The Bulli Seam EA - Illawarra Coal - PAC Inquiry COMMENTS BY DR ANN YOUNG, FEBRUARY 2010

1. The reliability of predictions and risk assessment methods In Appendix A Subsidence Assessment, (p 11), it is stated that non-systematic parameters, due to faults and other geological discontinuities, can be higher than predictable systematic ones. Also the transient parameters need to be considered as well as the incremental and systematic ones used in the Incremental Profile Method. This is made explicit in Figures 4.1 and 4.2 (p 24, 25). The observed transient tilts and curvatures (and hence strains) above Coal Cliff panels 221-226 are appreciable - 60% of maximum tilt and curvature of up to an absolute value of 0.1 when zero was predicted for the after-mining period. Predicted maxima may be exceeded or may occur away from the predicted location. Above Tahmoor panels 201 and 202, the position of the maximum tilt was 50-70 m away from the location predicted, and there were unpredicted spikes of upsidence both above and outside the goaf. The smoothed predicted subsidence profiles do not really predict the impact of subsidence on the ground surface.

Clearly the prediction techniques used - even if generally conservative - could seriously underestimate the tilts, strains and subsidence that actually affect the landscape features, and this is crucial for significant features such as cliff lines, stream beds and swamps.

BULLI SEAM PAC Feb 2010 Comments Ann Young 2 On p22 (Section 4.4) the conclusion is that systematic subsidence parameters, obtained by the Incremental Profile Method, ‘reasonably match the observed subsidence parameters’ and that predictions for wider panels can be done ‘with reasonable accuracy’. While I fully understand that the science is not exact and some qualification is valid, the issue is what is ‘reasonable? If the parameters lie well outside the predictions even in localised places, the prediction is not ‘reasonably accurate’. Similarly, on p 36, it says that ‘allowance should be made for the possibility of observed movements exceeding those predicted as the result of anomalous or non-systematic movements, or for greater subsidence, to occur in some places, such as was observed at Metropolitan and Tahmoor Collieries’. The question for the Department of Planning is ‘what is meant by allowance being made?’ Does it mean that any variations from prediction are allowed without penalty, being seen simply as unavoidable variation due to the complex nature of the environment? Or does it mean that everyone recognises that the predictions are not 100% accurate but that significant variations from them will bring immediate and meaningful penalty response from the relevant authorities? This raises the point I have argued in connection with other aspects of the impacts of longwall mining. The strong tendency of risk management and response in the industry and government is to say ‘if the parameters measured are on average within prediction, then all is well and the conditions of approval have been met’. So if the measured subsidence parameters are similar to those predicted, even if some lie outside the range, then no action is taken. Yet we know from experience that there are impacts that have not been predicted and these are put down simply to localised strains, tilts etc. There simply is NO stringent ongoing reappraisal of the impacts and assessment of whether the mining techniques should be modified. As this project is planned to continue over 30 years, this situation must change - the information that is gathered during current mining operations must be far more purposefully incorporated into the oversight of future mining practice. In my view, this is primarily the responsibility of government. My overwhelming impression is that the reporting of data on impacts is met with, at best, a ‘tick the box’ response and with no serious scientific assessment or evaluation. It is encouraging to see the development of subsidence prediction as detailed in Appendix A, such as the recognition of the importance of the shale component in Metropolitan leases vis-a-vis Appin etc (p 36). Also the research that shows the reduction in upsidence and closure when mining stops before a major stream is undermined both supports Illawarra Coal’s action in avoiding undermining major streams and also shows the environmental value of taking decisive action and changing mining patterns even before the full weight of evidence is available. I return to my November comments on swamps - we cannot keep procrastinating and waiting for definitive research when the indication of long term severe damage to the swamps is as compelling as the evidence was for the impacts of mining under rivers.

