Académique Documents
Professionnel Documents
Culture Documents
Abstract:
Baseow in the Andes is commonly considered to be related with the release of water stored in pramos. Pramo is the
predominant ecosystem above 3500 m a.s.l. and is characterized by a rainy and cold climate with low evapotranspiration.
However, this baseow concept is based on hydrological process studies in small Andean catchments of a few square kilometre
with a homogeneous land cover. Middle-sized Andean catchments, like the subcatchments of Tarqui and Yanuncay, Ecuador, are
rarely homogeneous or uniformly covered by pramo. The objectives of this study are therefore to investigate baseow
characteristics in heterogeneous Andean catchments and to identify relationships between baseow processes and physical
characteristics such as storage and recharge. Hereby, the contribution to baseow of pramo and other sources such as alluvial
aquifers is quantied.
This study uses nonlinear recession analysis, physically based lters and digital lters for comparison of baseow of
neighbouring but distinct subcatchments. The Yanuncay subcatchment shows a clearly different storage capacity and recession.
The storage capacity of Yanuncay is 50% higher than for Tarqui because of its higher coverage of pramo. On the other hand,
considerable storage capacity has also been found in the Tarqui subcatchment, which has a limited pramo area but a signicant
alluvial aquifer. It is shown that improved understanding of the specic baseow characteristics such as storage and recharge and
its relationships to the heterogeneity of the land cover in Andean catchments will lead to a better assessment of the water
resources and give new insights for effective management actions. Copyright 2014 John Wiley & Sons, Ltd.
KEY WORDS baseow separation; Andean catchment; pramo; mountain hydrology; Ecuador
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
BASEFLOW COMPARISON BETWEEN TWO ANDEAN CATCHMENTS, ECUADOR
analysis and contributes to the hydrological process The particular physical conditions of the two studied
literature with new insights in baseow characteristics subcatchments lead to differences in hydrological
of highly variable Andean catchments. behaviour. Thus, comparative analysis of both
subcatchments may explain differences in baseow
Study area characteristics.
The Paute River catchment is an Andean mountain
catchment situated in the southern part of Ecuador. It is Perceptual models
characterized by a high spatial variability of altitude, In order to illustrate the different physical conditions in
precipitation and temperature (Clleri et al., 2007). the subcatchments, three perceptual models have been
The adjacent Paute subcatchments of Tarqui and constructed from eld observations, data analysis and
Yanuncay (Figure 2) show a clear contrast in topography. previous studies (Beven, 2012). In these, the possible
Tarqui has a relatively at topography in the valley below pathways in which surface water and groundwater move
2700 m a.s.l., whilst Yanuncay has plateau areas at about through the Tarqui and Yanuncay catchment were
3500 m. Both subcatchments have a discharge station at identied (Figure 3).
their outlets just upstream of their conuence. It is hypothesized that the hydrological response of
Hydrogeologically, the Tarqui valley is characterized by Tarqui can be explained by the concurrence of the three
an alluvial aquifer in the lowest part of the subcatchment. models (A, B and C), whilst for Yanuncay mainly model
On the other hand, the neighbouring subcatchment of the A, with a secondary role for model B, is relevant.
Yanuncay River has a narrow valley without evidence of The perceptual model for the pramo dominated upper
an aquifer. Yanuncay has substantially more pramo areas catchment (A) is based on the concepts presented by
as compared with the Tarqui subcatchment (Figure 3). Buytaert and Beven (2011). The prevalent physical
The at valley of Tarqui represents an attractive place conditions are highly saturated soils and low evapotrans-
for urban expansion because of its limited distance from piration as indicated in Figure 3. The hydrological
the city of Cuenca, which has 500 000 inhabitants. The response of pramo is thought to be composed of a
Tarqui subcatchment will be potentially affected by combination of overland and shallow lateral subsurface
mining exploitation located in the pramo areas near its ow. As Yanuncay is mainly covered by pramo, the
watershed divide. Land use change by urban expansion expected response should predominantly follow model A.
and potential mining activities represent potential hazards Recharge to deeper groundwater is not likely in pramo
for proper water resources management. Better knowl- areas because of impervious bedrock at shallow depth,
edge of the hydrological processes is necessary to but a slow response is caused by the combination of a
mitigate and regulate these threats. high water holding capacity of the soil and a low
Figure 2. Location and elevation of Tarqui and Yanuncay subcatchments, the available meteorological, river level and precipitation stations are indicated
as well
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
P. GUZMN ET AL.
Figure 3. Perceptual models for the hydrological processes for upper, middle and lower part of the Tarqui and Yanuncay subcatchments in function of
their relief. The different processes are the following: evapotranspiration (ET); inltration (Inf); recharge (Rech); hillslope runoff (Qhill); overland ow
(Qo); lateral subsurface ow (Qi); and low ow (Qs). Wt is the water table
DATA
A digital elevation model with a 5 5 m spatial resolution
was available (Table I). The subcatchments have a similar
mean, standard deviation and maximum slope. Yanuncay
has a more elongated shape with, therefore, a larger
perimeter. Figure 4 compares the altitude in function of
Figure 4. Altitude versus cumulative area for Yanuncay and Tarqui
the cumulative area for both subcatchments. Yanuncay subcatchments
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
BASEFLOW COMPARISON BETWEEN TWO ANDEAN CATCHMENTS, ECUADOR
Figure 5. Monthly average precipitation for Yanuncay and Tarqui stations Figure 6. Monthly potential evapotranspiration (ET0) at two different
from July 1997 till 30 December 2011 altitudes for periods: June 2000 till June 2001 and August 2005 till
November 2012 at 2630 and 3761 m a.s.l., respectively
Both Yanuncay and Tarqui have two precipitation
stations each (Figure 2). The precipitation was monitored 2630 m a.s.l. is 2.6 mm/day, whereas in the pramos
from 17 July 1997 until 30 December 2011 by a tipping located at 3761 m, it is 1.96 mm/day. Buytaert et al.
bucket rain gauge with an accuracy of 0.5 mm (Figure 5); (2006a) obtained similar values (1.90 mm/day) for
other characteristics of the stations are given in Table II. pramos located in the Paute catchment.
