FEBRUARY, 1931 MONTHLY WEATHER REVIEW 73
experienced. Kerosene lanterns froze up at -5 5 O F.
(-48O C.) and dry-cell flashlights were rendered useless
also. Self-generating dynamo flashlights furnished the
only source of light for outside use during the severe cold weather.
I n conclusion it may be well to emphasize the fact that there is still an enormous field for ineteorologicd research
in Antarctica. Professor Hobbs has long pointed out
the great value which would accrue from a year's ob- servations on the polar plateau ; such work would be
possible although attended with extreme difficulties and
hardships to the personnel involved. Simultaneous
records at a number of points over the continent are also
needed before a really comprehensive and intelligent study may be made of the laws which govern atmospheric
act,ion in the southern latitudes. Each expedition has added a bit more to this knowledge yet the field of operations of one or two isolated parties has necessarily
been limited. The estent and rapidity of future re- search hinges, as heretofore, upon the generosity of the
people and of organizations in offering financial -support
for expeditions.
REPORT OF THE STREAM-FLOW PREDICTION SUBCOMMITTEE'
By A. STREIFF, Ccmiiltiiig Eugiiie~r
[Jackson, hIich.1
The past year was c,haract,erized by a greatly increased universal interest in the possibilit,ies of e,stiniat8ing stream
flow, which previously has been and, in inany professional
and scient>ific. quarters, still is re.garded as being wholly a fortuitous seque,nce. During the past dec,ade this view-
point had gradually undergone a radical change, until
to-day many earnest investigators recognize the presence
of definite syst,ematic elements in st,ream flow which per-
niit conclusions to be drawn as to future run-off well in
advai1c.e of occ,ui~e,nce.
PAST METHODS O F PREDICTION
In the planning of hydraulic projects, the manner of
estimating future discharge quantitie.s in the past has rested on a rathe,r questionable basis, unte.nable on scien-
tific grounds, and dictated by necessity rather t8han se.cure.,
basic knowledge. If streani flow is to be regarded as
wholly fortuitous sequeme, then the theoiy of probability
is directly applicable to the est8imate of future perform-
ance. Much has been theorized on future probability (probability a priori) based on past performanc,e or e,xperi-
ence (probability a posteriori) or so-called empirical determination of probability. It is a well-established fack
that only a voluminous series of data can furnish a secure
basis for future expectation. Such data must, moreover,
be wholly freed from systematic sequences. Instead of thousands of observations, hydraulic engi- neers usually have a t their disposal only a few records covering not more than 10 to 50 years. It is evident that
50 dice throws will not give the same average as several
thousand, and if systematic c.hanges are introduced, si1c.h as changing the throw, the dice, etc., no reliable estimat'e
of future averages based on previous performance is possi-
ble. Yet, this is the meager basis on which all hydraulic projects hitherto have bee,n based. The future is held to
have the same average, as well as limits of vaiiat,ion, as
the past records indicate.
N E W METHODS OF PREDICTION
Certain it is, that stream flow records never repeat
themselves, and ceaseless fluckuations exist, continually moving to higher and lowe,r levels apparently in often recognizable systemat,ic sequences. In Europe these have
1 Extracted from the 1929-30 Report Hydraulic Power Cpmmittee, Engineering 6eC- tion, National Electric Light Association, Great Lakes division. Presented at the tenth annual convention, French Lick, Ind., Sept. 2,527, 1930.
51079-31-2
been studied and applied by Dr. Axel W a l h (Twelve Ye.ars of Long Distance Prognosticat>ion of Rainfall and
Wnterlevels, Annals der Hydr. und Mnrit,im. hleteor-
ology, September 1926, pp. 89-92), direct,or of the hy-
drographic service in Sweden. Here, t800, muc.h research has 1iee.n applied to finding a hett,er wity of estimating
fut,ure stream flow. It is the, near fut>ure which is of the greatest importance in liydrnulic power projects, and t8his
may be radkally different froni the immediate past;
enough t80 cre.ate the differenc.e between earned or not
earned interest on outstanding bonds.
UTILITY OF PREDICTIONS
Esperieme of the past 10 years indicates that public utilit,ies which derive their power supply in whole or in
part froni hydroelect'iic plants can apply these studies
profitably to the appraisal of their future power supply.
