Response of Dr Hansen and Sandy Lake Conservation Association (SLCA) to Sandy Lake Watershed Study

The SLCA has serious concerns and questions about the Sandy Lake Watershed Study – Final Report (AECOM 2014) that we believe must be addressed before the document is accepted by HRM and before any form of development can be allowed in the Sandy Lake (Bedford) watershed.

In addition, we have related concerns about the general development process relevant to lakes. We would like to see improvements to the system piloted at Sandy Lake and other lakes to benefit all lakes in HRM. A new watershed policy and superior yet cost-effective practices can better avoid potential problems and can ensure healthy lakes in HRM for citizens to enjoy over the long term.

We lack confidence in the total phosphorus (TP) data set and in how it was used to draw the conclusions in the report.  The coefficient of variation[1] in the TP concentration is very high, indicating that the inherent variability in the data is 70% as large as the mean value of the TP concentration (Figure 8).  The data set is very small (17 samples) for the time period under consideration (about 10 years for Figure 8), with no consistency as to the time of year or the depth of sampling, as far as sampling strategy was concerned[2].  The results span all three trophic states (categories or levels of nutrient-richness), from eutrophic (far too rich), to mesotrophic, to oligotrophic (very nutrient poor).   Just over[3] one-quarter of the total phosphorus data collected since early in 2005 puts the lake in the eutrophic category (Figure 8). Such a data-set is hardly a sound scientific basis on which to base the broad conclusions that the lake “is” mesotrophic, nor is the data of sufficient quality or consistency to make reliable predictions.

The AECOM report omits quantitative statements of the level of uncertainty in their predictions, such as confidence intervals[4].   The early-warning TP value of 15 mg/L would be easily enveloped by the bounds of a common confidence interval (such as the 95% C.I.).  The report seems to be lacking with respect to quantifying such uncertainty, in both the backward and forward senses.  Given such uncertainty (both evident and unstated), to have a water quality objective for this lake that is the “upper limit of the mesotrophic range” (page 31) is not at all comforting to SLCA, especially given that:  (i) eutrophic urban lakes (the next trophic level) are not amenities but are liabilities, (ii) eutrophic lakes have a short life-span as lakes, and that

(iii) it is admitted in the WQM (water quality management) plan that Sandy Lake is “highly vulnerable” (page 42).

It is general knowledge that phosphorus and nitrates can worsen the trophic state of a water body.  However, the warming of bodies like Sandy Lake and Marsh Lake[5]:

  1. increases the rate at which nutrients such as TP are eventually used by algae, increasing their abundance,
  2. decreases O2 levels, altering the ecology of a water body, by affecting what flora and fauna are favoured and what species are even fundamentally viable[6],
  3. as a side-effect of the joint action of #1 and #2 above, the nutrient level a water body can be altered (worsened).  Page 14 of AECOM report admits this by saying that “Additionally, the increase in impervious surfaces, such as asphalt roads, and heat retention of these surfaces may increase water temperature, which can also adversely affect the lake’s aquatic health.” (bottom of paragraph 4).  In connection with possible changes to the water quality of Florence Lake on Vancouver Island (due to urbanization), the BC Dept of Environment has stated (by way of background information) that: “Challenges to water quality management on Florence Lake include phosphorus loading from non-point sources, shallow depths, warm temperatures and low oxygen levels, primarily during the summer months. Excess phosphorus can cause spring and summer algal blooms as well as the spread of aquatic vegetation. When the vegetation and blooms die off and settle to the bottom, this can lead to oxygen depletion in the lake which provides favourable temperatures and photic opportunities for algae growth throughout the water column. Furthermore, as O2 levels decrease near the bottom, internal nutrient loading occurs, whereby phosphorus is released from the sediment and enters the water column, exacerbating an already nutrient rich environment. With the lack of flushing of the lake in the fall and winter months, these nutrients are not removed and the process begins again.”

If the Terms of Reference prepared by HRM precluded consideration of temperature, they were fundamentally flawed.  If they did not, the modelling approach used by AECOM was flawed.

