|
<br />B34
<br />
<br />DROUGHT IN THE SOUTHWEST, 1942-66
<br />
<br />All streams in the Southwest are subject to summertime flash
<br />floods, but these are usually highl~' local and ephemeral phe-
<br />nomena, which nOl'lUuIl;y do not play an important part in the
<br />total annual runoff of the Gila, Salt and Verde Rivers. For,
<br />though the summer rainfall amounts to about half the annual
<br />total, its translation into usable runoff is very inefficient as
<br />('ompul'ed with winter rainfall. 'rhus, the fluctuations in the
<br />water-year data >1< >lo >I< are largely governed by the winter
<br />storms, and so may be safely compared with fluctuations in tree
<br />growth.
<br />Both the, Gila and the Salt are very nearly in the class of
<br />ephemeral streams. Draining regions where great extremes in
<br />seasonal weather are the rule, they can show fiuctations of even
<br />greater amplitude in the annual runoff in successive years. In
<br />very wet years the runoff reaches relatively great extremes" for
<br />when the characteristically larA'e losses in evaporation, tran-
<br />spiration, and other Vro('es~es are met, every inch of excess rain-
<br />fall is increasingly effective. Of course, many details affect this
<br />general relation, sucll as the distribution of storms during the
<br />year and the frequency of various intensities and duration of
<br />storms.
<br />Thus, wet wiuters, such as in 1905, lead to an exaggeration
<br />in runoff * * *. J..Iag effects also appeal' for these years, so
<br />that the trees fail to indicate extremes in runoff such as occurred
<br />in 1905 and 1915; there is evident, however, it general corre-
<br />spondence in lllaxima and minima of growth and runoff, as
<br />well as llluch agreement in the details of fluctuation for most
<br />of the years. 'Vhen the trees show a persistent maximum in
<br />growth, as in the late 1860's, it is probable that one or more
<br />years of extremely heavy runoff occurred. On the whole, how-
<br />ever, as elsewhere in the Southwest, the tree curves give the
<br />drought years with greatest fidelity.
<br />
<br />Schulman has emphasized the necessity of using only
<br />ring series from living trees for comparison with pre-
<br />cipitation or runoff records, because the validity of such
<br />comparisons depends entirely on the relative width of
<br />the rings, The indices used for hydrologic comparisons
<br />are, therefore, based almost entirely on ring records de-
<br />rived from living trees, thus avoiding the uncertainties
<br />of indices derived by combining records overlapping
<br />in time.
<br />
<br />RESULTS OF CORRELATION STUDIES
<br />Of the large number of tree-ring records available
<br />in the Southwest, some are for geographic areas for
<br />which records of natural streamflow are short, meager,
<br />Or completely lacking. On the other hand, tree-ring
<br />indices are not available in some arerus where long
<br />records of streamflow are available, as for example in
<br />Texas and eastern New Mexico, Tree-ring indices are
<br />available to represent several of the regions where
<br />streamflow records have been found to have a high de-
<br />gree of homogeneity-as for example southern and
<br />central California, the high plateaus of southwestern
<br />Utah, the San Juan Mountains of southwestern Colo-
<br />rado, the headwaters of the Gila River in New Mexico,
<br />and the middle Gila River basin and Mogollon Rim of
<br />Arizona. In addition, two indices appear to be excel-
<br />lent representatives of the upper Colorado River basin.
<br />
<br />One test of the agreement between the tree- ring index
<br />and the runoff from an area is the product-moment
<br />correlation of the two. Schulman (1956, p. 4'7) shows
<br />the correlation between tree-ring indices and yearly
<br />runoff for 9 streams, of which 3 are in the Southwest.
<br />Comparing Colorado River at Lees Ferry, Ariz" with
<br />the Douglas fir series in the period 1895-1950, he found
<br />coefficients of correlation of 0,'73 on a yearly basis and
<br />0,84 on a 3-year smoothed basis; using the pinyon series
<br />for the period 1895-1948, he found that the correspond-
<br />ing coefficients were 0,51 and 0,69. For Rio Grande
<br />near Del N orte, Colo" in the period 1890-1950, he found
<br />coefficients of 0,60 on the yearly basis and 0.43 on the
<br />3-year smoothed basis, For San Gabriel River near
<br />Azusa, Calif., in the period 1896-1950, the coefficient
<br />was 0.6'7 on the yearly basis and 0,88 on the 3- year
<br />smoothed basis. Correlations giving coefficients within
<br />the same range were found for streams studied during
<br />this investigation, For example, correlation on a yearly
<br />basis, for 1891~1951, of the upper Rio Grande tree-
<br />ring index with the runoff of the following streams
<br />gave these coefficients of correlation: Animas River
<br />at Durango, Colo., 0.62; Conejos River at Mogote, Colo.,
<br />0,81; and Rio Grande at Otowi Bridge near San Ilde-
<br />fonso, N, Mex" 0,'71.
<br />For this report correlations were made between yearly
<br />regional runoff and the yearly tree-ring index con-
<br />sidered to represent the region best, The correlations
<br />were made graphically on logarithmic paper by plot-
<br />ting tree-ring index figures against runoff in standard
<br />deviation units, (See table 3.) The regional correla-
<br />tions, as shown in table 6, gave coefficients that range
<br />from 0.35 to 0,88.
<br />
<br />T ABI..E 6.-Correlation of tree-ring indices with yearly runoff for
<br />hydrologic regions and the standard error of estimate of yearly
<br />and mean runoff based on tree-ring indices
<br />
<br /> .g~ ~ " j-s'SEij -6 ~ ~." '<lfJl
<br /> ~I ~~ ~ ",,0
<br /> - i ".~ ~
<br /> ". B'a~
<br /> ~:a I>. o....l=l~ ,,0_
<br />IIydrologlcregion ~"~I ~ g~ :e "S~~~ 'g'O~", ':E:8.
<br /> .~.g~ ~~<P~'s " ~ o~~
<br /> 'Cl o:a .00:;: ~ :<'~ .@~ig ~"" i
<br /> .~ ~ I>- ~S; !IM~ ~]~~ ~,.c ill ~~
<br /> . .~
<br /> ~ 8 "' '"
<br />(1) (2) (3) (') (5) (6) (7)
<br />Central C~l1fornla____~_ --- - - -
<br />1896-U141 77 0,66 -4lJ-.\}5 11 -16-29
<br />Southern Callfornla_____ 1896-1950 78 .82 -46-86 11 -14-26
<br />Southwest Utah~_~_~___ 1904-50 (') .78 -22-29 8 -8-10
<br />Mogollon Rlm___H_ ,___ 1889--1953 67 .79 -42-74 17 -10-18
<br />Upper Gila River_Hnn 1904-39 M .35 -38--62 12 -11-18
<br />Middle GUa Rlverm""_ 1909-tiO M .45 -52-109 10 -16-34
<br />San Juan Moutainsn_n 1890-1921 70 .83 -22-29 16 -&-7
<br />Ubper Colorado River 1897-1950 51 .71 -15-18 12 -4-5
<br />[\.SIn, 1897-1945 (') .88 -19-23 10 -&-7
<br />
<br />I Yearly mean ring widths from Schulman (1950, table 4-D, p, 14).
<br />2 Yearly mean rIng widths from Schulman (1945, table 5, p. 38),
<br />
<br />The highest coefficient, 0.88, was between runoff of
<br />Colorado River at Lees Ferry and the 1945 tree-ring
<br />series, A major reason for this high correlation al-
<br />
|