After Conflagration: Analysis of
Long-term Pollution
T.M.DANIELLV. KULIKCYPUM PTY LTD, G.P.O. Box 863, Canberra City, ACT, Australia 2601e-mail: [email protected]
CONTENTS
Created:
with Microsoft Word 2000; to print: volume=50 pages.
To
get better pictures: click CAPTIONS (go to the end of this paper).
1.
Introduction
2.
Relationship between topography and runoff hydrographs
3.
Runoff and water quality
4.
Mutual dilution
5.
Diurnal runoff variations, soils versus hydrochemistry
6.
Twins model
7.
Three Dimensional Technique
8.
Conclusions
9.
References
Appendices
ABSTRACT
Simple
graphical methods for organization, debugging and analysis of water quality
data are developed. An example of an application of these methods to
observations made on Licking Hole Creek Catchment, a mountain forest watershed,
burned by an intense conflagration, shows the sensitivity of the technique. It
is shown that some traps are in store for users of standard methods of
analysis.
The
technique could be applied:
1. To other conflagration affected watersheds. The objective
is to compare the data before and after conflagration.
2. To other catchments with major changes in land-use, such
as urbanized catchments.
3. To the analysis and organization of the data of the
hydrochemical archive for all stations, using the developed 3-D technique. The
objective in this case would be the creation of indicator diagrams which could
be used by analysts in the laboratory as "rating curves" and hence
ascertain quickly any changes in water quality.
1. INTRODUCTION
The
effects of conflagrations on the hydrological regime of rivers are reported in
many articles, published all around the world.
From
the point of view of pure logic, the problem seems quite simple. There are two
primary effects, firstly the elimination of vegetation and secondly that of
watershed pollution with ash and litter.
Further
application of pure logic allows many authors to forecast "obvious"
secondary effects. For example, the elimination of transpiration is supposed to
increase the runoff and base flow. An increase of sediment and ion
concentrations should follow. Later on, the watershed should be in a process of
recovery, the changes should gradually disappear and the sediment and ion
concentrations will return to the pre-fire values.
However,
some watersheds do not comply with the strength of this pure logic. Instead of
an increase of the runoff, a significant decrease is sometimes reported. In
other cases no significant change of the runoff occurs after conflagration, but
several years later 1/4 - 3/4 of the runoff disappears [5].
Reports
on the water quality effects of a conflagration are even more controversial.
Sometimes it seems that only the first flood after the conflagration has
increased ion and sediment concentrations. Then, the concentrations become the
same as before the fire. In contrast to this idea, hydrobiological
investigations of lakes show that even 10 years after a conflagration the transport
of chemicals continue to be much higher, producing "explosions" of
algae growth [8].
Of
course these contradictions may be explained with the statement "due to
different geographical conditions", etc.
ONE NUMBER IN
THIS TABLE IS WRONG. GUESS WHICH ONE.