The daily precipitation for each subcatchment was
estimated using Thiessen polygons.
For both Yanuncay and Tarqui, two river level gauging METHODS
stations are also available, one central station and one
near the outlet of the subcatchment (Figure 2). The Water balance
derived discharge records are based gauging of water The general water balance for the catchments can be
levels with 5-min interval and a calibrated stage stated as
discharge relationship. Data of precipitation, water level
and discharge have been supplied by the drinking water P Q ET G S (1)
supply company of the city of Cuenca. The discharge
time series covers the same period as the precipitation but where P is the precipitation (mm), Q is the river discharge
shows several gaps (Table II). Both precipitation and (mm), ET the actual evapotranspiration (mm), G the net
discharge data were processed and averaged using the groundwater ow leaving the catchment boundary
free software Hydraccess (Vauchel, 2005). without passing the river outlet (mm) and S the storage
Potential evapotranspiration (ET0) is calculated using change (mm) for a given period of time. The actual
the PenmanMonteith equation (Allen et al., 1998) for the evapotranspiration is calculated as
stations located at 2630 m a.s.l. in central Tarqui
(upstream of the outlets) and at 3761 m a.s.l. in the high ET ET0 Kv Ks (2)
Tarqui (pramo) very close to the Yanuncay border
(Figures 2 and 6). The average ET0 in the lower valley at with ET0 the reference crop evapotranspiration in
Table II. Characteristics of precipitation and river level gauging stations in the Tarqui and Yanuncay subcatchments
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
P. GUZMN ET AL.
millimetre, Kv a vegetation coefcient and Ks a stress The recession rate of baseow is studied during periods
factor. To take into account the spatial distribution of the with little or no precipitation (Tallaksen, 1995). In this
ET in the subcatchment, a weighted value of ET has been research, the starting criterion to select the moment for a
proposed: depletion period was initially set to ten consecutive days
without precipitation. However, this criterion was too
Ap Al
ET ET0p Kvp Ksp ET0l K l (3) strict and hence too few recession periods were found.
AT AT Therefore, the rule was relaxed to 10 days with an average
daily precipitation of less than 0.5 mm/day, which is
with A the area, where the subscript p denotes pramo, l substantially lower than the potential evapotranspiration,
the agricultural land (mainly pasture) lower than 3500 m even at higher altitudes.
a.s.l. and subscript T the total catchment. Kvp and Ksp are Although, recession discharge characteristics are by
0.62 and 1, respectively, and have been estimated by nature nonlinear, Wittenberg (1999) pointed out that
closing the water balance for two pramo because of the easier mathematical formulation, concep-
microcatchments located in upper Tarqui close to the tual models for runoff processes have been predominantly
Yanuncay boundary over a longer period. Studies in linear. Therefore, recession curves are often section by
pramo show values as low as 0.42 and 1.0 for Kvp and section approximated by linear reservoirs with different
Ksp, respectively (Buytaert et al., 2006a). We assume one recession constants (k). If the resulting set of k-values is
factor Kl, which averages the evapotranspiration in time different, there is a strong indication for the nonlinear
and space of the different vegetation types, located below character of the recession. This is especially true when
3500 m a.s.l. they vary in a systematic way according to the stage level
Over a longer period, the terms G and S in Equation 1 (Prasad, 1967; Wittenberg, 1994). The nonlinear storage
can be neglected, resulting in a Kl of 0.70 as estimated by equation is expressed as (Wittenberg, 1999).
closing the water balance. Then, the actual ET is
calculated with Equation 3. S aQb (4)
The altitude is considered as the main factor
inuencing the spatial variation of evapotranspiration. with S being storage in m3, Q discharge in m3/s, and a a
For similar altitude in Yanuncay and Tarqui, the average factor with dimension m33b sb. The exponent b is
evapotranspiration is expected to show the same temporal dimensionless. If b is equal to 1, Equation 4 changes
behaviour. into the well-known linear reservoir storage equation
(Maillet, 1905). Combining Equation 4 with the continu-
ity equation without recharge gives
Baseow assessment
There is no direct way to continuously measure baseow d aQb
(Furey and Gupta, 2001). Only during longer dry periods, Q (5)
dt
pure baseow can be observed. For other periods, baseow
separation methods are used. The majority of baseow integrating this equation results in an equation for the
separation methods do not have a clear physical basis. depletion of a nonlinear reservoir, starting at any initial
However, more recently, Furey and Gupta (2001) and discharge Qo (Coutagne, 1948)
Huyck et al. (2005) developed physically based lters for 1=b1
separating baseow from stream ow. 1 bQ1b
Qt Qo 1 o
t (6)
In order to investigate if baseow shows a distinct ab
behaviour in the contrasting but adjacent Tarqui and
Yanuncay subcatchments, we have estimated and com-
The coefcient a can be estimated by
pared baseow using nonlinear recession analysis,
physically based lters and digital lters.
