INTRODUCTION
Whenever purpose of planning is using of stream flow of rivers and also
river regime was distorted by domestic’s people then detection and
sufficient concept of characteristics and situation of low flow in targeted
catchment is vital. This concept must be quantitative which is more important
in urban areas that contagious diseases, chemical and thermal pollutions
are of significant importance. Certainly that water tension is more obvious
in drought and one of consequences of drought is decrease stream flow
of rivers and this make damages in different aspects. For example decrease
of stream flow of rivers cause increasing concentration of pollution and
since decrease in oxygen concentration. This is leading in aquatic mortality
and damage to environment and also to urban, industrial agricultural damage.
On the other hand in using management of storage dams and hydro powers
in drought period, an analyzing of low flow is very important.
Most of study shown estimate of low flow is more difficult of other flows (Henry,
1980). Characteristic of low flow is relevant to catchment’s characteristic
and climate variables. The most characteristic of catchments and climate which
influence on low flow include area, mean of annual precipitation, length of
main channel, percent of lakes and forest area, shape of catchments, girth of
catchment and mean height of catchment (Smakhtin, 2001).
To evaluate relation of low flow with catchment’s characteristic in 23
catchments in west Massachusetts and use 3 parameters such as: watershed area,
average basin slope and base flow recession constant (Vogel
and Kroll, 1992). In one study at 63 hydrometric stations on USA Texas river
due to evaluation of low flow, determine that effective factors of low flow
was drainage area, channel length, slope of catchment, basin shape factor, mean
annual precipitation, predominant hydrologic soil group and the 2 year 24 h
precipitation (Rifai et al., 2000). In another
study in South Taiwan evaluated regression models 34 catchments by Yu
et al. (2002) and they concluded that the most relation of low flow
is with area, slope of catchment. Suresh et al. (2003)
in Himalaya basins of Nepal, in order to study water resources and to estimate
low flow, used of regional analysis method and understand that the relation
between low flow with physiographic and geology in catchment is significant.
In Zimbabwe, Mazvimavi et al. (2004), in a research
on seasonal rivers in order to study low flow on 52 catchments, shown that low
flow has relative relation with mean annual precipitation, slope of catchment,
daring density and negative relation with evapotranspiration. Eslamian
et al. (2004) in order to study low flows in Mazandaran catchment used
of multiple regression to establish relation between low flows and catchment
characteristic and finally introduce 3 factors as important such as: area, average
of height catchment and average of slope catchment. Samiee
et al. (2005) in research in order to regional analysis of low flow
in the 12 hydrometric stations in Tehran find out 4 parameters which include
area, average annual precipitation, average weighted infiltration and average
slope of the catchment.
MATERIALS AND METHODS
Study area: Study area consists of two catchments, Karkheh and
Karoon which are one of Persian golf’s hydrology divisions. Karkheh
basin is located in between 30° 58’ to 34° 56’ north latitude
and 46° 60’ to 49° 10’ east longitude and Karoon basin is located
in between 30° 20’ to 34° 5’ north latitude and 48° 10’ to
52° 30’ east longitude, Fig. 1 shows present study
area. The present research was conducted from 2007 till 2008.
Methodology: With hydrologic plans of study area finally 138 hydrometric
stations were detected but 28 of them which have better condition about
statistical period were selected (Fig. 2).
Calculating of discharge with multi duration: For production necessary
parameters of related discharges (Q_{75%}, Q_{90%}, Q_{92%},
Q_{95%} and Q_{99%}). Firstly flow duration curve was
drew for all of the stations and then with that pertained curves we production
nominated parameters of discharge.

Fig. 2: 
Hydrometric station in the study area 
Factor analysis: In this method we could decrease so many variables
to fewer factors. And in this manner we could produce a summery of main
data. Factor analysis was done for 16 variables in 28 selected stations
in this study, because initially the results of factorial analysis is
very complicated and not lead to a good problem solving so in order to
maximize variance of any factor and simplifying the commentary of factorial
structure, a factorial axis was shown by selection and indicator in each
axis method. Final accepted method in order to establishment of factorial
analysis was using of ordinary data and production of main components
method, which were of similar answer to every of these methods: varimax
rotation, non rotation varimax, qartimax and eqamax methods and finally
varimax rotation method which is a popular one was determined as best
method for selection of any factor. Meanwhile nomination of factors is
also on the based on factor loads of varimax rotation. After selection
of necessary variables, factorial analysis was done on the basis of these
variables which the results shown.
RESULTS
That 5 of this express 80.567% of data variance. And so information is
summaries around these factors. The percentages of every factor are, 29.3,
24.4, 10.9, 9.7 and 6.3, respectively (Table 1). This
means that about 19.4% of whole variance didn’t explained that could
be increased by evaluation of additional variables. In order to varimax
rotation matrix table (Table 2), weighted average slop
of catchment with maximum weight load (0.844) on this factor and explanation
of maximum percentage from whole variance which means the value 29.285,
selected as the first factor.
Table 1: 
Total variance explained 