|
DAY |
FLOW |
T.BIDITY |
TOT.PHOS |
ORTH.PHO |
AMMONIA |
NIT(TKN) |
TOT.OX.N |
|
DDMM.YY |
CUMECS |
NTU |
MG/L |
MG/L |
MG/L |
MG/L |
MG/L |
|
1602.83 |
NA |
0.8 |
0.05 |
NA |
NA |
0.59 |
NA |
|
2302.83 |
0.039 |
6.9 |
0.1 |
0.003 |
0.171 |
0.8 |
0.006 |
|
103.83 |
0.008 |
8.2 |
0.092 |
0.048 |
0.028 |
0.61 |
0.01 |
|
2203.83 |
0.62 |
7.8 |
0.12 |
0.036 |
0.01 |
1 |
0.23 |
|
2403.83 |
0.2 |
4.3 |
0.061 |
0.024 |
0.017 |
0.44 |
0.082 |
|
3003.83 |
0.074 |
1.9 |
0.043 |
0.025 |
0.03 |
0.1 |
0.079 |
|
1104.83 |
0.4 |
2.6 |
NA |
0.026 |
0.059 |
1.37 |
0.26 |
|
1904.83 |
0.12 |
1.6 |
0.026 |
0.022 |
0.03 |
0.18 |
0.024 |
|
2604.83 |
0.12 |
1.9 |
0.067 |
0.04 |
0.001 |
0.26 |
0.02 |
|
205.83 |
1.14 |
4.7 |
0.06 |
0.032 |
0.04 |
NA |
0.047 |
|
405.83 |
1.52 |
4 |
0.13 |
0.032 |
0.018 |
0.22 |
0.17 |
|
505.83 |
1.05 |
2.3 |
0.036 |
0.03 |
0.011 |
0.12 |
0.12 |
|
3005.83 |
1.47 |
14 |
0.72 |
0.018 |
0.008 |
0.74 |
0.29 |
|
2608.83 |
1.32 |
12 |
0.081 |
0.01 |
0.038 |
0.69 |
0.18 |
|
2708.83 |
1.47 |
3.6 |
0.014 |
0.008 |
0.01 |
0.25 |
0.26 |
|
2908.83 |
0.913 |
2.2 |
0.036 |
0.005 |
0.01 |
0.14 |
0.15 |
|
111.83 |
0.733 |
2.6 |
0.021 |
0.004 |
0.016 |
0.224 |
0.059 |
|
911.83 |
0.798 |
2.5 |
0.083 |
0.006 |
0.034 |
0.345 |
0.106 |
|
112.83 |
1.05 |
1.5 |
0.02 |
0.01 |
0.03 |
NA |
0.008 |
|
712.83 |
1.18 |
2.4 |
0.04 |
0.003 |
0.02 |
0.23 |
0.021 |
|
2701.84 |
2.18 |
4.4 |
0.036 |
0.012 |
0.01 |
0.56 |
0.052 |
|
2203.84 |
0.26 |
1.9 |
0.008 |
0.001 |
0.01 |
0.24 |
0.006 |
|
404.84 |
0.201 |
1.2 |
0.06 |
0.002 |
0.015 |
0.27 |
0.001 |
|
405.84 |
NA |
NA |
0.02 |
0.01 |
0.011 |
NA |
NA |
|
106.84 |
0.157 |
1.2 |
0.024 |
0.01 |
0.013 |
NA |
NA |
|
307.84 |
0.118 |
1.3 |
NA |
0.01 |
0.01 |
NA |
0.03 |
However,
in many cases the cause is much simpler. The results are more the product of
the pure logic than an analysis of data. Take as an example the ion
concentrations data in Table 1. These are data for the Licking Hole Creek
Watershed (about 100 miles SSE of Canberra City, Australia. A wild
conflagration thoroughly burned this 20.6 km2 of eucalypt forest in
January 1983.
With
these data we can "support" almost any logical statement about water
quality. For example, comparing an "average" of the total phosphorus
concentrations in 1983 with that of 1984, it is possible to say that during one
year the concentration decreased so dramatically that it was sometimes even
below the sensibility of the observations.
If
there was no other data, this could be the end of an analysis by means of pure
logic. However, a Gamet automatic water sampler was installed at the gauging
station. The device was operated during many of the high flow events, taking
samples at hourly intervals once the water level reached a certain stage
height. The information from these samples (Figure 1) completely destroys all
results of the previous brilliant analysis.
In
fact the total phosphorus discharge appears to be at its maximum in 1984, when
supposedly the water was "clean".
Because
this great discharge of phosphorus appears during a flood, it seems necessary
to correlate the concentrations with flow. From the point of view of pure logic
there must be for every flood a certain concentration-flow relationship,
although with time the curves may creep down.
Reality
again contradicts this pure logic (Figure 1). Note that if there had not been
these floods in 1984, the automatic samples would have given an "almost
logical" picture. However, even in this case the manual samples do not
show any correlation with water discharge (Figure 1).
The
usual shelter against bad concentration-flow correlation is to use
(mass-flow)-(water discharge) relationships (Figure 2). However, it does not
help too much in this case. The mass flow of total phosphorus and Kjeldahl
nitrogen seems to show some correlation with discharge, but the accuracy of
this correlation is 10+-1.