Qi1 Qi t
a (7)
Nonlinear recession analysis 2 Qbi1 Qbi
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
BASEFLOW COMPARISON BETWEEN TWO ANDEAN CATCHMENTS, ECUADOR
Sivapalan (1999), Wittenberg (1999) and Chapman considering recharge as a delayed response to precipita-
(1999) is equal to 0.5. tion. 1 is the recession constant of the catchment
dened as a non-dimensional parameter, which describes
Physically based lters the rate at which stream ow decreases whilst recharged
Graphical and lter approaches lack a physical base by groundwater (Vogel and Kroll, 1996).
(Furey and Gupta, 2001). Comparing the subcatchments, To estimate the parameters , c1 and c3, Furey and
we expect that the different physical characteristics will Gupta (2001) use Equation 12 under the conditions of
produce different baseow patterns. Hence, a physical precipitation (PB; j ; PB; j1 ; PB; jd1 , resulting in
conceptual basis for the baseow separation method is QB; j
desirable. 1 if PB; j PB; j1 PB; jd1
The physical lter developed by Furey and Gupta QB; j1
(2001) is derived from a mass balance equation for 0 (13)
baseow from a hillside. Details on the development and
tests of this lter can be found in Furey and Gupta (2000,
2001, 2003). The main assumptions of the lter are that QB; j 1 QB; j1
c1 if PB; j > 0; PB; j1
the catchment is composed of n hillsides and that PB; j
channelled water can be routed instantaneously within PB; jd1 0 (14)
the time step of the model from each hillside to the outlet.
Under these conditions, mean baseow and streamow where c2 is calculated using Equation 10, and c3 by
are dened as Equation 9. The specic conditions of precipitation in
Equations 13 and 14 are very difcult to nd in the
qb B; j ni1 qb i; j (8a)
precipitation records of the studied subcatchments.
Therefore, we apply Equation 13 for
QB; j ni1 Qi; j (8b)
PB; j PB; j1 PB; jd1 0:5 (15)
where the subscript B indicates the basin, i the hillside, j
the time step, qb the baseow and Q total discharge. To we assume that during these specic events, the
preserve the mass inuence of precipitation in the discharge is negligible
(Equation 12).
c1 c2 c3 1 (9)
where c1, c2, c3 are fractions of precipitation transformed Digital lters
into overland ow, evapotranspiration and recharge, Alternatively, digital lters are attractive for long-time
respectively. The values of c1, c2, c3 for a hillside are series of data. Arnold and Allen (1999a) indicate that the
similar for a catchment. Over a long period, the net strength of these lters is the reproducibility, although they
change in storage of water in a basin is close to zero, do not have any physical basis. Furey and Gupta (2001)
hence c2, can be calculated as corroborate this idea and add that they are arguably more
objective than the graphical approach and often easier and
j0T
jj0 QB; j faster to implement. Results of digital lters look realistic
c2 1 j0T
(10)
jj0 PB; j because baseow is arbitrarily constrained such that it does
not exceed stream ow nor becomes negative.
The baseow lter BFLOW (Lyne and Hollick, 1979),
where P is the precipitation. The physical lter equations and the Eckhardt lter are based on the idea that high
become then (Furey and Gupta, 2001) frequency waves can be associated with direct runoff, and
qb B;j 1 qb B; j1 low frequency waves can be associated with baseow
(Eckhardt, 2005). However, the high-frequency compo-
c3
QB; jd1 qb B; jd1 (11) nent may represent only a part of the surface runoff rather
c1
than the total surface runoff, so that the physical
interpretation of baseow as low frequency component
QB; j 1 QB; j1 and direct runoff as high frequency component may be
incorrect (Spongberg, 2000).
n11 c1 Pi; j 1 c1 Pi;j1 c3 Pi; jd1 The digital lters of Lyne and Hollick or BFLOW
(12) (Equation 16) and Eckhardt (Equation 17) are integrated
in the web Geographical Information System hydrograph
where subscript d is a delayed time for recharge, analysis tool (WHAT) (Lim et al., 2005).
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
P. GUZMN ET AL.
The BFLOW lter is dened as 11), but the coefcients of physical lters provide the
possibility of explaining processes like recharge.
1
qdt qdt1 Qt Qt1 (16)
2
where, qdt is the ltered direct runoff at time step t (m3/s); RESULTS AND DISCUSSION
qdt 1 is the ltered direct runoff at time step t 1 (m3/s); Hydrograph correlation and water balance
is the lter parameter; Qt is the total streamow at time
step t (m3/s); and Qt 1 is the total stream ow at time Two time series of specic runoff (24 October 19988
step t 1 (m3/s). May 2002 and 4 February 200814 August 2011) for
The Eckhardt lter is dened by Tarqui and Yanuncay were compared. The specic runoff
in mm/day is calculated by dividing the discharge by the
1 BFImax qbt1 1 BFImax Qt area of the subcatchment. The linear correlation assessed
qbt with the R2 Pearson coefcient shows a similar response
1 BFImax
(17) in both subcatchments with correlations of 0.76 and 0.62,
respectively, for 19982002 and 20082011 (Figure 7).
