION PACKAGE IN THE BLACKWOOD PLATEAU

On the Swan and part of the Scott coastal plains recharge is specified as gross recharge, and the EVT package of MODFLOW is used to account for water loss due to evapotranspiration. In both these areas the Yarragadee aquifer is confined and there is upward pressure, there is no recharge to the Yarragadee aquifer.

For the Blackwood Plateau and part of the Scott Coastal Plain, EVT is not modelled.  Sun (2005) faced potential issues associated with EVT set up on the Blackwood Plateau, such as the stability of the model due to dry cells and the high variability in topography of the Blackwood Plateau. Although EVT was tested on the Blackwood Plateau, it was decided not to apply it and keep the original simple, but conservative, approach (Sun 2005).

Review: The most important parameter in the water balance of the Blackwood Plateau is evpotranspiration, where more than 80% of rainfall is lost. The water table is less than 5 in most areas, the gigantic trees of the South West have roots more than 20 m deep, it is expected that trees will use the stored water above the water table most of the dry season and will use groundwater in long drier periods. Although evapotranspiration is dominant, and the water table is less than 5 m the evapotranspiration package of MODFLOW is blocked, and some concocted values of recharge are used.

 CRITICAL ERRORS OF BOUNDARY CONDITIONS

Several critical errors were committed in the design of the boundary conditions:

  • DRAIN package was used for all the rivers instead of MODFLOW specially designed RIVER package.
  • The ocean boundary is modelled as a constant head coincident with the shoreline in layer 1, with a head of 0.0 m AHD. Aquifers below the superficial aquifer are presumed to extend offshore and discharge upwards into the ocean.
  • Blocking recharge to the upper-most active layer due to unrealistic heads in one hand and the problem of dry cells which was switched off.
  • Blocking evpotranpiration in the Blackwood Plateau.

Due to all these discrepancies in the initial and boundary conditions, the following happened:

  1. Most of the hydrographs show a significant disparity between measured and calculated water levels, particularly for the Superficial and Leederville aquifers.
  2. Substantial errors for the total flows especially to the Blackwood River between Nannup and Hutt Pool, and for the Donnelly River.
  3. The boundary conditions which are not conducive to discharge to the sea tend to elevate heads in the coastal zones. In turn, this impacted on the aquifer further upstream in all layers. This is one of the significant errors in this model as all simulated water levels are much higher than what it should be due to the damming effect.

 

 CREATIVE TERMINOLOGY TO INCREASE CONFIDENCE IN MODEL RESULTS

The modelers used novel terminology to add an element of security to their predictions of potential and current recharge and improve trust in the model results:

Masked effect:

Although the single aquifer test results showed that the differential pressures created by pumping will cause appreciable drawdown in all the above formations (which will affect groundwater discharge to wetland, rivers and biota), the hydraulic properties were manipulated to create lower drawdown in the upper formations and they termed that a masked effect.

 Rejected recharge:

All the surface water components of the water cycle; surface runoff, stream flow, bank storages, wetlands were seasonal springs were renamed rejected recharge. All this to prove that there is plenty of water available for additional recharge.

No one is disputing the fact that there are thousands of Giga litres of water; the problem is how, when and where it can be CAPTURED to recharge the aquifer that is if it is not already in equilibrium with the environment.

Also, there are many alternative paths for the water before it can become recharge if it is not lost by evapotranspiration, depression storages, perched storages, overland flow and stream flow, only what remains after that will become deep infiltration and part of it will become recharge. The main controls on all these activities and recharge is the hydraulic conductivity of the top and subsurface soils, the geological formation which will receive the recharge and the interconnectivity between the formations.

 

FUTURE WORK AND RECOMMENDATIONS

Too much work and money have already been spent on trying to estimate the recharge using different techniques and methods. We have been saying all the time, that the initial or current recharge is of no importance whatsoever in the future planning of the use of any aquifer in the South West or elsewhere. What is important is to find out how much additional recharge you need to CAPTURE, where, when and how you will do that. But the authors of the two reports are recommending further studies for the estimation of recharge. This will be waste of effort and money.

Their recommendations also include better LAI determination, which is rather academic if they are still intending to block the evapotranspiration package in the Blackwood Plateau. As for the additional core samples for more chloride balance analysis, this is of no use in these regional aquifers and reflect a very local attitude.

 The authors suggested that further full scale calibration of the SWAMS model is not required unless significant changes are made to the mapping of the hydrogeological units, primarily in the Leederville aquifer on the Blackwood Plateau.

In our opinion, the model is not representative of the processes which are taking place in the South West. Also, to continue using MODFLOW, the same problems of dry cells, unsteady steady state solution, failure to balance the water balance, the evapotranspiration package, the use of drains instead of rivers will still haunt the author. It is important to try to find another suitable model which can be used to model the different processes in the South West and a better interpretation of the hydro-geological processes and how to apply it in the construction of the model.

 In conclusion, it is essential to overlook the fairy tale of the 1,000,000 GL and the myth of downward leakage. The results presented in the multitude of reports in the Water Corporation Site indicate that the Superficial and Leederville aquifers are better sources of water than the Yarragadee.

It is also essential for the use of any aquifer that the CAPTURE philosophy be adopted for the guarantee of long term sustainable utilisation of the resource.