It is admitted in the report that the model used to develop the conclusions in this report is a steady-state model (pg 32).  That is, it inherently reaches some new equilibrium state with respect to its nutrient level or balance, as an outcome for a given set of imposed hypothetical conditions.  By contrast, the TP data (though very scattered) has a clearly upward trend, showing increasing TP levels over time, and this with no significant historic increase in the amount of actual urbanisation.  We are concerned that the Lake Capacity Model (LCM) results are accepted and believed even though they do not account for this temporal trend; they are used to suggest that TP will simply plateau in the mesotrophic range (we note again that the LCM does not consider temperature, nor associated micticity[7] changes, nor temperature-change-driven TP release from lake-bed sediments[8]).  Page 5 of the report states that the “retention time” of Sandy Lake is only about four months (paragraph 5 – White et al. 1984).  This means that the time scale of the replacement of the 6 million m3 in the lake is routinely encompassable by a single summer season, indicating a clear potential for higher summer lake temperatures, given that urban runoff has persistently higher summer temperatures.

The LCM, now 39 years old (Dillon & Rigler 1975. Appendix A), was developed for, and first applied to, various lakes/watersheds (and therefore soils and bedrock geologies) in southern Ontario.  It does not take into account microbiological contamination (Dillon & Rigler 1975), as is commonly caused by pets in urban areas (especially cats), nor does it account for phosphorus release from bed sediments, as can be caused by low oxygen in the hypolimnion (Dillon & Rigler 1975, page 1522, column 1)[9].  Its application to small temperature-sensitive lakes is questionable.  There is an abundance of more recent research that describes how urbanisation changes the quality of urban receiving waters, including temperature and ecological changes (Jones et al. 2012).

Firstly, the Sandy Lake Conservation Association is concerned that this well-known warming phenomenon was not considered.  Secondly, an important aspect of the LC model that must drastically affect results coming from it is the TP generation rates (“export coefficients”).  We are told that “large-lot residential” developments have export coefficients of 0.2 kg/ha per year, whereas “commercial” land uses have export coefficients of 0.6 kg/ha per year.  We are also told that good stormwater management practices would have the effect of reducing this phosphorus export rate by “50%” (page 40-41).  Clearly, such phosphorus ‘export’ rates (and hoped-for reductions in such rates) are educated guesses; changing them would completely change the conclusions of this report.  This is admitted by the progenitors of the LCM: “Uncertainty in the phosphorus export figures and in the loading from precipitation alone could result in a 100% error in the calculation of the natural phosphorus while factors such as the soil retention factor are still only approximations”.  We are concerned that no analysis of the amount of inherent uncertainty in the assumed export coefficients is presented in the AECOM report, nor are supporting references given for the values that were used (no authorities are cited beside the stated values of the coefficients).  Are these export coefficients well-supported by virtue of NS experience (post-development data)?

Page 1 mentions municipal services, and Figure 2 shows “water and sewer services”.  The frequently used word “sewer” in the report does not differentiate between the sanitary sewerage systems and the storm-sewer system (real or hypothetical), but it should have.  The bottom of page 1 apparently indicates that storm sewerage is not part of future development, in some cases.  Since TP does indeed come from stormwater, the SLCA would like to see where all future storm sewers will probably be located, especially the outfalls from same, and whether or not any stormwater will be directed into the lake (and/or its tributaries), and if so, how it will end up in the lake.  The sanitary sewerage system should have been portrayed in a figure, noting in addition whether a given area will instead remain on septic systems, under a given scenario.

The future area to be serviced by storm sewerage system(s) should have been portrayed in a separate figure, noting the most probable locations of any outfalls and stormwater retention facilities.

The report appears to advocate development Scenario 2 (page 45, chapter on satisfying HRM’s E-17 policy), but it is hard to find an explicit recommendation to this effect.