where bt and bt 1 are the baseow (m3/s) at time step t and Table III shows the cumulative precipitation, discharge
t 1, respectively. BFImax is dened as the maximum and evapotranspiration for Tarqui and Yanuncay. The
value of the baseow index, that is, the long-term ratio of analyzed 1230 days correspond to the longest common
baseow to total streamow. An extended explanation of period of continuous series of precipitation and discharge.
the index is given in Eckhardt (2005). In order to minimize Total precipitation and specic runoff for Tarqui are
the inuence of a subjective choice of BFImax, BFImax lower than for Yanuncay, 6% and 55%, respectively.
values have been suggested for different hydrogeological Higher specic runoff for Yanuncay is likely related to
conditions (Eckhardt, 2005). BFImax values vary from the pramo and is consistent with results presented by
0.25 over 0.50 to 0.80 for perennial streams with hard rock Buytaert and Beven (2011).
aquifers, intermittent streams with porous aquifers and The evapotranspiration estimated using Equation 3 is
perennial streams with porous aquifer, respectively. higher in Tarqui, the difference is related directly to the
As the studied subcatchments lack groundwater level altitude distribution of the subcatchments.
data, we applied WHAT using different values of BFImax
0.25, 0.5 and 0.80, in order to infer if the baseow
response of the subcatchments corresponds to one of the Table III. Long term water balance for Yanuncay and Tarqui for a
aquifer types. period of 1230 days (2 April 200814 August 2011)
The described methods for baseow analysis and Yanuncay %P Tarqui %P
separation (nonlinear recession analysis, physically based
lter and digital lters) were applied at a daily basis. Precipitation P (mm) 3343 3147
It should be noted that the expressions derived for Specic runoff Q (mm) 1854 55 800 25
digital lters (Equations (16) and (17)) have a similar Evapotranspiration ET (mm) 1697 51 2154 68
P-Q-ET -6 7
form to the physically based lter expression (Equation
Figure 7. Correlation between specic runoff of the Tarqui and Yanuncay subcatchments for the periods 24 October 1998 till 8 May 2002 and 4
February 2008 till 14 August 2011
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
BASEFLOW COMPARISON BETWEEN TWO ANDEAN CATCHMENTS, ECUADOR
Table III shows that the water balance (P Q ET) is Table V. Variation of the ratio a/Area in function of b for
not completely closed, pointing to inaccuracies in the Yanuncay and Tarqui subcatchments for the middle reach and
downstream station
measured data, estimation of the evapotranspiration or
change in storage and net groundwater (Equation 1). a/Area (m33b sb m2)
Despite uncertainties in the estimation of the water
balance terms, Yanuncay and Tarqui show clearly Yanuncay Yanuncay Tarqui Tarqui
b downstream middle downstream middle
different hydrological patterns.
0.4 2.53E-07 2.03E-07 1.49E-07 1.53E-07
Nonlinear recession analysis 0.5 1.89E-07 1.65E-07 1.13E-07 1.24E-07
0.6 1.48E-07 1.39E-07 9.03E-08 1.09E-07
Wittenberg (1999) shows that the analysis of ow
recession curves allow the determination of characteristics
of the groundwater reservoir. In an analysis comparing
Chinese and German catchments, the last showed substantially more pramo area than Tarqui, 322 versus
signicantly higher values of a, and hence higher storage 52 km2, respectively. The at sedimentary valley of Tarqui
capacity than the Chinese catchments (Wittenberg, 1994). (25 km2) below 2700 m a.s.l. is not large enough to
Similarly, we aim to characterize the storage of the compensate the storage capacity of the pramo of
subcatchments of Yanuncay and Tarqui by comparing the Yanuncay.
values of a for gauging stations in the middle reach and We have calculated the ratio between a and the
downstream (total) for each of the two subcatchments. corresponding drainage area (a/Area) for each station
Comparing a in function of b values shows that for (middle and downstream) in order to compare the storage
each of the four subcatchments, a consistently reduces capacity per unit of area (m2) (Table V).
with increasing b (Table IV). This consistency was tested The ratio a/area shows that Yanuncay downstream has
in order to compare the storage capacity of different the highest storage capacity per unit of area. Comparing
subcatchments xing the value of b. with tendencies of a (Table IV), the a/Area ratio in
The Yanuncay downstream discharge series shows Yanuncay middle is now closer to Yanuncay downstream
higher a values than for the stations of Tarqui and the and is 46% higher than Tarqui downstream. On the other
Yanuncay middle. The storage capacity of Yanuncay hand, the Tarqui middle a/Area ratio is closer to the
downstream is 55%, 46% and 216% higher than Tarqui downstream ratio.
Yanuncay middle, Tarqui downstream and middle, Values in Table V conrm the inuence of different
respectively (for b equal to 0.5). Larger subcatchments ecosystems on the storage of the subcatchments.
generally show higher storage. This is noticed when
comparing the total and middle stations within the same Physically based lter
subcatchment. We assess the delay time for recharge d by applying
Tarqui downstream and Yanuncay middle show a Equation 11 on the discharge and baseow series whilst
similar storage capacity. There is a 6% difference for b gradually varying d. With d equal to zero days, the
equal to 0.5. The area of Tarqui is 167 km2 (54%) larger simulated baseow tends to be shifted to the right. Whilst
than Yanuncay middle. d increases, the shift also increases. As a consequence,
Besides the area, the composition and proportion of the adopting d equal to zero seems appropriate. This indicates
different ecosystems in the subcatchments appear to that the response of the subcatchment occurs within the
inuence the storage. The Yanuncay subcatchment has same day as the rainfall.