It appears that for the Great Lakes division it is quite
possible to e,st)imate hydraulic, power output a year in ad-
vanc.e to within 5 per cent. The general trend can be
forecmt for seve.ra1 years to follow, and thereby the ste,aIn power and coal supply re,quire,ments can be bud-
geted more accurately. No errors need be made as by
one utility, which constructe,d an espensive booster pump installation for their circulabing water just in advance of
t,he rise in levels of the Great Lakes, or by another in the
Great Lakes region, which hurriedly installed expensive
additional boiler capacity to take care of threatening shortage just in advance of a rising hydraulic power out-
put.
~
STREAM FLOW I N GREAT LAKES REGION
The Great Lakes region appe.ars to be dist>inguished by
a singularly regular niultiannual sequence of stream flow, which enable,s close estimates of wat,er power. Naturally
such estimate,s are of greater value, t'he greater the amount of available storage. Without storage the an-
nual (seasonal) variation determines the requirements of ste,ani power; peak capacity is not affe.cted by a variation
of t,he annual mean, although even in such cases the
annual coal supply is still subject to calculation in ad- vance. Members of the Gmat Lake,s division are enabled
to utilize the knowledge of these systematic. sequences of st,reaiii flow in the Great Lakes region to the estent above
indicate,d. The discovery of this regular c.ycle in stream flow is due to the hydrologist, Robert E. Horton (United States
74 MONTHLY WEATHER REVIEW FEBRUARY, 1931
Geological Survey Water Supply Paper No. 30). Since
1875 the cycle has faithfully recurred 10 times. At pre,sent a minimum is approached whic,h will occur in
1931. A further maximum will occur in 1935, and a high
niaximuni in 1940. The hydroelectric coiiipanies in the
Great Lakes region will profitably note that. nest year d l
be a poor water year, and their hydro output will c,onipa,re,
with the previous low around 1925.
On the other hand, abundant hydro output, will he
a.vailable in the Great Lakes region aronnd 1934-1936.
STREAM FLOW IN MICHIGAN
For Michigan the graph shown in Figure 1 is iliuminat- This graph represents the flow of the Muskegon
It may be seen that the relation with ing. River in Michigan.
It should be mentioned that the lower peninsula of hlichigan has, on account of its heavy cover of glacial
drift, a very great volume of ground storage, which favors the exact realization of these estimates. Next year a riiininiiim output of hydro iw Michigan will occur, the
deficiency being as high as 37.5 per cent, of the 1928 hydro
011 tput. In a personal letter to the editor, hlr. Streiff comments
on the methods of stream-flow prediction as follows:
JACKSON, MICH., Juikunry 66, 1931,
DEAR PROFESSOR HENRY: Hydraulic engineers thus far have
liceii i n a quandary. In a contineiitnl climate, asthot of theunited
States, the water problem is, for great 1J:wtS of the country, that of
n deficiency. The economics of projects costing niillions depend
on oil accurate estimate of the quantity of 11 ater xvliich is available, or will be, in the future. The theory that future averages will be equal to past averages has led to fiiraiicial failures in a number of
FIOURE I.-Stream-flow prediction for Muskegon River, Mich.
the ma.Gma and minima of the Wolf numbers is striking.
periodical Power of April, lg29- The dotted line shown
thereon is the probable extension of the annual mean stream flow as it appeared to the chairnlan of this subcorn-
mittee in the year 1928. The dotted line drops sharply,
indicating yea]% in lg30 and lg31* The present
drought fully supports the estimate made two years ago. The estimates of power output for the Consumers Polver
co. have checked as follows: F~~ 1928,
for 1929, error of the
year exactly corresponded with the estimate. Respectfully yours, A. STREIFF.
cases. I have personally been involved in two of them during the
and while we believe we are on the right path, this is strictly an engineering belief. Engineers are in every branch striding far ahead of scientific certainties and proceed on “opinion,” “judg- ment,” “hunch,” and so forth. Hydrodynamics is still in a very
elementary stage a6 far as the possibility of computing flow of water is concerned, but still engineers build water turbines of 93 per cent efficiency, largely empirically. It
5 per cent; is a technical expedient. I fully disclaim its scientific value. And it should be regarded as wholly in the field of hydraulic engineering.
I really feel very little informed about meteorology.
This graph was prepared in 1928 and first published in the p a ~~~~~~h a v e beell forced to cast abollt for other solutions,
Hence the report should be regarded from that standpoint.
per cent; for l930 the first