Abstract
This study evaluates the efficacy of Palmer drought indices to monitor hydrological droughts in river discharges
and soil moisture in selected watersheds with varying geomorphologic characteristics in region of Thessaly, Greece.
The Palmer four indices (PDSI, Weighted PDSI, PHDI and the moisture anomaly Z-index) were used as indicators
of meteorological drought severity. The hydrological drought severity was evaluated from the outputs of the monthly
UTHBAL conceptual water balance model. The UTHBAL model was calibrated with the available observed runoff
data to extend, reconstruct and produce runoff and soil moisture timeseries for the hydrologic period 19602002 at
the study catchments. The produced hydrologic variables were normalized through Box-Cox transformation and
standardized to normal distribution. The standardized hydrologic variables were used as an indicator of hydrological
drought severity and were compared with the Palmer drought indices timeseries estimated by basin-wide
meteorological data. The results showed that, in general, the Weighted PDSI and the moisture anomaly Z-index were
found to better represent river discharges and soil moisture, respectively, for all study watersheds irrespectively to
their area, geophysical, and hydroclimatic characteristics. However, the results were quite variable in the
identification of specific historical drought periods. Although, the Palmer indices were successful in the identification
of drought severity of historical events, they failed to identify the drought duration.
Keywords:
Palmer drought indices; Runoff; Soil moisture; Meteorological drought; Hydrological drought; Drought
severity; Drought duration
1. Introduction
In many Mediterranean countries environmental policy and water resources and drought
*Corresponding author.
Presented at the conference on Water Resources Management: New Approaches and Technologies, 1416 June 2007,
Chania, Greece.
0011-9164/09/$ See front matter 2008 Elsevier B.V. All rights reserved
doi:10.1016/j.desal.2007.12.019
Thessaly, respectively [5]. The annual precipitation for the 19761977 year was 467 mm and for
the year 19891990 was 521 mm. The driest
January and February and the second driest
March in record occurred during the hydrological
year 19891990. The prolonged and significant
decrease of monthly and annual precipitation has
a dramatic impact on water resources of the
region. Usually, the dry periods are accompanied
with high temperatures, which lead to higher
evapotranspiration rates and dry soils. These conditions inversely affect both the natural vegetation and the agriculture of the region as well as
the available storage of the reservoirs. Severe and
extremely dry conditions result in irrigation
cutbacks, overexploitation of groundwater and
Table 1
Characteristics of the study watersheds
Watershed
Area
(km2)
Elevation
range (m)
Mean
elevation (m)
Mean annual
precipitation (mm)
Mean annual
runoff (mm)
Runoff
coefficient
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
133
145
2055
2869
6591
615
1155
3001800
2002000
1802000
1502000
702000
7002300
5002300
949
838
660
555
451
1400
1288
1820
1450
985
890
720
1860
1875
1130
830
600
420
290
1150
1290
0.62
0.57
0.61
0.47
0.40
0.62
0.69
hydrological problems. Recently, these hydrological models have been employed to explore the
impact of climatic change. They also have been
utilized for long-range streamflow forecasting.
Although such applications may use hourly or
daily models, these models are, however, more
data intensive and have more parameters than the
monthly models. A complete review of water
balance model applications could be found in Xu
and Singh [33].
In this study the monthly conceptual water
balance UTHBAL model [34] has been used. The
water balance model allocates the watershed
runoff into three components, the surface runoff,
the interflow runoff and the baseflow runoff
using a soil moisture mechanism with the first
priority of the balancing being the fulfillment of
actual evapotranspiration. The model separates
the total precipitation into rainfall and snowfall,
because the correct division of precipitation is
essential for accurate runoff simulation. The rainsnow percentage is estimated using a logistic
relationship based on mean monthly temperature
and the snowmelt is calculated estimated using
the simple degree-day method [35]. The
UTHBAL model requires monthly values of
mean temperature, precipitation, and potential
evapotranspiration and produces values for actual
evapotranspiration, soil moisture, groundwater
and surface runoff. The input timeseries were
estimated using the methods presented in the
previous paragraphs. The UTHBAL model has
six parameters to be optimized in order to estimate watershed runoff. The optimization was
performed using the Generalised Reduced Gradient Algorithm and the Nash-Sutcliffe Model
Efficiency was used as the objective function.
The UTHBAL model was calibrated with the
available observed runoff data to extend, reconstruct and produce runoff and soil moisture
timeseries for the period of analysis (19602002)
and for all study watersheds. A detailed description of the UTHBAL model has been presented in
a recent paper [34].