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
P. GUZMN ET AL.
The estimated recession constants (1 ) (Table VI) Table VII. Values of BFI ratio for Tarqui and Yanuncay middle
are consistent with literature values, slightly higher values and total river level stations
for (1 ) were found for the subcatchment of Yanuncay. Yanuncay Yanuncay Tarqui Tarqui
Overland ow and recharge (c1 and c3) in Yanuncay downstream middle downstream middle
are, respectively, 10% lower and 75% higher than in
Tarqui. As a consequence, the ratio c3/c1 is higher in Eckhardt lter, 0.26 0.27 0.28 0.25
Yanuncay. Average recharge (c3) is calculated on basis of BFImax = 0.25
Eckhardt lter, 0.45 0.45 0.47 0.43
c1 and c2, c1 is estimated based on Equation 14, and c2 is BFImax = 0.5
calculated using rainfall and discharge data. Higher c2 in Eckhardt lter, 0.63 0.64 0.65 0.59
Tarqui means higher evapotranspiration and less water BFImax = 0.8
available for recharge, which is also consistent with the BFLOW lter 0.65 0.66 0.66 0.60
values estimated in Table III.
The resulting baseow for both Yanuncay and Tarqui
appears to be remarkably high. The ratio between total BFImax equal to 0.25 and 0.5 underestimates baseow as
discharge and baseow (BFI) is 0.93 and 0.87 for it is consistently below the minimal discharge for Tarqui
Yanuncay and Tarqui, respectively. These values point to and Yanuncay stations.
an overestimation of c3/c1. Whilst BFImax is 0.25 or 0.5, the obtained BFI index
Assumptions for estimating c3/c1 do not take variations tends to be close to BFImax, but when BFImax is 0.8, BFI
in time into account. Furey and Gupta (2001) also noticed is on average 22% lower than BFImax (Table VII).
the need to consider c3/c1 as a time-dependent variable Digital lters use high and low-frequency components
because of changes in evapotranspiration (c2), soil moisture to separate baseow, and BFImax constrains the maximum
and precipitation. c2 estimated for wet and dry months baseow. It means that BFI reaches BFImax when the
varies from 0.41 to 0.76 in Yanuncay and from 0.47 to 1 in waves have a predominantly low frequency. If BFImax
Tarqui with lower values during April and higher during increases, more high-frequency waves are included from
August. c1 shows an important variation, it even reduces to the discharge hydrograph, and hence BFI becomes smaller
zero for Tarqui during the month of August. than BFImax, which is shown for the case when BFImax
The ratio c3/c1 can be estimated by an iterative equals 0.8.
procedure in which baseow is tted to the total ow Eckhardt (2008) stated that the true BFI values are
during dry periods. In Yanuncay and Tarqui, we have unknown. Therefore, one cannot say which one of the
estimated c3/c1 and found values of 15.9 and 12.8, values is the best approximation, but for the studied
respectively. A major problem constitutes of assessing c3/ subcatchments, the physical geomorphological character-
c1 during wet periods. Including measurements of soil istics are known to be different, and the results coming
moisture could help to estimate c3/c1 in wet periods but is from digital lters do not reect this as have been shown
especially difcult at subcatchment level because of its in other case studies by Eckhardt (2005).
spatial and temporal variation.
The comparison between Tarqui and Yanuncay Baseow comparison
consistently shows a higher recharge in Yanuncay Measuring baseow in the eld is very difcult (Li
compared with Tarqui. This can be explained by the et al., 2014). For assessing and comparing baseow in
lower potential evaporation at higher altitude. both subcatchments, we assumed that the total ow of the
subcatchment is equal to baseow during the lowest
Digital lters discharge periods, which coincide with the periods of
WHAT was applied for baseow separation from daily high-potential evapotranspiration and low precipitation.
discharge series for the middle and total gauging stations These periods were most common from July until
in both subcatchments. For the BFLOW lter, we have November of 2009 and from August until October of
used a lter parameter of 0.925, and for the Eckhardt 2010 (Figure 8).
lter, we have gradually changed the parameter BFImax During both dry periods, the Tarqui subcatchment
from 0.25 to 0.8 in order to infer whether the baseow showed the lowest specic runoff. Its average specic
response of the subcatchments corresponds to one of the runoff is about 60% lower than Yanuncay average
aquifer types described previously in methods section. specic runoff.
The baseow index shows only very small differences Despite the dry weather conditions, low specic
between Tarqui and Yanuncay considering the same runoff peaks are noticeable in Tarqui as well as in
period and the same lter approach (Table VII). Also, the Yanuncay. The low peaks might be considered ground-
BFI values found for BFLOW and Eckhardt lters are water discharge from close proximity to the river
very similar. We consider that the Eckhardt lter with (Chapman and Maxwell, 1996). They coincide with
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
BASEFLOW COMPARISON BETWEEN TWO ANDEAN CATCHMENTS, ECUADOR
Figure 8. Comparison of estimated baseow and specic runoff for Tarqui and Yanuncay during lowest discharge periods. Q is specic runoff.