Extraction method: Principal component analysis 
Also the second factor was the area of catchment
because of weight load of 0.941 and percentage of 24.374 for variance.
And then three other factors such as average elevation, compactness coefficient
and slope of main channel was determined as third, fourth and fifth factors.
Multiple regression models: In regional analysis multiple regression
method, is usually used for gain of revelation between flow characteristics
and catchment’s characteristics and some models introduce for flow
estimation (1, 2 and 4). For accomplish regression, selected parameters
which mentioned in Table 2 as an estimating variables
and discharges of flow duration, main variable evaluated with two scales
such as cubic meter per second (m^{3} sec^{1}) and millimeter
(mm) and two simple logarithmic scale in analysis and finally with this
method we achieve 60 models for 5 parameters discharge with different
durations. Them finally we understand that logarithmic models with cubic
meter per second (m^{3} sec^{1}) scale were the best
and eventually we submitted one model for every evaluating duration (Table
3).
Comparing calculated with observed discharges: With use of gained
equations, as finial models, discharges with duration of Q_{75%}Q_{99%}
of time for 28 stations of Karkheh and Karoon catchments which gained
from the model, we compared them with discharge flow duration statistics,
then we got that they have significance of 99% and there relation coefficient
between 7779%. Result are shown in Fig. 37.
Table 2: 
Rotated component matrix 


Fig. 3: 
Predicted discharge versus observed discharge values
for Q_{75%} 

Fig. 4: 
Predicted discharge versus observed discharge values
for Q_{90%} 
Evaluation and accuracy test model: Efficacy of models in assessment
and approving were calculated with use of R square indexes, standard error
and efficacy coefficient. In Table 3 shown calculations of
indexes above with selected models. Whatever Rsquare is higher, standard error
is lover and efficacy coefficient is more near to one, shows that a model is
better.

Fig. 5: 
Predicted discharge versus observed discharge values
for Q_{92%} 

Fig. 6: 
Predicted discharge versus observed discharge values
for Q_{95%} 
Table 3: 
Final model for low flow by multivariate regression
in the Karkheh and Karoon basins 

SE: Standard Error; CE: Coefficient Efficacy; R^{2}:
R square; I: Permeability, BFI: Base Flow Index, H_{min}:
Minimum of Elevation, A: Watershed Area 

Fig. 7: 
Predicted discharge versus observed discharge values
for Q_{99%} 

Fig. 8: 
Predicted discharge versus observed discharge values
for Q_{75%}( Accuracy test) 

Fig. 9: 
Predicted discharge versus observed discharge values
for Q_{90%} (Accuracy test) 

Fig. 10: 
Predicted discharge versus observed discharge values
for Q_{92%} (Accuracy test) 

Fig. 11: 
Predicted discharge versus observed discharge values
for Q_{95%} (Accuracy test) 

Fig. 12: 
Predicted discharge versus observed discharge values
for Q_{99%} (Accuracy test) 
In addition to mentioned parameters for evaluating gains models, we
used of 9 stations parameters but not from 28 stations which were studied and
didn’t inter free in regression analysis. For this purpose we calculated
7599% of time of discharges of flow duration, with the aid of equations which
shown in Table 3 and parameters of 9 catchments that mentioned
before and then compared with observed discharges of the same stations which
finally we understand that they have 99% significance and a relation coefficient
between 95 to 95%. Result are shown in Fig. 812.
DISCUSSION
Physiographic characteristics of catchment is always one of effective factors
which has been mostly focused in earlier studies by Eslamian
et al. (2004), Samiee et al. (2005),
Vogel and Kroll (1992) and Suresh
et al. (2003), but in this research in addition to physiographic
parameters used climatic and geologic parameters as well. Finally in this research
we used of 21 parameters to selected effective factors on low flows, we also
we used of two parameters such as Base Flow Index (BFI) and Recession Constant
Index. Which we can’t see these 2 parameters in other Iranian studies.
BFI is as an indicator to catchments out come and its own specific qualification,
we can compare flow characteristics is different catchments with the aid of
these index. Finally we introduce 5 parameters as effective factors on low flows
which include: area, average slope, average height, coefficient gravilluse and
slope of main channel.
Generally a most of researches the characteristics of area, average slop and
average height include catchment area more important than others (Eslamian
et al., 2004; Samiee et al., 2005;
Smakhtin, 2001; Vogel and Kroll, 1992;
Rifai et al., 2000; Suresh
et al., 2003; Yu et al., 2002), but
in this research in addition to these parameters the factors such as: coefficient
gravilluse and slope of main channel was also introduced as important parameters.
Which we can say that although main height of catchment doesn’t has direct
revelation with the low flow but it could illustrating how the catchment is
snowy and also we could explain the cause of addition of the factor main slope
of catchment with effect of average slope on daring speed (Eslamian
et al., 2004; Dingman and Lawlor, 1995; Vogel
and Kroll, 1992).
ACKNOWLEDGMENTS
We thank Trabiat Modares University (TMU), (http://www.modares.ac.ir)
and we thank Soil Conservation and Watershed Management Institute of Iran
(SCWMI), (http://www.scwmri.ac.ir),
we thank of our wives whom support us every time. We thank of Dr. Mohammad
Ashrafi (MD) whom helps us translation of our English.