2. Impacts on stream beds I draw attention to the comment on Appendix A p. 52 that ‘the level of impact to a stream .. can be managed by limiting the amount of extraction and caving that occurs’. This limiting is exactly what conservationists are seeking and the reason is clear in the sorry list of major stream impacts and pools drying up on p 54. While the subsidence history of those streams has helped to plan to avoid some major impacts, part of the problem lies in the reliance on those subsidence calculations. There is almost no attention in the EA to the well-documented and reasonably well-understood geomorphological processes of stream and valley erosion in sandstone landscapes. This means that the impacts of subsidence are not properly understood.

BULLI SEAM PAC Feb 2010 Comments Ann Young 3 A good example is the discussion about cracking of rock bars. It is depressing to see diagrams such as Figs 5.2 and 5.3 (Appendix A, p. 58) in a professional assessment. Making every allowance for a desire to make the diagrams easy to understand, they are overly simplified and give little understanding of processes. Worse still, they reassert the simple linear model of flow through bedrock below stream beds when we know that mining-induced cracking makes available a greatly expanded 3-D network of fractures in the bedrock. Nor does this linear model take into account the changes of input from valley sides. Nor does it recognise the complex interaction between impermeable sandstone beds and the connectivity of joints and bedding planes.

The overly simplified picture is further reinforced by the ‘formula’ on p 59: upstream flow = diversionary flow + surface flow in the affected section = downstream flow This ‘formula’ implies that all water diverted in an affected area necessarily returns (‘reports’ in correct military manner) downstream and so the only real impact is loss of surface flow for some distance. This is an assertion! It is not an established fact! And even if the total flow returned, the alteration to water quality from passage through the bedrock (eg added iron content) means that the environmental impacts may be long-lasting and significant. The counter argument to permanent loss of flow is to invoke the idea that ‘Provided that surface water does not flow into the mine, water merely fills the voids created by mining-induced fractures ‘ (p 60). While this may be so under stored waters (for which the comment was made), I am less convinced that it is a valid assumption for stream beds or under swamps (about which it also often has been asserted). Consider the following likely results of cracking:

• The dispersal of flow into a network of cracks allows the possibility of emergence at multiple sites and increased loss to evapotranspiration.

• It means that the impact of dry spells is accentuated because there is less surface water.

BULLI SEAM PAC Feb 2010 Comments Ann Young 4 • It changes the rate of flow from extremely slow movement through subsurface

pathways that are tortuous, narrow and often partly blocked by sediment or debris, to more rapid flow through multiple open fractures.

• Near catchment heads, it may even allow water to move to another sub-catchment from the ridges, given that the depth of induced fractures can be 10-15 m. This is especially likely if the water is diverted into strata which dip away from the valley axis.

• It implies that once the fractures have been filled by a good rainfall and flow, then the problem ceases - but in fact they may well dry out again and in the future there will be multiple losses into the network.

In short, even if the water ends up further downstream (if evapotranspiration losses are not significant), there may be significant impacts on the quality of the water, the flashiness of flood flow and the percentage of base flow - let alone the ecological consequences for the affected reaches. It is asserted that even if water is diverted into dilated strata, there is unlikely to be net loss of water from the catchment by flow into deeper strata or the mine, and that there are data from ‘detailed geological database..and studies of water flows in major streams’ that support this assertion (p 79). I suggest that this detailed empirical evidence is very limited. I still await a full explanation of why major inflows to other mines have occurred at times of high rainfall. I accept that there is not a direct link from surface to mine, but interconnection of naturally separate aquifers and/or altered pressure relationships seem very likely. In the project area, there is a dearth of information about the aquifers in the Hawkesbury Sandstone. This was evident in the technical papers by Sydney Catchment Authority consultants when they were researching the extraction from the Kangaloon aquifer. Yet the dilation of the strata will occur almost entirely in this poorly understood formation. As a passing comment, it is surprising that the major work on the Hawkesbury Sandstone’s geology, hydrology, landforms and geotechnical characteristics has not been consulted during the preparation of the EA. (G.H. McNALLY & B.J. FRANKLIN (eds.) Sandstone City , Geological Society of Australia, Sydney, 2000) And it is outrageous to claim that ‘remediation of fractured rock bars has been successfully undertaken within the Southern Coalfield’ (p79)! The industry may consider the improvement in pool levels after massive volumes of material were injected into a rock bar at Waratah Rivulet to have been worthy of an award, but the result was trivial in comparison to the extent of damage caused. Nor are cracks in the rocks bars the only issue. The widespread fracturing of the stream bed, the opening of joints both in the bed and the valley sides, the cracking of previously intact blocks of sandstone - all these impacts remain even if a few holes are plugged in a few rock bars. The impacts of subsidence on streams are much more than a few cracks in a few rock bars that can perhaps be plugged afterwards. The impacts affect not only the rock bars but also the stream beds and the valley sides. And it is nonsense to write as if the processes are well understood and as if remediation techniques were proven and available.