Qb_eckh0.8, Qb_N.L.Rec and Qb_Phys.Fil are baseow estimated by Eckhardt digital lter, nonlinear recession analysis and physical lter, respectively
events when precipitation exceeds potential evapotrans- capacity analyzed through the parameter a and the ratio a/
piration. Area shows differences between the Yanuncay
The low peaks in Yanuncay and Tarqui have a different subcatchment and the more heterogeneous Tarqui
source or origin. The most marked peaks in Yanuncay subcatchment. This parameter can be explained by the
ratify the response of pramos characterized by high water balance and physical characteristics as indicated by
saturation and high hydraulic conductivity, whereas in the perceptual models.
Tarqui, mobilization of groundwater from the alluvial The Yanuncay subcatchment, which is largely repre-
valley is the dominant process. sented by perceptual model A (pramo), shows a storage
Baseow assessed by different methods showed minor capacity approximately 50% higher than Tarqui. This is
differences, whilst the dynamics varied from method to linked to the amount of pramo (three times larger or
method. 270 km2 more than Tarqui). Hence, pramo ecosystems
Digital and nonlinear recession lters are not capable to play a key role in the hydrological response. Because the
simulate low peaks. For digital lters, this is caused by pramo area in Tarqui is one third of that in Yanuncay,
the constrictions imposed by BFImax and the ltering of we would expect lower storage if this came exclusively
low frequencies as baseow. Nonlinear recession lters from pramo. However, considerable storage capacity
fail to capture peaks because their lter process does not other than in the pramo has been found in Tarqui in the
allow reaching a maximum value in a peak, rather, it is form of the presence of an alluvial aquifer.
limited to a recession curve pattern. Physical lters better Baseow presented in the Tarqui subcatchment shows
simulate these peaks. the combined effect of an area of pramo with the alluvial
Higher values of the c3/c1 ratio in physical lters allow aquifer in the lower valley (perceptual models A and
approaching baseow estimates close to the low dis- C). The contribution of perceptual model B to
charges, however, it results in an overestimation of baseow is considered negligible because of the domi-
baseow in other periods as discussed previously. In our nance of higher slopes and shallow soils.
comparison between Tarqui and Yanuncay for low The parameters proposed by physical based lters also
discharge periods, the assessed ratio c3/c1 for Yanuncay explain characteristics involved in baseow response such
is around 24% higher than for Tarqui. This conrms the as recharge (c3, c3/c1) and recession constant (1 ).
hypothesis of the high inuence of pramo in the slow The ratio c3/c1 is estimated using precipitation and
response of the catchment and the capacity of pramo to streamow data, thus accurate and good representation
sustain river baseow during low precipitation periods. of temporal and spatial distribution of precipitation seems
crucial. Moreover, the current method assumes a constant
ratio. Further research should investigate the time-
dependence of c3/c1 in function of catchment wetness
CONCLUSIONS
and/or soil moisture data.
Baseow characteristics of two middle sized Andean Although recharge in Yanuncay is larger than Tarqui
catchments have been investigated and compared using because of differences in potential evaporation, the
different baseow separation approaches. parameter c2 for Tarqui for August demonstrates the
Recession characteristics of baseow are clearly importance of the alluvial aquifer for sustaining baseow,
differentiated from the comparative analysis. Storage whilst there is neither recharge nor overland ow. In
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
P. GUZMN ET AL.
contrast to Buytaert et al. (2006c), we found evidence that Chapman T, Maxwell A. 1996. Baseow separation-comparison of
rd
numerical methods with tracer experiments. 23 Hydrology and Water
baseow in the Paute catchment has additional sources and Resources Symposium, Australia; 539545.
is not completely supplied by the pramo ecosystem. Chapman T. 1999. A comparison of algorithms for stream ow recession
The importance of the pramo ecosystem for sustaining and baseow separation. Hydrological Processes 13: 701714. DOI:
10.1002/(SICI)1099-1085(19990415)13:5<701::AID-HYP774>3.0.
baseow is a signicant contribution and reason for CO;2-2
conserving them. However, the alluvial aquifer contribu- Chow V, Maidment D, Mays L. 1988. Applied Hydrology. McGraw-Hill
tion to storage capacity and baseow in Tarqui is also Series in Water Resources and Environmental Engineering. McGraw-
Hill: New York; 552.
important and demands further attention for study and Coutagne A. 1948. Les variations de dbit en priode non inuence
management. par les prcipitations-le dbit dinltration (corrlations uviales
internes) 2me partie. La Houille Blanche 416436. DOI: 10.1051/
lhb/1948053
Crespo P, Feyen J, Buytaert W, Bcker A, Breuer L, Frede H, Ramrez M.
ACKNOWLEDGEMENTS 2011. Identifying controls of the rainfallrunoff response of small
catchments in the tropical Andes (Ecuador). Journal of Hydrology 407:
We thank Electric Corporation of Ecuador-Hidropaute 164174. DOI: 10.1016/j.jhydrol.2011.07.021
(CELEC EP Hidropaute) for their nancial support of this Dewandel B, Lachassagne P, Bakalowicz M, Weng P, Al-Malki A. 2003.