The amount of hydrologic detail in the report is also disappointing.  No pre-development and post-development hydrographs or other graphics are presented.  To get some sense of volumes involved, one should compare the volume associated with the expected amount of urban development to the lake volume.  If the volume of 6 million m3 (page 5) of Sandy Lake is spread over the expected newly-developed area of 361 hectares, a depth of 1662 mm results.  This means that an annual runoff figure of 850 mm (for example[10]) represents about half of this depth.  This means that if the annual runoff from the 361 hectares becomes both greater and warmer, this volume of warm water will represent a relatively large percentage of the volume of Sandy Lake.  We would like to see a description and discussion of studies on lakes of comparable relative size.

HRM’s Regional Plan’s E-17 (Appendix B) requires that specific recommendations be made from studies such as this one.  Contrary to the fact that changes to the stormwater runoff behaviour is the very reason that studies such as this one are done, we find the following vague statement on page 42: “The meaning of the term Advanced Stormwater Management does not reflect any specific methods of stormwater management….” (second paragraph).

Also contrary to the need for specific recommendations plural, page 45 of the report merely mentions undifferentiated “best forestry practices” (other than a simple buffer zone recommendation).

This report does recommend (i) that the effluent from the two wastewater treatment (WWT) plants in the watershed be ‘sent’ elsewhere.  In connection with the WWT plant overflows (due to an ‘overflowing’ sanitary sewerage system, or ‘SSO’), page 47 of the report states “Overflows typically occur during extreme weather events.  The timing, frequency, and severity of these events are not possible to predict and so the water quality impacts from overflows cannot be quantified or modelled.”  This statement is true if one is limited to the LCM model, but completely false in general.  Halifax Water and HRM have paid for extensive studies to quantitatively address just such SSO problems, and these studies are on their shelves.  (These have been done because the CCME has asked that municipalities put an end to all SSO’s, as per the statement at the bottom of page 51).

Related Questions and Concerns

Note: The watershed study is required to address E-17 of the Regional Plan.

  1. It is acknowledged in the report that urban development is currently increasing phosphorus concentrations in Sandy and Marsh Lake.  As stated in item E-17, the purpose of the watershed study is to determine the carrying capacity of the watershed and determine the amount of development.  All three development scenarios are full development scenarios for the areas targeted for development.  Given the current negative water quality trend and the low-flush rate of the lake, a development scenario that takes a precautionary approach and is not full-build out would be advisable.  Why was this type of scenario not considered? 
  2. All the scenarios contain a constraint of 20 metres around watercourses, wetlands, and waterbodies.  Given the objective in item E-17-i states to “identify and recommend measures to protect and manage natural corridors and critical habitats for terrestrial and aquatic species, including species at risk”, why was research conducted by Rideout (2012) stating “a wider buffer of >50m is required to provide terrestrial habitat services” not taken into consideration?  Our Sandy Lake advisors recommend a 60m minimum, given the vulnerability of the lake, and that more would be preferable.
  3. Also, item E-17-j states to “identify appropriate riparian buffers for the watershed”.  Given the intensity of the proposed development and the rising levels of nutrients, again, why is only the minimum distance of riparian buffers (i.e. 20m) recommended?
  4. It is acknowledged in the report that there was no comprehensive study of wildlife undertaken in the Sandy Lake area.  Within the Birch Cove Lakes watershed study, sensitive areas were listed as development constraints.  In the Sandy Lake report, merely stating that an analysis for Sandy Lake was not completed therefore there are no constraints does not satisfy the requirement of E-17-i. Why was there not an ecological study and/or GIS analysis conducted to identify sensitive areas to satisfy the E-17-i requirement?
  5. Within the Birch Cove Lakes watershed study old growth forest was considered a constraint.  The Sandy Lake report mentions a mature hemlock forest on the southern peninsula of Sandy Lake.  Mature hemlock represents a late-successional stage forest which is indicative of old-growth forest.  Data was collected on October 3, 2014 by a Department of Natural Resources (DNR) employee to quantify the mature hemlock (See Edward Glover’s submission on old-growth forest).  Bruce Stewart, manager of research and planning at DNR, states that this forest stand ranks high as old growth.  Item E-17-k states “identify areas that are suitable and not suitable for development”. Omission of old growth forests from the report implies that old-growth forest is suitable for development. Why was this old growth forest, or any other for that matter, not considered a constraint for the Sandy Lake watershed study?
  6. We are resubmitting our March 22, 2014 (Appendix C) letter because the final report did not include a table of concordance documenting how those issues from our letter were addressed.