This
means that the highest value of a mass flow, corresponding to a water
discharge, may be 100 times greater than the lowest. The graphs for other
elements are much worse. This makes impossible any comparison with a control
watershed or pre-fire observations. It would seem that this is why it is
impossible to detect any significant increase of the chemical transport, except
for a first flood.
The
example, shown above, demonstrates the state of the art of conflagration effect
studies. It seems obvious that there will be no progress if we continue to use
just pure logic and conventional methods of analysis. Therefore the main
purpose must be not to analyze a particular conflagration or effects but to
develop a simple and more accurate technique, to allow a reasonable judgement
to be made. An analysis of data and a closer integration of different types of
observations must be the primary objectives. As to pure logic and predetermined
ideas - we have nothing against them on the condition that they are of help in
reaching these objectives.
2. RELATIONSHIP
BETWEEN TOPOGRAPHY AND RUNOFF HYDROGRAPHS
The
watershed has a very distinctive topographical structure (Figure 3 and 4). It
consists of 10-20o steep mountain slopes and swampy
"flats". About half of the mountain slopes deliver water to these
flats.
Some
reservations, usual for a mountain watershed, must be made on the accuracy of
the water balance components (Figure 5), especially the rainfall data. The
runoff data are more accurate. The rating curve for the gauging station,
situated in a rocky gorge, is stable, although the discharges above 1.6 cumecs
are calculated by an extrapolation of the rating curve.
An
eucalypt forest on volcanic podzolic soils existed prior to the conflagration
of 1983.
Recession
curves (Figure 5 and 6) show that the watershed continues to produce runoff
even several months after a flood, despite of an absence of rain. Figure 6
shows a recession curve in a "natural" co-ordinate system. Although
the initial period of the recession is, as usual, curvilinear, several days
after the flood the curve became more or less linear with the time decrement of
about 10-20 days.
Up to
this point everything looks only too decent, but this is again the result of a separate
analysis of different data. As soon as we try to link them, we shall have
troubles. For example, lets try to compare the map data and recession curves.
Several
days after a flood the "quick flow" is supposed to cease. The
subsurface runoff becomes more or less stable and we can use Darcy equation in
a simple form
Q =
KHlI
to
estimate the runoff Q (m3/s), as a function of the average
coefficient of filtration K (m/s), the height of the water table H (m), the
length of the drainage network l (m) and the slope I (m/m).
The
water balance equation for a recession is
Q = m
A (dH/dt),
where
m is the specific yield (m/m) and A is the area of the watershed (m2).
Hence
Q = Qo
- C2 V,
where
Qo
(m3/s) is a discharge at the moment to ,
C2=KIl/mA,
dV/dt
= Q.
We
can estimate C2 from a recession curve (Figure 6). The hill slopes
are 10-20o steep and I = sin 15o = 0.25. There are not
any data on m but usually for podzolic soils m=0.1 - 0.3. The length of the
stream network, measured on the 1:10 000 map, is 20 km. Then,
K = C2
m A/Il = (0.05x0.1x20x106/0.25x20x103) = 20m/day
This
is about 1000 times greater than any reasonable value. In fact, 25m/day is a
coefficient of filtration for gravel. Some direct field measurements [9] in
this region (Cotter River watersheds) gave, as a maximal value, 0.7m/day.
Although
all the data, used in these calculations, are not very accurate, any reasonable
correction of them may not change the result significantly. For example, we
cannot suppose that the slope of the groundwater flow is greater than 15o.
Most probably it is less than this value. But even if we suppose that the
slopes are vertical, then I=sin 90o = 1 instead of 0.25. This does
not help too much.
The only way to fix this contradiction is to suppose that the watershed has implicit drainage structures. It would follow that the most of the subsurface runoff has to be delivered to the creek by these structures. This means that the watershed has to have springs. The 1:10 000 map, however, does not show any springs. As a last resort, we may suppose that the watershed has more than 1000 springs, so little that they are not worth mapping. Then the underground drainage network of these springs will increase the length of the drainage system by a factor of 103.