Evaluation of aquifer thickness by analysing recession hydrographs.
research as well as the VLIR IUC programme between Application to the Oman ophiolite hard-rock aquifer. Journal of
the Flemish universities and Universidad de Cuenca, Hydrology 274: 248269. DOI: 10.1016/S0022-1694(02)00418-3
Ecuador. Dunne T. 1983. Relation of eld studies and modeling in the prediction of
storm runoff. Journal of Hydrology 65: 2548. DOI: 10.1016/0022-
1694(83)90209-3
Eckhardt K. 2005. How to construct recursive digital lters for baseow
REFERENCES separation. Hydrological Processes 19: 507515. DOI: 10.1002/hyp.5675
Eckhardt K. 2008. A comparison of baseow indices, which were
Allen R, Pereira L, Raes D, Smith M. 1998. Crop evapotranspiration- calculated with seven different baseow separation methods. Journal of
guidelines for computing crop water requirements-FAO irrigation and Hydrology 352: 168173. DOI: 10.1016/j.jhydrol.2008.01.005
drainage paper 56; 2640. Frhlich K, Frhlich W, Wittenberg H. 1994. Determination of
Arnold J, Allen P. 1999. Automated methods for estimating baseow and groundwater recharge by baseow separation: regional analysis in
ground water recharge from streamow records 1. JAWRA Journal of northeast China. IAHS Publications-Series of Proceedings and Reports-
the American Water Resources Association 35: 411424. DOI: Intern Assoc Hydrological Sciences 221: 6976.
10.1111/j.1752-1688.1999.tb03599.x Furey P, Gupta V. 2000. Space-time variability of low streamows in river
Bako M, Hunt D. 1988. Derivation of baseow recession constant using networks. Water Resources Research 36: 26792690. DOI: 10.1029/
computer and numerical analysis. Hydrological Sciences Journal 33: 2000WR900136
357367. DOI: 10.1080/02626668809491259 Furey P, Gupta V. 2001. A physically based lter for separating base ow
Bako M, Owoade A. 1988. Field application of a numerical method for the from streamow time series. Water Resources Research 37:
derivation of baseow recession constant. Hydrological Processes 2: 27092722. DOI: 10.1029/2001WR000243
331336. DOI: 10.1002/hyp.3360020404 Hall F. 1968. Base-ow recessions a Review. Water Resources Research
Beven K. 2012. Rainfall-Runoff Modelling: The Primer. John Wiley & 4: 973983. DOI: 10.1029/WR004i005p00973
Sons: Chichester; 488. Huyck A, Pauwels V, Verhoest N. 2005. A base ow separation algorithm
Buytaert W, Beven K. 2011. Models as multiple working hypotheses: based on the linearized Boussinesq equation for complex hillslopes.
Hydrological simulation of tropical alpine wetlands. Hydrological Water Resources Research 41. DOI: 10.1029/2004WR003789
Processes 25: 17841799. DOI: 10.1002/hyp.7936 Lamb R, Beven K. 1997. Using interactive recession curve analysis to
Buytaert W, De Bivre B, Wyseure G, Deckers J. 2004. The use of the specify a general catchment storage model. Hydrology and Earth
linear reservoir concept to quantify the impact of changes in land use on System Sciences 1: 101113.
the hydrology of catchments in the Andes. Hydrology and Earth System Li L, Maier H, Partington D, Lambert M, Simmons C. 2014. Performance
Sciences 8: 108114. DOI: 10.5194/hess-8-108-2004 assessment and improvement of recursive digital baseow lters for
Buytaert W, Wyseure G, De Bivre B, Deckers J. 2005. The effect of land- catchments with different physical characteristics and hydrological
use changes on the hydrological behaviour of Histic Andosols in south inputs. Environmental Modelling & Software 54: 3952. DOI: 10.1016/
Ecuador. Hydrological Processes 19: 39853997. DOI: 10.1002/ j.envsoft.2013.12.011
hyp.5867 Lim K, Engel B, Tang Z, Choi J, Kim K, Muthukrishnan S, Tripathy D.
Buytaert W, Iiguez V, Clleri R, De Bivre B, Wyseure G, Deckers J. 2005. Automated Web Gis Based Hydrograph Analysis Tool, WHAT.
2006a. Analysis of the water balance of small pramo catchments in JAWRA Journal of the American Water Resources Association 41:
south Ecuador. Environmental role of wetlands in headwaters 63: 14071416. DOI: 10.1111/j.1752-1688.2005.tb03808.x
271281. DOI: 10.1007/1-4020-4228-0_24 Lyne V, Hollick M. 1979. Stochastic time-variable rainfall-runoff
Buytaert W, Deckers J, Wyseure G. 2006b. Description and classication modelling. Hydrology and Water Resources Symposium. Institut of
of nonallophanic Andosols in south Ecuadorian alpine grasslands Engineers Australia. Perth.
(pramo). Geomorphology 73: 207221. DOI: 10.1016/j. Maillet E. 1905. Essais dhydraulique souterraine & uviale. Hermann.
geomorph.2005.06.012 Nathan RJ, McMahon TA. 1990. Evaluation of automated techniques for
Buytaert W, Clleri R, De Bivre B, Cisneros F, Wyseure G, Deckers J, base ow and recession analyses. Water Resources Research 26:
Hofstede R. 2006c. Human impact on the hydrology of the Andean 14651473. DOI: 10.1029/WR026i007p01465
pramos. Earth-Science Reviews 79: 5372. DOI: 10.1016/j. Prasad R. 1967. A nonlinear hydrologic system response model. ASCE
earscirev.2006.06.002 Journal of Hydraulics Division 93: 201202.