“How Lakes Work” as applied to Sandy Lake:

“No Swim” advisories are often associated with high water temperatures, because bacteria multiply much more quickly when it is warm (the same reason why we refrigerate our food).  Protozoans eat bacteria and bigger ‘microscopic’ life-forms eat protozoans, and on up the food chain. The types of benthic invertebrates present (spineless bottom dwellers that live in sediment) are important indicators of water quality; persistently warmer water will alter the type and diversity of the limnic fauna (lake creatures) broadly speaking.  High water temperatures inherently limit the saturation level (upper limit) of Oxygen (O2) in the water, altering what kinds of fish will be happy in a given lake or stream.  All salmonidae (this family includes trout) like cold high-O2 water, whereas carp and suckers and perch are much more tolerant of poor O2 levels (but are considered less desirable as fish).  Salmonidae want gormet benthic invertebrates like dragonfly nymphs, but carp will eat garbage.

Warmth also tends to change the flora (botanic ecology); some aquatic weeds do well in warmer water, so when water warms one can get post-development ‘blooms’ of plants that you previously did not see much of, or any of, before.  These weeds increase the rate at which a lake traps sediment, matter that would normally just pass through the lake.  This trapping of sediment and the way that lakes become marshes and then bogs and then land is called ‘natural succession’.

There are more factors. At many Canadian beaches (such as Mooney’s Bay on the Ottawa River) there are regular closures right after it rains, in the summer.  This is because cat faeces have accumulated in the watershed and get washed into the river.

Dark south-facing surfaces should be avoided in new developments. One can intelligently use trees and orient the streets appropriately to minimize the warming of runoff.  One can orient roofs so that most of their surfaces are not south-facing surfaces (even though this increases people’s heating bills).  Rain barrels should be on all properties. A golf course should be limited as to what it can apply (like farmers, golf courses tend to waste fertilizer by putting on too much fertilizer at a time).

If the two wastewater treatment plants within the Sandy Lake watershed cannot do their job from time to time due to “I and I” (inflow & infiltration = surface water and groundwater leaking into the sanitary sewerage system) occurring every time it rains, methods exist for relining the pipes in the existing system (trenchless technologies).

New sanitary sewerage systems should be tested every 5 years and leaks fixed.

Another factor is Sandy Lake’s dimicity (It turns over twice a year), a function that can also be altered by watershed urbanisation.

We understand that the annual precipitation in the Sackville area is about 1200 mm per year (this number varies from year to year, of course).  This published annual precipitation value includes the water equivalent of the annual snowfall – itself a highly variable number.  If one has a watershed area which ‘suddenly’ has 30% more impermeable surfaces, one’s surface runoff proportion from that area goes up by 360 mm (and the groundwater system receives that much less of the annual precipitation).  As an example, if the amount of newly developed area is 1 km2, that represents an additional volume of water of 360,000 cubic meters (water likely with cat faeces etc. in it).  Whether or not a given lake is going to be strongly affected or not depends on ITS own volume.  A small-volume lake could be ‘killed’ by such a change.

Overall, we would have thought we would see some clearer work on the relative volumes involved (by AECOM); i.e. basic hydrologic budgets as compared to Sandy Lake’s own typical volume (Note: Bachiu’s “hydraulic” budgets, hydraulic is the wrong word).  There are many useful studies out there on this kind of thing.  To just study total phosphorus and use it as one’s sole criteria is too narrow.  HRM would be wise to catch up to the times on data collection.  Use of equipment now available for continuous monitoring of surrogate parameters is quite cost-effective and very informative, compared to having a guy fill a bottle from the lake.  AECOM had very little data to work with.  One does not get a picture of what is going on from a couple-of-dozen grab samples. The scatter would have been due to when in the year the sample was taken and how long since it had rained.  Drawing conclusions of this importance, that is, when development of this magnitude is being considered near a lake, from so little data is very questionable.