The synthetic runoff and soil moisture timeseries were used for the estimation of hydrological drought. The synthetic runoff and soil
moisture were normalized through Box-Cox
transformation and standardized. The transformed
and standardized runoff and soil moisture timeseries were used as an indicator of hydrological
drought severity and were compared with the
Palmers drought indices timeseries estimated by
basin-wide meteorological data.
3.2. Meteorological drought assessment
The Palmer Drought Severity Index (PDSI)
was introduced by Palmer [36] for the assessment
of meteorological drought and it is a well-known
and widely used drought index. Although PDSI is
referred to as an index of meteorological drought,
however, the procedure considers precipitation,
evapotranspiration, and soil moisture conditions,
which are determinants of hydrological drought,
i.e. the period during which the actual water
supply is less than the minimum water supply
necessary for normal operations in a particular
region. The PDSI measures the departure of the
moisture supply from normal conditions. Moisture supply is calculated from the water balance
of a two-layer soil model using monthly mean
precipitation and temperature data as well as the
local available soil water content. From the input
data, all the basic water balance terms, namely
evapotranspiration, soil recharge, runoff, and
moisture loss from the surface layer are
estimated.
The PDSI varies roughly between !6 (drought
periods) and +6 (wet periods) with the zero value
indicating the normal conditions. The PDSI is
standardized for different regions and time periods to facilitate direct comparisons of the PDSI
between different regions [37]. The Palmer
method used for calculating the PDSI has a
number of limitations and deficiencies [38]. The
limitations of the method can be classified into
two categories: the water balance model defi-
4. Application Results
The UTHBAL model was calibrated with the
available observed runoff data to extend, reconstruct and produce runoff and soil moisture
timeseries for the period of analysis 19602002
for all study watersheds. The simulated and
observed hydrographs compared graphically and
statistically. Three statistical indices were used to
assess the accuracy and performance of the simulation method, namely, the model efficiency [43]:
n
Eff 1
Qobs Qsim
i 1
n
Qobs Qobs
i 1
(1)
% DV
Vsim Vobs
100
Vobs
(2)
10
Table 2
Calibration statistics of the UTHBAL model with the available observed runoff data
Watershed
Eff
DV (%)
Calibration period
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
0.72
0.72
0.80
0.78
0.76
0.71
0.72
0.96
!0.80
!1.98
!1.28
0.39
!0.75
!0.99
0.85
0.85
0.90
0.89
0.87
0.84
0.85
Oct 60Sep 94
Oct 60Sep 85, Oct 87Sep 94
Oct 60Sep 71
Oct 60Dec 69, Feb 72Jun 94
Oct 60Feb 67, Jan 69Mar 91
Oct 60Sep 94
Oct 60Sep 94
X 1
, 0
Y
ln X , 0
(3)
Y Y
Y
(4)
11
(a)
(b)
Fig. 2. Comparison of observed and simulated hydrographs for (a) Mesohora watershed and (b) Larissa watershed.
12
(i.e. standardized river discharges and soil moisture). Hence, the strength of the anomaly for all
calculated drought indices is classified as set out
in Table 3. This table also contains the corresponding probabilities of occurrence of each
severity arising naturally from the normal probability density function. Thus, at a given location
for an individual month, moderate dry and drier
periods (index value #!1) have an occurrence
Table 3
Drought classification by standardized hydrological
variables and Palmer drought indices with corresponding
event probabilities
Drought Index
value
Category
Probability
(%)
2.00 or more
1.50 to 1.99
1.00 to 1.49
!0.99 to 0.99
!1.49 to !1.00
!1.99 to !1.50
!2 or less
Extremely wet
Severely wet
Moderately wet
Near normal
Moderately dry
Severely dry
Extremely dry
2.3
4.4
9.2
68.2
9.2
4.4
2.3
probability of 15.9%, whereas extreme dry periods (index value #!2) have an event probability
of 2.3%. Extreme values in all calculated indices
will, by definition, occur with the same frequency
at all locations.
Table 4 shows the Pearson correlation coefficients between the continuous standardized series
of river discharges, soil moisture and the four
Palmer indices. For river discharges and soil
moisture, correlations are positive, but there are
important differences with regard to examined
meteorological drought indices. Considering river
discharges, higher correlations have been obtained with the Weighted PDSI (average R = 0.80
for all watersheds, ranging from R = 0.78 to 0.82)
followed by the PDSI (Table 4) for all study
watersheds. Considering soil moisture, higher
correlations have been obtained with the moisture
anomaly Palmer Z-index (average R = 0.87 for all
watersheds, ranging from R = 0.85 to 0.89).