3. Rock bars The diagrams in the EA document give a totally inaccurate picture of the rock bars found in the sandstone valleys of the project area. The ‘rock bar’ in the diagram looks like a heap of different material, sticking up on a sloping stream bed, and with a few pathways for water sloping through it. In fact, the rock bars are quite different. Usually they sit on very gently sloping reaches, above abrupt knickpoints (small waterfalls). The pools behind them are not impounded behind a barrier so much as lying in excavated sections behind the rock bar. The

BULLI SEAM PAC Feb 2010 Comments Ann Young 5 pathways through the rock bars are not sloping but vertical via potholes and then horizontal via bedding planes. In fact similar pathways exist upstream of the rock bars, and at least part of the excavation of pools occurs not by joint block plucking during floods but by solution of the underlying sandstone. The potholes in the base of pools indicates this, and potholes are both most numerous and deepest close to the knickpoints. At times, potholes may be deeper than the fall to the next pool downstream, showing that solution occurs in the subsurface along the valley floor. The subsurface flow is concentrated down existing joints and along existing bedding planes. This is obvious from the emergence of subsurface flow (often marked by red-orange iron oxide tufas) at discontinuities on the face of knickpoints. BUT this solution takes place very slowly, on a geological time scale. The iron oxide skin on a pothole in the bed of Lizard Creek was dated at 33,000 years old (U/Th dating*), showing clearly that the stream bedrock channels form extremely slowly. The disruption caused by subsidence cracking is a massive impact on the landscape, far exceeding any natural process in speed and severity. (*R.W. YOUNG, S.A. SHORT, D.M. PRICE, E.A. BRYANT, G.C. NANSON, B.H. GARDINER & R.A.L. WRAY 1994 “Ferruginous weathering under cool temperate climates during the Late Pleistocene in southeastern Australia.” Zeitschrift für Geomorphologie 38, 45-57.) What happens when the stream bed is fractured is that multiple additional pathways are created by splitting of the joint blocks, joints and bedding planes are widened so flow through them may be faster, cracks extend well away from the valley axis so sideways leakage occurs, new sites for seepage to emerge may be created. The sketch below illustrates these points:

BULLI SEAM PAC Feb 2010 Comments Ann Young 6 4. Rock bars and swamps The implication of the discussion of rock bars in swamps (p O-14 of Appendix O Upland Swamp Risk Assessment.) is that rock bars are responsible for swamp formation, at least in most cases. This error persists because of the view that rock bars sit up on the stream bed and pond water behind them. In fact, at the downstream exit of most swamps, the sediments extend right to, and sometimes drape over, the top of a knickpoint. This knickpoint may or may not have a pool upstream of it. The swamps do NOT occur because rock bars block downstream movement of sediment and water! Rather, the gentle slope and low discharge of headwater creeks mean coarse sandy sediment is not flushed into the steeper valleys below, and the sediments retain water seeping into them, especially once the wet conditions lead to accumulation of organic matter. I certainly think that cracking of the knickpoints will cause damage to the swamps, but the cracking at the exit is only part of the problem when subsidence fractures the bedrock surface underlying the whole swamp. The emphasis on rock bars means that this fact is forgotten, and the myth of easy remediation is enhanced.