SLCA recommends putting a halt to all further development, subject to (i) implementation of the data collection program recommended by AECOM in Section 9 of this report plus an associated follow-up modelling effort of a more complete type, and (ii) the full and formal assessment of the state of health and abundance of the four species-at-risk, already identified as being at risk on page 50 of the report itself (salmon and turtles, asters and stitchworts).

If development is to eventually go ahead, HRM and Halifax Water should:

  1. switch households who are on septic systems (especially those that fail a timely inspection[11]) to a new sanitary sewerage collection system  before allowing further development,
  2. move both sources of WWT plant effluent before allowing further development (see on page 47).

The above would lessen two major existing nutrient sources before the addition of any new nutrient sources,

  1. increase the minimum buffer size to 60 m (given the stated vulnerability of the lake),
  2. pass a by-law requiring use of rain barrels on every residential lot in the Sandy Lake watershed,
  3. require the implementation of state-of-the-art stormwater management measures for any proposed sub-division developments, such as the use of permeable pavements (Ferguson 1994), and underground detention systems (Poornima and Davis 2010),
  4. expect the use of tree canopies (Jones et al. 2012) and the adjust orientation of developments at the sub-division-layout stage so as to lessen the warming of runoff (adjustment of azimuths of roofs and roads),
  5. disallow uncontrolled tree removal,
  6. ban the use of road salt in the watershed,
  7. disallow the bulldozing of drumlins, so as to preserve groundwater recharge sites (see last paragraph on page 9),
  8. limit the use of lawn fertilizers (amount applied, time of year),
  9. ban herbicides and limit the use fertilizer by the golf course (alluded to on page 47),
  10. specify a minimum lot size for all new lots[12], to a size that will clearly lessen adverse hydrologic changes.

The above might be ensured by passing by-laws before any further watershed change, thus communicating to potential developers what they are ‘getting themselves into’, before they begin construction.

In Closing

We trust that readers will see that we deeply care about the welfare of this lake that has been in existence for a very long time and that is being put at risk by human activity. Thorough evaluation along with careful application of current science and knowledge can retain its health as a continuing natural habitat as well as provide an enduring beautiful resource for recreation and enjoyment for citizens.

There is a need for better monitoring, better data, better analysis, better modelling, and better literature reviews. There is modern equipment for monitoring multiple parameters continuously (rainfall, lake depth, specific conductivity, temperature,…) and it is far more clever and affordable than it used to be.

As was noted in the second watershed meeting (Appendix D), the absence of policies that might be applied to HRM watersheds is part of the problem. This absence of policy is a political problem that HRM councillors, now aware of it, can rectify.  

Sandy Lake is already at significant risk from the uncontrolled clear cut that occurred in 2013. Putting into place policies and controls to prevent such events in other areas of HRM would logically be a part of watershed protective policies. We ask that such protective policies for trees and wooded lands be created and implemented across HRM.

Regarding the protection of trees and lakes:

  1. Florence Lake (1980‐2009) Water Quality Monitoring Program (Appendix E)
  2. Ottawa Tree By-law:

  1. Saskatchewan Lakeshore Development:

We respectfully request that HRM decision makers follow our lead to protect Sandy Lake and will develop policies that will benefit and protect lakes throughout HRM from this time on. We believe that our document provides much information that can be used to develop these improvements to policy and practice.

We in the SLCA would like to be part of the solutions and we offer our ongoing efforts and input to this end.

In the meantime, we request that this watershed report not be accepted by HRM, and that no development be allowed in the Sandy Lake watershed until our questions and concerns outlined in the bolded parts of this document are addressed satisfactorily, and until up to date lake protection policies and practices are put into place.