No
such springs were found in the gorge around the gauging station. On the 24/9/85
(the third day after the rainfall of 21/9/85), a survey was made along the line
shown on Figure 3. Walking in the middle of the slopes for several hours,
nobody could find any trace of a spring until by chance the first one was
found, giving us an idea where and how to find others. The spring could be
traced only due to a very narrow strip of moss vegetation. Walking upslope to
the beginning of the strip, we found a little puddle 30-40 cm in diameter.
Walking down slope we could find the end of this drainage path and a terminal
point with a bigger puddle. During the next hour 36 such springs were found.
Any little gully, which is not shown on the map as a hydrographical network had
a little stream. Walking along the gully, we could find many of these little
puddles feeding the stream.
If
there are 36 little springs along 2-3 km on one side of the river, then along
the 20 km of the river network it has to be hundreds of them. So, it is indeed
a watershed of hundreds of springs, where the subsurface flow plays a very important
role.
3. RUNOFF
AND WATER QUALITY
At
first sight, hydrographs from this catchment do not show much evidence of
quickflow runoff. An attempt to correlate peak flows with hourly intensities of
rainfall was completely unsuccessful. Instead, the peak flows correlate very
well with the sums of precipitation. So it seems that the subsurface component
plays a major role even during high flow.
Very
high concentrations of sediment were recorded during flood events but neither
the sediment concentration, nor the sediment load show any correlation with
discharge (Figures 1c, 2) therefore it seems that the sediment is not a product
of riverbed erosion. Hence, some surface runoff has to exist, probably on the
flat parts of the watershed.
From
this, the idea to use a hydrograph separation technique to establish a
relationship between water quality and a separated flow originated. There are
many different methods of hydrograph separation and a variety of these could be
used, trying them and reasoning why some of them give better results than
others.
Perhaps,
we could even find the best method through "pure logic". But what we
really need to do is just to organize our data.
Suppose
that the separation of the base flow of the hydrographs, is done so that for
each time increment a value of, say, "sep-flow" is found then suppose
that there is some relationship between the concentrations and the
"sep-flow" q. This would mean that, if we have such a relationship
for the concentration of one chemical element
C1
= f1(q),
and
for another
C2
= f2(q),
then
C1
= f3 (C2).
Therefore,
if any of the hydrograph separation techniques are successful, we should have a
good relationship between the concentrations C1 and C2.
The
data show that there is some weak correlation between the concentrations, but
the accuracy of such relationships is less than 10+-1.
So,
the success of any method of hydrograph separation may be only very
limited.
4. MUTUAL
DILUTION
If
there are surface and subsurface components of the flow then the surface component
produces "dirty" water and the subsurface flow generates the
"clean" one, although the clean one may contain a lot of dissolved
chemicals. This does not mean however, that the concentrations or
turbidity will depend on the amount of the surface flow. The subsurface
component may dilute the surface component and the final turbidity will be low.
Reciprocally,
if the subsurface flow carries most of the dissolved chemicals and the
"dirty" overland flow is relatively clean from these chemical
components, the concentrations may not be correlated with the base flow.
Because
the hydrograph has a subsurface component that seems important even during
flood events, we cannot have good results until this mutual dilution is taken
into account.
Let
us organize the data to eliminate this mutual dilution using the simple fact
that any dilution effects only concentrations and cannot change the mass flow.
Then, the relationship between the mass-flows should be much better then than
between concentrations. Figure 7 seems to support this idea.
The best correlation is between the mass flows of total phosphorus and Kjeldahl nitrogen. This probably represents the "quick" component of the runoff, because most of the phosphorus moves riding on the surfaces of fine particles. Hence, there is a simple way to improve this relationship. Instead of the total phosphorus, one may use the value of particulate (non-filterable) phosphorus. Similarly, it is better to use instead of the Kjeldahl nitrogen, the value of the organic nitrogen, because the ammonia may be partly delivered by the subsurface flow. Figure 7 shows that this improves the relationship in the lower part of the graph.
