Go to National Library of New Zealand Te Puna Mātauranga o Aotearoa
Volume 10, 1877
This text is also available in PDF
(1 MB) Opens in new window
– 180 –

Art. XVII.—On Floods in Lake Districts and Flooded Rivers in general, with Methods adopted for their Prevention and Control.

Plate XIV.

[Read before the Wellington Philosophical Society, 2nd February, 1878.]

While at Queenstown, on Lake Wakatipu, during last November, a heavy rainfall was experienced, which, together with the melted snow on the main ranges, caused a rapid rise in the level of the lake. On Saturday morning, the 17th November, the rain-gauge at Queenstown registered .66 of an inch, which fell during the previous night; and on Sunday morning an additional 1.61, making 2.27 inches during forty-eight hours. The rain being from the north-west melted the snow with great rapidity, causing an immense rush of water into the lake, the level of which rose over two feet in the two days.

– 181 –

While watching the water rising on the morning of the 18th, I observed that a mark which I had placed on the shore became rapidly covered and shortly afterwards exposed again, the surface of the lake being at the time perfectly calm, not a ripple ruffling it.

As I found it difficult to account for this, I made more careful observations for upwards of an hour, the result of which was that I ascertained there was a rise and fall of three inches in the level of the water at intervals of five minutes, which was maintained with perfect regularity. This rise and fall was extremely gradual, so that it was quite imperceptible on the smooth surface. After 1 p.m. a breeze sprang up, preventing a further continuance of my observations. I was unable at the time to account for this disturbance satisfactorily, but was inclined to attribute it to a slight earthquake, possibly too faint to be noticed.

In conversation with Mr. Worthington, the Meteorological Observer, the next morning, I mentioned what I had remarked. He informed me that he had himself noticed the same rise and fall on a larger scale, after one of the heaviest floods experienced; consequently it at once pointed to the floods being in some manner the cause, though at first sight it seemed impossible for any flood to have such an effect on a body of water nearly fifty miles in length and of great depth.

Having repeatedly thought over the matter without being able to account for it in a satisfactory manner, I put together a few facts relating to the natural features of the lake with reference to this flood, in order to assist me in coming to some conclusion, which I therefore beg to suggest to you. I have also tabulated the effect of this flood in conjunction with some that have occurred in the European Alps, in districts bearing similar features, the information being taken from Beardmore's “Manual of Hydrology.”

Lake Wakatipu is nearly fifty miles in length, and varies between one-and-a-half and three-and-a-half miles in breadth, its area scaled from the map being 113 square miles. Its drainage area is about 1,200 square miles, principally at the northern extremity, where the water-shed is the main range of Southern Alps drained by the rivers Dart and Rees. These two rivers flow into the extreme northern apex of the lake, having a drainage area of 400 square miles or a third of the whole commanded by the lake, including the portion covered with perpetual snow and glaciers.

The rise of two feet in the level of the lake means 1,823 cubic feet of water per minute per square mile impounded. At the same time the river Kawarau at the outlet was discharging 500 cubic feet per minute per square mile based on the rainfall for the first sixteen days in November. This makes a total of 2,323 cubic feet per minute per square mile run off the drainage area

– 182 –

of the lake during the 17th and 18th November. This quantity is equivalent to a depth of 1.439 inches per diem run off the whole drainage area.

The observed rainfall at Queenstown during the two days being 2.27 inches or 1.135 per diem, of which we can only count upon three-fourths as having run off (this being the proportion observed to do so in similar districts in Europe) we can only attribute .851 inch over the whole area to rainfall. Consequently the difference must be derived from the melted snow and ice running from the mountains.

I may observe here, that Queenstown being nearly at the centre of the drainage area, the registered rainfall may be taken as a fair average; and that at the late period of the season when this flood occurred, the snow on the mountains immediately surrounding the lake had almost entirely disappeared.

The following table will give the above results more clearly:—

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Cubic feet per minute per square mile. Cubic feet per minute. Equivalent depth run off per diem. Inches.
Rainfall 1,373 1,647,600 .851
Snow 950 1,140,000 .588
Total 2,323 2,787,600 1.439

From the above figures we can approximate the proportion of the flood-water running into the head of the lake by the rivers Dart and Rees:—

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Sq. Miles.
Rainfall 1,373 × 400 = 549,000
Snow 950 × 1,200 = 1,140,000
Total, Cubic Feet per Minute 1,689,000

As in the former table it was shown that the total flood-water amounted to 2,787,600 cubic feet per minute, it is evident that considerably more than half the whole quantity rushed into the lake at the extreme north end, or head as it is termed. Accepting these to be the facts, the question is,—Would this mass of water flowing rapidly (possibly in much less than forty-eight hours) into one extremity have sufficient effect to cause the disturbance observed?