We also request a meeting with appropriate HRM staff and our local councillors to chart the way forward.

[1]  standard deviation divided by the mean.

[2]  it is significant that a detailed sampling program, one which includes temperature, is recommended by AECOM (page 43).

[3]  if the threshold data point of 20 mg/L is counted, 5/17 or 29.4 %  of the data indicates eutrophic conditions.

[4]  we note the vague admission of the presence of “inherent uncertainty” in the LCM model (page 42).

[5]  Marsh Lake is surrounded by a wetland that is quite large compared to its own surface area.

[6]  e.g. catfish and suckers are quite happy in warm eutrophic waters.  Page 22 of the AECOM report admits the possibility of oxygen deprivation in the case of Sandy Lake, and that it would promote TP release from sediments in the bed of the lake.

[7]  i.e. mixing regime, most Canadian lakes being dimictic – ‘turning over’ twice a year.  The AECOM report contains recognition of the fact that this actually occurs in Sandy Lake (top of page 18) but takes the matter no further.  The twice-a-year turnover prevents stagnation of the deepest water layers of a lake, and is temperature-driven.

[8] a significant fraction of sediment will probably accumulate in a lake, not simply flush through.  AECOM assumed a phosphorus sequestration rate by sediment in Sandy Lake of 33% based on an oxygenated hypolimnion.  Oxygen content decreases exponentially with increasing temperature, and increased lake temperatures are a normal expectation.  Sediment accumulation, over time, together with a warmer lake (lower O2) could have a significant effect on the actual TP sequestration/desequestration rates, but no modelling was done in this regard.

[9]  More capable models exist, such as the EQuIS LakeWatch Limnology Decision Support System, and the GLM-FABM ( model.

[10]  approximately the value found from the runoff map for NS, in the Hydrologic Atlas of Canada, AECOM estimates 755 mm.  The mean annual precipitation is about 1350 mm, with higher variability in recent years.

[11]  as per item 9 page 52.  We also note the statement on pg 32 of the report that: “Unfortunately, mitigation (sic) measures to reduce total phosphorus concentrations are seldom instantaneous or completely effective so …early warning values are often used to manage lake quality, rather than waiting for the … water quality objective to be met.”

[12]  one that is relatively large.  We note the following statement on page 40 “In Scenario 2 the contribution from small lot residential increases by a factor of three and is the dominant source of phosphorus, at 25% of the total load, in this scenario.”

Note: Assistance in the preparation of this report was provided by David Hansen, Ph.D., P.Eng.


AECOM (T. Bachiu) 2014. Sandy Lake Watershed Study – Final Report. Project report 60303077, AECOM Canada Ltd, Halifax NS, 64 pp plus appendices.

Dillon P.J. and Rigler F.H. 1975. A simple method for predicting the capacity of lake for development based on lake trophic status. Journal of the Fisheries Research Board of Canada, 32(9):1519-1531.

E, Rideout (2012), Setbacks and Vegetated Buffers in Nova Scotia,

Ferguson B.K. 1994. Stormwater Infiltration. Lewis Publishers, Boca Raton, 269 pp.

Jones M.P., Hunt W.F., and Winston R.J. 2012. Effect of urban catchment composition on runoff temperature. 138(12):1231-1236.

Poornima N. and Davis A.P. 2010. Thermal reduction by underground storm-water retention system. ASCE J. of Environmental Eng., 136(5):520-526.

List of Appendices

Appendix A: Dillon P.J. and Rigler F.H. 1975. A simple method for predicting the capacity of lake for development based on lake trophic status. Journal of the Fisheries Research Board of Canada, 32(9):1519-1531.

Appendix B: E-17 From The Regional Plan

Appendix C: Sandy Lake Conservation Association response to the February preliminary report

Appendix D: Sackville River’s Association, 2014 Sep Big Sandy Lake Watershed Water Quality Study

Appendix E: Florence Lake (1980‐2009) Water Quality Monitoring Program

Appendix F: Additional Resources