7c – absent
5. DIURNAL RUNOFF VARIATIONS & SOILS VERSUS
HYDROCHEMISTRY
In
summer the "harmonic" diurnal runoff variations are due to
transpiration [2]. Naturally, immediately after a conflagration the harmonics
of the runoff cease and resume only when the new generation of vegetation
appears on a burned watershed [5].
Usually,
these fluctuations have not been considered worthy of any serious analysis.
They represent a sort of curiosity. However, it is possible to show that these
data contain considerable information, which has to be integrated with other
observations. As a first step, they can be used as an indicator of the recovery
of the transpiration.
After
the January 1983 conflagration, the harmonic variations were not registered
during February - May. Then, the winter period (May - October, 1983) occurred
and the vegetation was dormant. During the spring month of the November, 1983
the diurnal harmonic variations recommenced (Figure 8).
By
itself, this would not be very interesting; the young vegetation simply began
to drink water. However this also means that a nutrient consumption by the
vegetation may have occurred. The nitrogen content in podzolic soils is usually
very poor and any young undergrowth depends strongly on the nitrogen supply. It
is natural that the consumption of these building blocks of chlorophyll has to
be very high.
The trees and grass cannot consume organic nitrogen. Hence, the production of nutrients depends on decay. The simplified formula of the decay is:
ORGANIC
NITROGEN --> AMMONIA
--> NOx.
If the supply of organic nitrogen is sufficient, then the loss of ammonia to the atmosphere regulates ammonia concentration. This may explain the organic nitrogen-ammonia correlation of Figure 7. The amount of NOx depends mostly on leaching and vegetation consumption. Hence, the mass flow of NOx has to drop as soon, as the diurnal runoff variations recommence.
Figure
9 shows that the NOx data in fact consist of two sets, each are
having correlation with other data, which is no worse than the other graphs.
The initial difficulties were due to the mixing of these two sets.
6. TWINS
MODEL
Up to
now we were operating with discharges and mass flows. However, there is some
opposition to this approach [6, 7]. There is an opinion that because mass flow
is calculated by multiplying concentration by discharge, a sort of
"non-existing" correlation between mass-flow and discharge may be
produced. This opinion is mostly due to some works, where the accuracy of the
relationships was estimated by means of standard statistical criteria, i.e. coefficient
of correlation between logarithms. If, instead of this, maximal deviations were
used to estimate the accuracy (as on the Figure 2 and 7) then the artificial
correlation would not appear.
Anyway,
let us use for the next stage of the analysis the concentrations instead of
mass flows. This will make graphs more varied and less smooth but, as one can
see, the accuracy will not be less.
If
the genesis of the water quality is due to the two different sources, it is
better to suppose that the both of these sources give a contribution to the
mass-flow. Then, the mass flow for i-component is
Ri
= fi (Q1, Q2)
and
depend on the flow from the "flats" Q1 and from the
springs Q2.
Note,
that now the formula Ri = f (Rj) for the mutual dilution
model does not work.
The
concentration of i-chemical at the gauging station is Ci = Ri/Q,
where Q = Q1 + Q2.
Then
Ci = fi (Q1, Q2)/(Q1 + Q2)
= fi (Q1, Q2).
For
any three components one can obtain:
C1
= f1 (Q1, Q2) (1)
C2
= f2 (Q1, Q2) (2)
C3
= f3 (Q1, Q2) (3)
Eliminating
the unknown components Q1 and Q2 results in C1
= f4 (C2, C3) (4)
Also,
because Q2 = Q - Q1, the equations (1) and (2) can be
rewritten as
C1
= f1 (Q1, Q-Q1) = f11 (Q, Q1);
C2
= f2 (Q1, Q-Q1) = f22 (Q, Q1).
Eliminating
the unknown component Q1, one can obtain
C1
= f44 (C2, Q) (5)
The
equations (4) and (5) may be compared with our data.