I have been unable to find any other explanation, and believe that the following are the reasons. I would ask you to examine the conformation of the lake, and observe that it is throughout a long and narrow one, and that opposite Queenstown there are two bends or elbows of more than right angles; in point of fact, a wave passing down the lake would be deflected

– 183 –

about 100° from the axis of its course, about eighteen miles from where generated, in order to pass round the first of these. This, therefore, is not probable. The result would be, I think, different, and would possibly account for the pulsations noticed.

We will presume that this large body of water in passing into the lake created, while endeavouring to find its level, a gentle swell or wave which, from the conformation of the lake, could only be propelled in the direction of its length; this, upon arriving at the first bend would impinge against the southern shore, and instead of passing round the bend would be reflected back towards the northern side of the lake, and so create an oscillating wave which, upon reaching Queenstown Bay, would rise and fall at interval corresponding to the time occupied by the wave crossing backwards and forwards.

Beardmore, the hydraulic engineer, while describing the effect of tidal disturbances in rivers, remarks,—“When the reaches of the rivers are straight the bore travels evenly up the river; but at turnings it is thrown off towards the further side, where it rises higher than in the straight reaches. Thence it recoils and impinges upon the opposite shore, and so, like a disturbed pendulum, it oscillates from side to side, and only regains its steady course when the reaches lengthen.”

Were the shores of the lake flat and sloping, with the depth of water gradually shoaling off, a wave of this description would be carried by its impetus up the slope, consequently rapidly parting with its energy. The shores of the lake are, on the contrary, almost perpendicular rocky cliffs, with deep water close up to them, thus assisting the transmission of an oscillating wave.

The width of the lake opposite Queenstown is about three miles, but a wave as before described would, in consequence of travelling a diagonal course, considerably increase the distance. On the accompanying sketch of the lake (Plate XIV.) I have indicated what I consider the probable direction that this wave would take, which in crossing opposite Queenstown measures about five miles, or ten miles during each pulsation; and having observed the intervals to be five minutes, it naturally follows that its velocity would be 120 miles per hour. Assuming this to be the case, according to Professor Airy's formula, it would necessitate a depth of 1,000 feet.

Soundings have been taken, * and the greatest depth off Collins Bay is given as 1,296 feet, the bottom rising gradually towards the head of the lake, so that I assume the depth of that portion between the two bends to

[Footnote] * Hector: Report Geol. Surv. Otago, Prov. Council Papers, 1864, p. 86; and “Trans, N.Z. Inst.,” vol. II., p. 373.

– 184 –

be between 1,000 or 1,200 feet, which is sufficiently near to support my theory. I do not attach any scientific importance to what I have described, though, there being reasons for every function of nature, it is both interesting and our duty in the cause of science to seek for and obtain an explanation of those reasons.

While considering the magnitude of the floods in these mountainous districts, I should like to draw attention to a fact that I have no doubt is well known to most of you, namely, the beneficial effects that these lakes exert in restraining the rush of heavy floods. They act as reservoirs, in which are stored up the enormous bodies of water pouring off the precipitous slopes of the mountains, gradually allowing it to find its way to the sea in restrained quantities. They also do good service in arresting the shingle and debris washed off the hills, and carried into the torrents by glaciers and land-slips.

Taking the case of the Molyneux, or, as it is generally termed, the Clutha, it requires no great stretch of imagination to picture to oneself what the aspect of the lower valleys would have become were it not for the influence exerted over the floods by Lakes Wakatipu, Hawea, and Wanaka. These valleys would, in all probability, have been deserts of shingle and sand where not water.

The Rivers Rangitata, Rakaia, and Waimakariri, in Canterbury, have few lakes on their tributaries—the former and the latter none of importance whatever. The nature of their beds is well known, and requires no description by me. They are a continual source of anxiety to the settlers in their proximity, as the flood-channels alter their course during each successive flood, inundating and destroying the land near them, and costing large sums of money annually in endeavours to restrain and control them.

There are many other smaller rivers in both islands having somewhat the same characteristics, and which almost annually do great harm both by flooding and encroaching on the cultivated land near the banks; consequently the question of how they should be treated, in order to regulate and control them, becomes more and more serious, as it is, I believe, an established fact that the high floods are becoming still higher, as well as oftener repeated.

We need not go far to seek for an explanation. The rapid destruction of timber and brushwood along the banks will give rise to encroachment, as on the Hutt River. The extensive bush fires on the ranges, and even the felling of timber for use, allows the heavy rains to flow more quickly off the surface into the streams, as, consequent upon the destruction of the larger trees, the smaller ones perish. As a bush country becomes settled and the timber cleared, so will the floods become more violent in their nature; and,

Picture icon

Probable direction of wave.
Boundary of drainage area

– 185 –

what will eventually prove more serious, should the destruction of timber not be checked, many of the streams will dry up during the summer months.