Graphical examination of temporal evolution
and scattergraphs confirms that the Weighted
PDSI and the Palmer Z-index conform better to
Table 4
Correlation between continuous meteorological drought indices and the standardized hydrological variables
Watershed
River discharge (Runoff Z-index):
Mouzaki
Pili
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Soil moisture (Soil moisture Z-index):
Mouzaki
Pili
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Palmer Z-index
PDSI
Weighted PDSI
PHDI
0.69
0.70
0.74
0.74
0.69
0.72
0.74
0.73
0.77
0.77
0.79
0.78
0.77
0.77
0.78
0.81
0.80
0.82
0.81
0.81
0.80
0.69
0.71
0.69
0.72
0.72
0.71
0.71
0.86
0.85
0.89
0.89
0.88
0.86
0.87
0.74
0.73
0.72
0.75
0.77
0.73
0.74
0.75
0.75
0.74
0.76
0.77
0.75
0.74
0.63
0.62
0.61
0.65
0.67
0.63
0.63
13
Fig. 3. Scattergraphs of Palmer drought indices with river discharges (Runoff Z-index) for Larissa watershed.
ber and January. Contrarily, the worst correlations were obtained for spring and summer
months.
This noticeable seasonality in the strength of
correlations is, probably, the result of the hydrological behavior of the study watersheds. A
possible reason could be the seasonal variability
of precipitation and the accumulation of snow and
its consequent melting. The snow accumulation
and the snowmelt are not accounted in the
calculation of the Palmer indices but they are
taken into consideration in the simulation of
runoff. These processes cause redistribution of
14
Fig. 4. Scattergraphs of Palmer drought indices with soil moisture (Soil moisture Z-index) for Larissa watershed.
Table 5
Monthly correlations between Palmer meteorological drought indices and the standardized hydrological variables for
Larissa watershed
Palmer Index
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
0.88
0.88
0.85
0.82
0.64
0.40
0.58
0.62
0.57
0.52
PDSI
0.85
0.86
0.90
0.91
0.81
0.76
0.71
0.67
0.76
0.72
0.72
0.73
Weighted PDSI
0.86
0.88
0.90
0.91
0.81
0.81
0.72
0.67
0.81
0.82
0.80
0.76
PHDI
0.83
0.82
0.83
0.83
0.75
0.72
0.64
0.60
0.72
0.63
0.68
0.68
0.92
0.88
0.94
0.96
0.95
0.94
0.89
0.84
0.87
0.88
0.84
0.78
PDSI
0.88
0.88
0.89
0.84
0.74
0.68
0.67
0.73
0.75
0.72
0.77
0.70
Weighted PDSI
0.88
0.88
0.88
0.83
0.69
0.70
0.66
0.77
0.81
0.77
0.75
0.72
PHDI
0.81
0.80
0.80
0.75
0.63
0.60
0.55
0.66
0.65
0.58
0.62
0.60
15
(a)
(b)
Fig. 5. Identification of major hydrological drought episodes for Larissa watershed (a) river discharges and (b) soil
moisture.
16
Table 6
Correlation between meteorological drought indices and the standardized hydrological variables for the three identified
drought episodes:( a) Oct 1976Sep 1978, (b) Oct 1987Sep 1990, and (c) Oct 2000Sep 2002
Watershed
Palmer Z-index
a
PDSI
c
Weighted PDSI
PHDI
0.73
0.88
0.49
0.50
0.60
0.86
0.87
0.56
0.58
0.64
0.73
0.76
0.73
0.77
0.81
0.79
0.93
0.89
0.82
0.92
0.80
0.79
0.87
0.58
0.57
0.65
0.90
0.89
0.66
0.70
0.80
0.74
0.75
0.70
0.75
0.72
0.73
0.88
0.82
0.91
0.86
0.80
0.70
0.73
0.41
0.41
0.55
0.82
0.83
0.57
0.53
0.67
0.65
0.68
0.59
0.65
0.67
0.66
0.77
0.71
0.88
0.76
0.75
0.76
0.89
0.48
0.48
0.64
0.83
0.82
0.49
0.53
0.68
0.74
0.80
0.75
0.74
0.79
0.78
0.91
0.87
0.86
0.91
0.74
0.83
0.87
0.57
0.55
0.69
0.86
0.85
0.60
0.67
0.76
0.74
0.77
0.67
0.70
0.73
0.74
0.88
0.80
0.91
0.86
0.79
0.67
0.72
0.38
0.37
0.53
0.76
0.77
0.49
0.49
0.60
0.59
0.65
0.53
0.57
0.66
0.69
0.76
0.67
0.82
0.79
0.72
show the higher correlations with the runoff Zindex, and the Palmer Z-index and the PDSI are
better correlated with the soil moisture Z-index.