5. The impacts on swamps I do not wish to repeat material I have dealt with previously, but several issues need attention. On p 12 Appendix O, several statements are made about the hydrology of the swamps. I do not know how well the Australian Water Balance Model estimates the relative inputs of runoff from rainfall v. baseflow, nor surface flow v. groundwater contribution. However, while any data is helpful, it is hard to place much reliance on data form only 4 years in the headwater creeks and 10 years well downstream. The conclusion that the hydrological input from swamps is low relative to non-swamp areas is at odds with the generally held view, I think. And the statement that swamp vegetation has a higher capacity to transpire available water is - as far as I know - completely unsupported by empirical data. On p O-23, swamps at West Cliff and Dendrobium are quoted as having had closure >200 mm without damage other than cracking and minor erosion. I have no information about the West Cliff swamps but Dendrobium swamp 12 is in a yet-to-be mined area (Area 3A, LW7 as shown in the Area 3 SMP fig 5.1 below) with a predicted closure of 335 mm (as in this EA). An error like this reduces confidence in the conclusion that only in-valley (valley infill) swamps are likely to suffer damage.

BULLI SEAM PAC Feb 2010 Comments Ann Young 7 Indeed while the quoted comment from the Metropolitan PAC Report (p O-22) suggest the members of the Panel are tired of my repeating that headwater swamps are at risk, I continue to be frustrated by my inability to persuade them otherwise. I know of only one headwater swamp (as now defined) affected by the levels of subsidence that are now routine in the Southern Coalfield (Dendrobium swamp 1), and in my view that has been severely damaged. After the recent rain, the piezometer readings from this swamp should over the next month or so give us a clear indication about whether the swamp can still retain water or not.

5. Assessing the significance of swamps The modified Approval for Dendrobium Area 3 makes this condition for swamp 15a:

The Applicant shall ensure that subsidence does not cause erosion of the surface or changes in ecosystem functionality of Swamp 15a and that the structural integrity of its controlling rockbar is maintained or restored, to the satisfaction of the Director-General.

Presumably this swamp would, in the terms of the Metropolitan PAC Report quoted on p O-33, be either ‘of special significance’ or one with ‘the potential to result in significant negative environmental risk’ and where ‘the potential environmental consequences are to be avoided by avoiding the impact’. Swamp 15a is a large (1.5 km long) swamp that extends from the headwater section near the ridge into the valley floor. In my opinion, it is not a valley infill swamp, in the definition of Tomkin and Humphreys (2006). The swamps of similar rating are CRE-S3b, CT1-S2, CT2-S7, DAC-S3, OHC-S5a, STC-S24, STC-S36 and WOR-S5a. At a minimum, these 8 must receive the same protection as swamp 15a in Dendrobium Area 3A. In fact the establishment of a category of ‘swamps of special significance’ has added a new dimension to the decision-making on swamp protection. The EA concludes that there are no such swamps in the project area. This would be hotly disputed by many in the community. The complex DAC-S5a/WOR--S3 covers 52 ha, and if joined by WOR-S4 as it is naturally, gives a swamp of 78 ha. The CT1-S4,5,6 complex covers 49 ha, yet does not even rate a mention with the other ‘top 8’. Dahlia Swamp is smaller (23 ha) but the community value placed on this swamp is very high. The category of ‘special significance’ must recognise all values associated with the swamps - their hydrological significance, their perceived value by the community, their ecological value, their significance as geomorphological and pedological features. I believe the dismissal of all the swamps in the project area as less than special undervalues them greatly. The ‘inclusion of community values’ into Choice Modelling (p O-43) may have led Gillespie Economics to conclude the swamps are not worth more than the coal underlying them, but I suggest their assessment of community values may underestimate community sentiment - not to mention the government’s responsibility to use the precautionary principle in managing the water resources of the catchment, and the long-term benefits of nature conservation in a long-protected area. The over-riding of long term water resource stability by short term economic benefits should not continue. A responsible government must take consider inter-generational equity and conserve irreplaceable resources. Also the separation of the 8 swamps with >200 mm closure throughout + high erosion risk seems rather arbitrary. All that Chart O-5 (p O-30) does is to add a ‘changed erosion risk’ cutoff to the existing closure parameter. The text gives no real justification for using 0.2 rather than, say, 0.1 for changed erosion risk. And while the erosion risk is a logical criterion, it is derived from the relationship of pre to post mining gradients. Presumably the other factors used to estimate bed shear are generalised estimates from the sediment characteristics and catchment size. If however we analyse some of the enormous mass of data in the spreadsheets of Attachment O-B somewhat differently, it is hard to see why these 8 swamps