Clleri R, Feyen J. 2009. The hydrology of tropical Andean ecosystems: Roa-Garca M, Brown S, Schreier H, Lavkulich L. 2011. The role of land
Importance, knowledge status, and perspectives. Mountain Research use and soils in regulating water ow in small headwater catchments of
and Development 29: 350355. DOI: 10.1659/mrd.00007 the Andes. Water Resources Research 47: W05510. DOI: 10.1029/
Clleri R, Willems P, Buytaert W, Feyen J. 2007. Spacetime rainfall 2010WR009582
variability in the Paute basin, Ecuadorian Andes. Hydrological Smakhtin V. 2001. Low ow hydrology: a review. Journal of Hydrology
Processes 21: 33163327. DOI: 10.1002/hyp.6575 240: 147186. DOI: 10.1016/S0022-1694(00)00340-1
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
BASEFLOW COMPARISON BETWEEN TWO ANDEAN CATCHMENTS, ECUADOR
Spongberg M. 2000. Spectral analysis of base ow separation with digital Wang D, Cai X. 2010. Comparative study of climate and human
lters. Water Resources Research 36: 745752. DOI: 10.1029/ impacts on seasonal baseow in urban and agricultural watersheds.
1999WR900303 Geophysical Research Letters 37: L06406. DOI: 10.1029/
Sugiyama H. 1996. Analysis and extraction of low ow recession 2009GL041879
characteristics. JAWRA Journal of the American Water Resources Wittenberg H. 1994. Nonlinear analysis of ow recession curves. IAHS
Association 32: 491497. DOI: 10.1111/j.1752-1688.1996.tb04047.x Publications-Series of Proceedings and Reports-Intern Assoc Hydro-
Sujono J, Shikasho S, Hiramatsu K. 2004. A comparison of techniques for logical Sciences 221: 6168.
hydrograph recession analysis. Hydrological Processes 18: 403413. Wittenberg H. 1999. Baseow recession and recharge as nonlinear storage
DOI: 10.1002/hyp.1247 processes. Hydrological Processes 13: 715726. DOI: 10.1002/(SICI)
Tallaksen L. 1995. A review of baseow recession analysis. Journal of 1099-1085(19990415)13:5<715::AID-HYP775>3.0.CO;2-N
Hydrology 165: 349370. DOI: 10.1016/0022-1694(94)02540-R Wittenberg H. 2003. Effects of season and man-made changes on
Van Dijk A. 2010. Climate and terrain factors explaining streamow baseow and ow recession: case studies. Hydrological Processes 17:
response and recession in Australian catchments. Hydrology and Earth 21132123. DOI: 10.1002/hyp.1324
System Sciences 14: 159169. DOI: 10.5194/hess-14-159-2010 Wittenberg H, Sivapalan M. 1999. Watershed groundwater balance
Vauchel P. 2005. HYDRACCESS: Software for management and estimation using streamow recession analysis and baseow separation.
processing of hydro-meteorological data. www.ore-hybam.org/index. Journal of Hydrology 219: 2033. DOI: 10.1016/S0022-1694(99)
php/Software/Hydraccess 00040-2
Vogel R, Kroll C. 1996. Estimation of baseow recession constants. Water Zecharias Y, Brutsaert W. 1988. Recession characteristics of groundwater
Resources Management 10: 303320. DOI: 10.1007/BF00508898 outow and base ow from mountainous watersheds. Water Resources
Wang D, Cai X. 2009. Detecting human interferences to low ows Research 24: 16511658.
through base ow recession analysis. Water Resources Research 45:
W07426. DOI: 10.1029/2009WR007819
Data:
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Tmax (C) 11.2 11.4 11.3 10.6 10.4 9.6 8.7 9.2 10.0 11.6 11.7 11.6
Tmin (C) 4.0 4.2 4.3 4.1 3.8 3.7 2.8 2.3 2.8 3.2 2.8 3.4
Tdew (C) 5.3 5.5 5.6 5.5 4.8 4.8 3.8 3.4 3.8 4.3 3.5 4.7
Ws (m/s) 3.7 3.4 3.5 3.4 4.0 4.5 5.7 5.5 5.1 3.5 3.3 3.3
Srad (W/m2) 163.2 158.4 151.3 147.2 139.9 137.4 124.3 135.4 133.4 175.5 174.2 157.3
Tmax, maximum temperature; Tmin, minimum temperature; Tdew, dew temperature (dew point); Ws, wind speed;
Srad, solar radiation
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)
P. GUZMN ET AL.
Data:
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Tmax (C) 18.2 18.3 18.3 17.7 18.2 17.7 18.0 17.5 18.0 18.7 19.5 19.3
Tmin (C) 7.6 8.4 7.9 8.4 7.8 7.2 6.1 5.9 6.1 4.7 1.6 7.2
Tdew (C) 8.1 8.2 8.3 8.1 8.0 6.9 6.3 6.4 7.5 6.7 4.2 8.0
Ws (m/s) 0.6 0.8 0.5 0.6 0.5 0.6 0.7 0.6 0.5 0.6 0.8 0.7
Srad (W/m2) 175.6 188.6 172.8 144.0 143.0 139.3 156.7 151.4 163.4 186.8 237.4 176.8
Tmax, maximum temperature; Tmin, minimum temperature; Tdew, dew temperature (dew point); Ws, wind speed;
Srad, solar radiation
Copyright 2014 John Wiley & Sons, Ltd. Hydrol. Process. (2015)