The Commission of Italian engineers which has lately investigated the cause of the recent terrible floods in the River Po, reported that during the present century these floods have been progressively increasing in height; that the expedient of attempting to confine them within the channel of the river by continual additions to the height of the artificial banks, has been considered inapplicable, as being a method tending to increase the very dangers it is intended to prevent.

For this reason the object sought by the commission was rather the reduction of the floods themselves, or at all events the arrest of their increase. This involved the investigation of the influence on the volume of the river of the denudation of the growth of wood on its banks; the suggestion of legislative measures; the construction of storage basins or lakes to retain the flood-water for subsequent distribution for the purposes of irrigation, together with numerous other matters affecting the river. The result of this examination was that they estimated it was necessary to spend £600,000 in strengthening and restoring the banks, and a further sum of £320,000 for subsidiary works.

Another Commission also lately appointed to report on works necessary to prevent inundation of the River Tiber in Rome, had the following proposed remedies to investigate and report on—


Re-wooding the banks of the Tiber


Storage lakes or basins of reserve


Total deviation of the course of the Tiber


Partial diversion of the water


Limitation of the flow through the city


Rectification of the channel


Additions to the banks and lateral defences.

The Commission decided to increase the discharging capabilities of the two channels existing, by widening and clearing the bed within the city and embanking and regulating the course above it.

These particulars are obtained from the Italian Civil Engineers' Journal, and a writer in the same remarks, with reference to the remedies mentioned in the before-mentioned list, that with regard to the first remedy of rewooding the banks too little is known of the requisite details, and that the result of any operation of this nature would be too slow in its development for it to be relied on as a prevention of flood; as to the second proposal of storing storm water in artificial lakes, the only experience cited in modern times is in the basins of the Upper Loire in the south of France, which works were executed in 1711.

– 186 –

After the disastrous floods of 1846 and 1856 these basins were taken as models for numerous projects, one of which was the establishment of eighty-eight storage lakes of a similar kind on the Loire and its affluents, proposed by Mons. Comoy. One of these basins was commenced in 1861 on the river Furens, an affluent of the Loire near St. Etienne, but before the completion of the work its destiny was altered, and the water, which was retained by a masonry dam 170 feet high, was applied as a motive force.

The greater number of Italian engineers are not favourable to the plan of storage basins. Signor Lombardini, the famous hydraulic engineer, remarks, with reference to the case of the Loire, that the basins proposed in the valleys of that river would have cost £2,600,000 to store a mass of water, the withdrawal of which from the supply would result in the lowering at the point of juncture of the two streams, of the Allier by 2.3 feet, and of the Loire by 3.3 feet, which is an insignificant result when compared with the expenditure required to effect it.

I have quoted these particulars regarding what has been done in the matter of late years in Italy and the South of France, in order to show the importance with which the subject is regarded where the rivers bear very much the same character that they do in New Zealand.

Before undertaking works of this nature here, we should as far as possible reap the benefit of the extensive experience gained in the older countries. Volumes could be filled with examples of the works that have been undertaken with a view of regulating and controlling the floods in rivers, and recording the results; a thorough study of the whole subject is, however, the only manner in which to understand the measures employed, including if possible a personal examination of the works themselves.

It may be said that though these European rivers bear somewhat similar features to those in New Zealand, they differ sufficiently to require a distinct mode of treatment. This is true of all rivers to a great extent, as the slightest variation in the slope of their beds, the different nature of the country through which they flow, the geological features of the mountains in which they rise, and many other circumstances, render it necessary that each should be considered separately and treated differently.

As previously mentioned, one method generally proposed for the protection of river banks is by planting them, which in very many cases means re-planting them, as most rivers have been more or less wooded originally. This method has been practised on some of the Alpine tributaries of the Italian rivers, but I am not aware of the results. It would be a most difficult plan to carry out successfully, and would necessitate the purchase or reserving of the land for a considerable width on both banks, and that it should be fenced in to protect the young trees

– 187 –

from cattle. How much simpler would it have been to have reserved the original bush!

I do not suppose that the banks of the rivers on the Canterbury plains have been ever wooded—certainly not since the settlement of the country—and I doubt if this method of planting would be of any avail in their case, as the impetuosity of the floods and loose character of the banks would be fatal to the growth of plantations. In some places willows have been planted as a protection to the railway banks, and promise to succeed, though had they been introduced on a larger scale there would have been a better chance of ultimate good results. It is only in the lower reaches of these rivers that they break over their banks, as they flow for the greater portion of their course across the plains between well-defined terraces, which gradually die out when the sea coast is approached.