Nevertheless, there are a few cases for which
other Palmer indices have similar correlation or
they are even better correlated with the hydrological Z-indices. Furthermore, this variation of
the results is not consistent either for the same
event or for the same watershed. This indicates
that identification of the evolution of runoff and
soil moisture drought by the Palmer indices is
affected by the meteorological/hydrological
conditions during the event and the characteristics
of the watershed.
Identification of the drought onset and termination can be based on the value that the variable
under analysis takes in a single interval without
memory of the past, e.g. run analysis [46] or by
taking explicitly into account the memory of the
past intervals [36]. For these historical drought
episodes the threshold drought value is defined
17
Fig. 6. Response of Palmer drought indices with standardized hydrological variables for identified drought episodes in
Larissa watershed.
18
Table 7
Average drought severity of the identified hydrological drought periods
Watershed
Period Oct 1976Sep 1978:
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Period Oct 1988Sep 1990:
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Period Oct 2000Sep 2002
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Runoff
Z-index
Soil moisture
Z-index
Palmer
Z-index
PDSI
Weighted
PDSI
PHDI
!1.61
!1.67
!1.82
!1.86
!1.79
!1.76
!1.70
!1.59
!1.79
!1.79
!1.81
!1.77
!1.81
!1.66
!1.55
!1.52
!1.41
!1.37
!1.37
!1.36
!1.60
!1.81
!1.43
!1.60
!1.53
!1.65
!1.51
!1.59
!1.79
!1.48
!1.63
!1.62
!1.64
!1.52
!1.67
!1.56
!1.40
!1.55
!1.51
!1.62
!1.45
!1.52
!1.75
!1.99
!1.67
!1.57
!1.65
!1.76
!1.71
!1.68
!1.97
!1.69
!1.66
!1.85
!1.72
!1.62
!1.31
!1.50
!1.40
!1.41
!1.48
!1.48
!1.47
!1.42
!1.68
!1.56
!1.62
!1.53
!1.51
!1.40
!1.47
!1.74
!1.57
!1.71
!1.55
!1.56
!1.47
!1.35
!1.64
!1.52
!1.57
!1.51
!1.50
!1.34
!1.20
!1.49
!1.29
!1.46
!1.34
!1.12
!1.41
!1.29
!1.35
!1.26
!1.38
!1.39
!1.08
!1.12
!1.43
!1.30
!1.45
!1.29
!1.44
!1.24
!1.20
!1.47
!1.44
!1.55
!1.29
!1.42
!1.24
!1.28
!1.41
!1.45
!1.58
!1.34
!1.48
!1.27
!1.33
!1.39
!1.41
!1.47
!1.26
!1.38
!1.26
!1.26
19
Table 8
Drought duration (in months) of the identified hydrological drought periods
Watershed
Period Oct 1976Sep 1978:
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Period Oct 1988Sep 1990:
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Period Oct 2000Sep 2002:
Pili
Mouzaki
Mesdani
Ali Efenti
Larissa
Mesohora
Sykia
Runoff
Z-index
Soil moisture
Z-index
Palmer
Z-index
PDSI
Weighted
PDSI
PHDI
10
9
10
10
13
8
7
10
8
10
10
12
6
7
5
5
8
8
9
6
4
10
17
18
18
19
11
10
11
14
16
15
19
10
9
13
16
18
18
19
11
10
9
13
13
14
14
9
10
9
13
12
13
13
9
10
6
8
8
9
9
7
5
20
23
17
25
18
12
15
18
21
20
23
19
13
13
20
22
20
27
20
19
15
6
5
7
8
9
3
3
5
7
5
6
7
2
4
3
4
6
7
7
2
3
7
7
16
10
16
11
8
11
9
18
12
16
10
9
9
9
20
13
20
10
9
20
References
[1] D.A.Wilhite, in: D.A. Wilhite, ed., Drought: A
Global Assessment, Routledge, 2000, pp. 318.
[2] X. Lana and A. Burgueno, Int. J. Climatol., 18
(1998) 93110.
[3] T.W. Kim, J.B. Valdes and J. Aparicio, Water Int.,
27(3) (2002) 420430.
21