The
easiest and the worst method to do this is to use some statistical software,
calculating multiple regression coefficients. Any of these programs supposes
that the relationship between variables is linear or at least could be made
linear after some simple transformations of the variables.
The
best way to do this is to draw 3-dimensional graphs. There are some limitations
to this technique. As is shown later, the main condition of success is to have
enough patience to complete all the sets of the 3-D graphs, despite some bad
results.
Figure
10 shows an example of such a 3-D picture. The graph is the same as Figure 1a,
but now there is a label near every point. This is the value of the Kjeldahl
nitrogen, mg/l. The isopleths are drawn very approximately, by hand. Despite
this the graph allows the calculation of TKN as a function of discharge and
total phosphorus with accuracy of about 1.3+-1. This is much better
than the 10+-1 accuracy of mass flow graphs 2a.
Although
the isopleths have a peculiar form, most points fit quite well with the
exception of four. The deviated points are shown on the Figure 10 in frames. It
is easy to see from the Table 1 that these are the first four samples. They all
have a greater TKN concentration.
There
is a great temptation to jump back to pure logic and to say that we have found
a trend. It seems that before the flood event at the end of March 1983, there
were some greater concentrations. Then the situation stabilized. But let us
suggest that it is not our business to speculate, but simply to organize the
data. By the way, it would be good to have some more measurements in the space
near the "square" initial points.
The
main result is that now we do not have "too much data".
7. THREE
DIMENSIONAL TECHNIQUE
Let's
introduce an abbreviation for a title of a 3-D graph in the format (X, Y) Z.
Then, the graph 10 will have the title: (DISCHARGE,TOTAL PHOS.) TKN.
How
many graphs could be drawn if we had data for n elements? The amount of
combinations (X, Y) of n elements is n(n-1)/2 because a swapping of X
and Y do not produce any new results. Then any of the (X, Y) planes could be
combined with any other variable, except X and Y. The number of the graphs
therefore, is:
N =
0.5 n(n-1)(n-2).
For
the eight variables in the Table 1, we could have 168 graphs. It is obvious
that the full set of these 3-D graphs has to be made by a computer, although
the drawing of the isopleths and analysis may be done by human means.
Figure
11 shows another 3-D image. Figures 11a, 11b and 11c are just the front, side
and plane view of the same 3-D surface. Mathematically, it is sufficient to
have only one of them. Practically, as we can see later, all three views are
necessary.
Again, most of the data seems to fit in a logical place. The first four observations still reveal a trend. There is, however, at least one other point, which deviates significantly. The value of the total phosphorus for 30 May, 1983, is much greater than it is supposed to be. Note that on the graphs (TURB, TOT.PHOS) TKN and (TKN, TOT.PHOS) TURB the point is simply placed above other points. It is still possible that this is a result of special conditions and some additional isopleths could be drawn around it. However, the graph (TURB, TKN) TOT.PHOS. is quite different. The point is closely surrounded by others and all the neighbours have about 10 times less TOT.PHOS. concentration.
A
request to the chemical laboratory for the original records showed that this
was simply a bug, created during an input to the computer database. The
original value of total phosphorus is not 0.72 but .072.
This
example shows that three reciprocally perpendicular views are useful for a
debugging and organization of the data. However, there is something more
important behind this routine. An attempt to draw these graphs may be easily
discouraged at an early stage by systematically bad results. Instead of a
peculiar but organized "map" one may have a chaotic set of labels and
give up. It is necessary to draw at least two other perspectives of the
"chaos" and the result of this simple "repetition" may be
much more successful.
To
explain why this may happen, we shall use an analogy. It is possible to think
of a 3-D image as a topographic map. If it is a map of a flat slope
descending, say, to SW, all three views (from above, from S and from W) will be
organized sets of parallel lines. A horizontal look from N will also show a
well-organized picture, although we have to understand that we are now looking
on the slope from the point of view of a mole, buried underground. We see now
an "internal" surface of the slope.
