|
<br />278
<br />
<br />J. LAVABRE ET AL.
<br />
<br />HYDROLOGICAL RESPONSE OF A MEDITERRANEAN BASIN
<br />
<br />279
<br />
<br />.,'. . .:/
<br />~'':'<\~:
<br />
<br />,4". ...,
<br />:....,;,.~~:'r\'
<br />., .
<br />
<br />.:,\"
<br />
<br />TABLE 2
<br />
<br />,:' \.
<br />::!.ct
<br />
<br />Main statistical characteristics of the 23 year reference period (1967-1989)
<br />
<br />Annual
<br />precipitation
<br />
<br />Annual
<br />runoff
<br />
<br />Annual
<br />losses
<br />
<br />'>':~".
<br />
<br />,i-,: 'i
<br />~',,'-, '(!,;";-f I'~
<br />_ . ,::~\'~A.nk.,;;:.,,_.. ~~". .?~>>:~-
<br />'Jf-~,..-:--o""~ t"", ,r'};.\i"'...... tv~' I~
<br />. --~.~~ti,(~:~ -_-E,,~... .lLy4 ...-:~~;,.~::J" .:6f_~"~'~
<br />,'_,' ~_ _ _.,.~ .......".., A'
<br />_ _~.,.,,-....., ::.:~-r~...,-...~ <::;,'" ," ~ ,",:# -~
<br />.., ""'-. '<.,~<; "'... ~ _ _ _...~'4..r,.....olr-..-'.:' ,.-... ~ ~ V,
<br />.,;;J~~:~r~~t~::;1i7:3:~
<br />
<br />_. ,_,_,' ~_<J...,-+.':J"'f\.~~..:I~)1~;(i~':.1~~~\< '. /
<br />. ~ ~-,:::, '~,._..~ .~_...;;: ~<:;:...._ _~~;;;~",",i' ;._!-~_' ':1-:";..011
<br />...",=:."", , ~-I~. .....:,.Sfl ~~.;;,:..,~~'i:.''';!:'''' " \.,.-~
<br />, . J "V ;,.: ....~~ -f~~,-~;-<.~ (fr......~:e-....A...!!'f.i ..-'..t:.....~ ~ '....,.
<br />"~~ ,..--:- ",,'::'X',~-;"",-"
<br />
<br />Mean (mm)
<br />Standard deviation (mm)
<br />Variation coefficient (%)
<br />
<br />1164
<br />366
<br />31
<br />
<br />626
<br />325
<br />52
<br />
<br />538
<br />67
<br />12
<br />
<br />Annual analysis
<br />
<br />.:;:~:- -
<br />.;;, ,\,
<br />
<br />A hydrological year can be simply described by the annual precipitation
<br />(AP) and by the annual runoff (AR). We can define the annual losses (AL) as
<br />
<br />AL = AP - AR
<br />
<br />(I)
<br />
<br />(a)
<br />
<br />Now, our purpose is to set a prediction of the AR for 1990 based on the
<br />reference period and compare it with the observed AR. Table 2 shows the
<br />main statistical characteristics for the 23 year reference period, 1967-1989.
<br />The relatively high rainfall is a consequence of the proximity to the sea and
<br />the orographic role of the massif in the genesis of precipitation. The small
<br />variation coefficient of the AL shows that they are relatively stable around the
<br />mean value. This suggests that the annual variability of the AP is practically
<br />absorbed by the AR, and that the AL are little influenced by the pluviometry.
<br />So, given that the geological structure of the basin allows us to ignore the
<br />existence of deep losses. the AL may be considered as an estimate of the real
<br />evapotranspiration. (Note that the annual potential evapotranspiration
<br />measured by Colorado bac is 1155mm, see Table 5.) On the other hand, on
<br />the basis of the AL stability, we can propose that a linear regression could be
<br />a good tool to relate the variability of AR to that of AP. Thus the established
<br />eqn. (2) will be used to predict the AR for 1990.
<br />
<br />AR (mm) = 0.88AP (mm) - 396 R' = 0.98; a", = 51 mm(2)
<br />
<br />The high correlation coefficient between AR and AP (R 0.988) confirms the
<br />good quality of the equation, which accounts for 98% of the initial AR
<br />variance and that has a residual standard deviation (urcs) of 51 mm. It is worth
<br />noting that the reference period covers a wide range of situations. thus
<br />increasing the robustness of the relation. Table 3 shows the extreme values of '
<br />AP recorded during this period. illustrating this statement and, at the same
<br />time, the well known high variability of the precipitations characterising the
<br />Mediterranean climate.
<br />As a consequence. we will expect that this equation will be valid in the case
<br />
<br />
<br />(b) i
<br />Fig. 2. The Rimbaud pluviograph: area (a) before and (b) after Ihe fire,
<br />
<br />..
<br />
<br />
|