As the slope of these river-beds generally becomes less upon nearing the sea, the consequent reduced velocity of the current affords a better opportunity for trees to succeed if exposed to it. Upon approaching the hills the velocity and consequent strength of the current during floods may be instanced in the case of the late rise in the Rangitata, where square blocks of concrete, nearly two tons in weight, were carried upwards of a quarter of a mile down the stream. In such a position no tree-planting could be of any avail.

The construction of storage reservoirs, in imitation of the lakes alluded to at the commencement of this paper, seems to me the most effectual manner in which to control floods, and which would at the same time arrest the shingle perpetually travelling from the mountains to the sea. This travelling shingle is generally the cause of diversions of the rivers from their proper course. An obstacle, such as a fallen tree grounding in the bed, causes a reduction in the velocity of the current, immediately causing the shingle in motion to deposit behind it; this shingle bank will increase in size till it causes the stream to branch off in a new direction, in many instances to the destruction of valuable land.

In the case of the storage reservoir on the River Furens, one of the latest examples (as previously alluded to), it was constructed of sufficient capacity to impound sixty-five per cent. of the average yearly rainfall. This was a most extravagant manner in which to arrest the floods, though most effectual. It, however, was so constructed for the double purpose of storing water, for supplying power, and for use otherwise in the manufacturing town of St. Etienne situated immediately below it. It was thus made remunerative. The cost is stated to have been £63,600, and it is paying two-and-a-half per cent on that amount. Had it been constructed only for the purpose of regulating the floods in that river, it need only have

– 188 –

been of sufficient capacity to impound the amount of water brought down by a maximum flood of a few days' duration, allowing it to pass gradually down the river in such a volume as the channel might be capable of carrying without injury; after which, when empty, it would be available to repeat the process. Though this plan appears to have been condemned for the purpose on large rivers such as the Loire, it is quite possible that there may be cases where, on smaller streams, subject to sudden heavy floods, it might be carried out with success.

To attempt this plan in New Zealand, where there is little probability of its proving directly remunerative, nature must first be called upon to furnish a site where a storage basin would have a wide-spread area, easily enclosed at some narrow rocky gorge on the river to be treated. Such a site being found, it would (presuming that the fall of the ground be not too great) require a dam of only moderate height to impound sufficient water for the purpose. Unless such natural sites can be found, the cost would be out of all proportion to the results. Earthen dams would be sufficient provided proper discharge tunnels could be driven through rock, they, however, do not answer so well as masonry, where liable to stand dry for any length of time, as sun-cracks and vermin imperil their stability when the next floods are impounded.

The other methods of dealing with the case, such as enclosing the flood waters by enbankments running parallel with the course of the river, enlarging the channels, etc., are the most universally adopted, but are only applicable to rivers with a moderate fall, where the velocity of the current is not too great to destroy them; they are by far the easiest to construct, though from too limited a knowledge of the floods and their effects, and from being usually undertaken piece-meal, without any general and well-considered design, they are liable to failure. The maintenance of this description of work is also very costly, entailing as it must do extensive reconstruction and repairs after each successive high flood.

It is a most essential thing that all such works wherever contemplated should receive great consideration. It is impossible to collect too much information, both regarding the behaviour of rivers when low and in flood, as well as for a complete system of surveys and levels, in order to enable the manner of treatment to be determined. There is no branch of engineering so difficult to undertake successfully.

– 189 –

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

Table of Comparison of Great Floods in European Alpine Lake Wakatipu During the Flood of the 17th and 18th November, 1877.
Locality Drainage Areas, Quantity impounded per square mile. Estimated discharge per square mile. Total quantity run off per square mile. Depth run off per diem. Date of flood Remarks
Areas of lake, Drainage area Total
Square miles. Square miles. Square miles. Cubic feet per minute. Cubic feet per minute. Cubic feet per minute. Inches
Lake of Geneva 208 2,792 3,000 1,120 530 1,650 1.02 17th and 18th Sept., 1840. Observed rainfall at Geneva in 24 hours =2.83 inches.
" " " " " 757 618 1,375 .90 28th May to June 1st, 1856. Do. do. do.,=1.476 in. in 24 hours.
" Maggiore. 77 2,418 2,495 2,550 400 2,950 1.83 21st and 22nd October, 1857 Due principally to melting snow.
" Waktipu.. 113 1,087 1,220 1,823 500 2,323 1.44 17th and 18th Nov., 1877. Averagerainfall observed for two days = 1.135 in. The results are in a great measure due to melting snow.