In the case of a hill, only a view from above gives an organized picture. Looking "through the hill" from the South, we superimpose ascending and descending slopes. The labels near "experimental points" are now a horizontal distance (or latitude) and every point can have at least two labels: one for the front side of the hill and another for the opposite ("invisible") part of the hill. This will create an incredible chaos of the labels and may make it impossible to draw any isopleths.
For a
more complex sort of surface all three projections may be bad and it will be
necessary to find a right perspective in order to have an organized picture.
But even in the case when all three views are organized, as for Figure 11,
usually one of them is better and more suitable for a particular type of
analysis.
It is
possible to make some minor improvements on Figure 10 and 11, but most of these
corrections would not be very important. It is improbable that we can create a
relationship between concentrations, which will have more than 20-30% accuracy
and the graphs are very close to this limit. Some results may be obtained by an
analysis of the points, which have much greater deviations.
8. CONCLUSIONS
Further
paths of the analysis are obvious. We can now use this graphical technique for
the traditional comparison of the hydrochemical regime before and after a
conflagration and for hydrograph analysis. Let's make some concluding remarks.
Although,
after a conflagration, hydrochemical data looks chaotic they contain much
information. The disaster gives one a unique chance to analyze hydrologic
behaviour using the chemical indicators more or less evenly spread over the
watershed surface.
The
general understanding of this "chaos" can be achieved by a systematic
integration of all sorts of data, analysis of interconnections between them.
However, use of a standard statistical technique, averaging or over-indulgence
in pure logic may lead to completely wrong conclusions.
9. REFERENCES
1. Blong R.J., Riley S.J., Crozier P.J. (1981). Sediment
yield from runoff plots following bushfire near Narrabeen Lagoon, NSW. Search
13(1-2): 36-38, 1979.
2. Bren L.J., Flinn D.W., Hopman P., Leitch C.J. (1979). The
hydrology of small forested catchments in NE Victoria. Bulletin No. 27, Forest
Commission, Victoria, 48p., 1979.
3. Brown J.A.H. (1972). Hydrologic effects of a bushfire in
a catchment in south-eastern NSW. J. Hydrology 15: 77-96, 1972.
4. Burgess J.S., Reiger W.A., Olive L.J. (1981). Sediment
yield change, following logging and fire effects in dry sclerophyll forests in
southern NSW. Int. Ass. Hydrol. Sci. : Pub. N132 : 375-385, 1981.
5. Langford K.J. (1976). Change in yield following a
bushfire in a forest of Eucalyptus Regnans. J. Hydrology, 29, 87-114, 1976.
6. O'Loughlin E.M., Bren L.J. (eds.) (1982). The First
National Symposium on forest hydrology. Melbourne, Australia, 152 p., 1982.
7. O'Loughlin E.M., Cullen P. (eds.) (1982). Prediction in water quality. Proceedings of Symposium, Australian Academy of Sciences, 453 p., 1982.
8. Small I. (1978). The flora of impoundments. Proc. AWWA
Summer School, Hobart.
9. Talsma T. (1983). Soils of the Cotter Catchment Area, ACT
: distribution, chemical and physical properties. Austr. J. of Soil Res., 21,
1983.
Some preliminary results were published
promptly in:
Water
International, 1988,v13, No.2, pp.74-79.
And
CAPTIONS:
5Mb
Figure
1a Concentration - discharge relationships. All graphs are black and white -
YAK!
Figure
1b Concentration - discharge relationships.
Figure
1c Concentration - discharge relationships.
Figure
2 Mass-flow as function of discharge.
Figure 7 Relationships between mass flows (g/s).
ORG.NITROGEN=TKN-AMMONIA;
PARTICULATE PHOSPHORUS = TOTAL PHOS. - ORTHO PHOSPHORUS.
Figure
8 Diurnal runoff variations: 19-22 November 1983.
Figure
9 Total oxidized nitrogen as a two set record.
Figure
10 Kjeldahl Nitrogen as function of discharge and total phosphorus.
Figure 11. Three views on 3-D image TURBIDITY - TOTAL PHOSPHORUS - KIELDAHL
NITROGEN
Figure
11a