BULLI SEAM PAC Feb 2010 Comments Ann Young 8 should be singled out from others. I plotted data for 2 groups of swamps - the 8 identified, and another 11 which were 1 km or more in length. Using tilt as an indicator of erosion risk, and closure, the 2 series of data are not nicely separated like those in Chart O-5. In fact you might argue that the risk to those not included may be higher than that for the 8 identified. Furthermore, the swamps not included are generally larger than those that are, or are almost contiguous with them. It is also misleading to discuss the swamps as though they are discrete. This particularly applies to the CT1 - S4,S5,S6 complex, the Dahlia Swamp complex DAC-7a&b, the DAC-S4/OHC-S6 complex and the WOR-S4 which joins the WOR-S5a/DAC-S3 complex. These large complexes obviously have a higher value ecologically and hydrologically than smaller individual ones. Yet the analysis does not take this into account. For these reasons, I consider that the analysis of the risks to swamps are inadequate and rest far too heavily on a selective use of a mass of comparative data.

BULLI SEAM PAC Feb 2010 Comments Ann Young 9 6. Cliffs and steep slopes In Appendix A, Section 5.3 (p 80 ff), the rockfalls in several mining areas are analysed. The comment that rockfalls can be due to natural causes such as high rainfall is obviously correct - but it is a fact that natural rockfalls outside mining areas are extremely rare. While cliff collapse is a natural process, subsidence increases its frequency and its severity greatly. The estimate of 3-5% of cliff length to be affected is interesting, given that the extent of damage in Dendrobium Area 2 was considerably higher. Also I note that toppling is the mechanism proposed for rockfalls. Yet in natural conditions, and certainly between Dendrobium LW 3&4, back-tilting and sliding was the mechanism involved, as was also the case in the spectacular mining-induced Nattai failure. Furthermore, while stress concentration in valley floors certainly means that upsidence is highest there, the cliff lines are not exempt from the impacts of closure. Along cliffs, the release of locked-in horizontal stresses, and the very fact that the rocks are unsupported on one side, mean that joints open naturally. The phenomenon of block gliding shows this dramatically. Clearly when the area is undermined and valley closure occurs, the widening of joints may be accentuated. Hence the likelihood of cliff collapses increases. Also it is very surprising to see that Section 5.4 on steep slopes gives minimal attention to the steep slopes on Wianamatta Group strata in the Razorback area. Given the high natural instability of these slopes and that surface cracking seems anticipated, then geotechnical assessment of the impacts of changed surface and soil water flows, tilts and slope angle changes, possibility of flow diversion to deeper-than-pre-mining levels and altered stress levels, would seem appropriate.

7. Summary Even allowing for the fact that this a general blueprint for a long term project, there are major flaws in the EA:

1. It lacks any real analysis of process and relies too strongly on optimistic assessments of past impacts (eg the sections on rock bars and swamps)

2. It draws conclusions from inadequate data and presents them as facts (eg the section on the contribution of swamps to hydrology)

3. It appears to assume that only comparative analysis of subsidence parameters is needed to predict impacts of mining on the landscape (eg the absence of any geotechnical assessment of the Razorback area)

4. Its risk management systems involve techniques that the community does not regard as proven or always feasible (eg grouting of cracks) and monitoring systems that are not in place now and may not be in time to provide useful baseline data

5. It continues to argue that short term economic benefit outweighs the long term value of water resource protection.