=Paper=
{{Paper
|id=Vol-1152/paper14
|storemode=property
|title=Spatiotemporal Classification of Drought Severity
|pdfUrl=https://ceur-ws.org/Vol-1152/paper14.pdf
|volume=Vol-1152
|dblpUrl=https://dblp.org/rec/conf/haicta/NicolasBSPB11
}}
==Spatiotemporal Classification of Drought Severity==
https://ceur-ws.org/Vol-1152/paper14.pdf
Spatiotemporal classification of drought severity
Dalezios R. Nicolas1a, Anna Blanta1b, Nicos Spyropoulos2, Nicholas Pismichos1c,
Evangelia Boukouvala1d,
1
Laboratory of Agrometeorology, Department of Agriculture, Ichthyology & Acquatic
Environment, , University of Thessaly, Volos, Greece, email: adalezios@uth.gr,
b
amplanta@uth.gr, cpismicho@uth.gr, dboukouva@uth.gr
2
Agricultural University of Athens, Department of Natural Resource Development and
Agricultural Engineering, Athens, email: nicosp@hol.gr
Abstract. The growing number and effectiveness of earth observation satellite
systems, along with the increasing reliability of remote sensing methodologies
and techniques, present a wide range of new capabilities in monitoring and
assessing droughts. In this paper, several drought features are analyzed and
assessed by using remotely sensed Reconnaissance Drought Index (RDI). The
developed methodology is applied to the region of Thessaly in Central Greece,
which is the major agricultural area in the country. In particular, severity, areal
extent, duration, onset and end time are analyzed from monthly RDI images
over the period 1981-2001. The results show an increase in the areal extent of
drought during each episode and that droughts are classified in two classes,
namely mild to moderate and severe to extreme, respectively, lasting about one
hydrological year. The onset of severe to extreme droughts coincides with the
beginning of the hydrological year, whereas the onset of mild to moderate
droughts is in spring.
Keywords: RDI, Remote Sensing, drought features
1 Introduction
Drought is one of the major natural hazards with significant impact to
environment, agriculture, economy and society. There is a great variety of sectors
affected by drought and its spatiotemporal variability (Heim, 2002). The basic cause
of drought is the lack of precipitation events over a period of time in a region.
Droughts occur in both high and low rainfall areas and virtually all climate regimes.
There are several definitions and types of drought as a hazard, namely
meteorological, agricultural, hydrological drought, as well as socioeconomic impacts,
which are based on the temporal and spatial scale of the selected approach. Drought
impacts are very critical affecting societies more than any other natural disaster
(Keyantash and Dracup, 2002). However it is difficult to determine the effects of
drought as it constitutes a complicated phenomenon evolving gradually in any single
region. Monitoring and assessing drought conditions is usually performed through
drought indicators and indices.
_________________________________
Copyright ©by the paper’s authors. Copying permitted only for private and academic purposes.
In: M. Salampasis, A. Matopoulos (eds.): Proceedings of the International Conference on Information
and Communication Technologies
for Sustainable Agri-production and Environment (HAICTA 2011), Skiathos, 8-11 September, 2011.
171
The quantification of drought is not an easy task. It is common practice that
several indicators can be synthesized into a single indicator in quantitative terms,
called a drought index. There are several widely used drought indices using
conventional and/or remote sensing data (Kanellou et al., 2008). Historically, drought
quantification methods are based on conventional hydrometeorological data, such as
precipitation and temperature, which are limited, often inaccurate and unavailable in
near real-time (Thenkabail et al, 2004). On the other hand, satellite-based data are
consistently available and can be used to detect several features (Kanellou et al.,
2008). In fact, over the last decades, remote sensing has gradually become an
important tool for the detection of the spatial and temporal distribution and
characteristics of drought at different scales. Thus, the growing number and
effectiveness of earth observation satellite systems, along with the increasing
reliability of remote sensing methods and techniques, present a wide range of new
capabilities in monitoring and assessing droughts.
Remotely sensed drought features and characteristics have recently become key
parameters in any drought preparedness and mitigation plan in the framework of
drought risk management (Rossi, 2000). In particular, in order to assess and monitor
drought episodes and to alleviate the impacts of droughts it is necessary to detect
several drought features such as severity, duration, periodicity, areal extent, onset and
end time and to link drought variability to climate and its variability (Loukas et al.,
2002). It is clear that there is a need for proper quantification of drought impacts.
Thus monitoring of drought development is of critical importance in economically
and environmentally sensitive regions. In this paper, the remote sensing potential in
terms of data and methods is explored in order to quantify drought and classify
drought severity based on several drought features by using RDI.
RDI provides information for the water deficit in a region as it is based not only
on precipitation, but also on potential evapotranspiration. In the computation of RDI
the innovation consists of employing Blaney-Criddle method for potential
evapotranspiration instead of Thornthwaite method, since it is more appropriate for
the Mediterranean region with dry and hot summers (Blaney and Criddle, 1950).
Several drought features and characteristics are analyzed from monthly remotely
sensed RDI images for the period 1981-2001 for Thessaly, Greece. Specifically, areal
extent and severity during drought episodes signify the spatiotemporal variability of
droughts in Thessaly. The paper is organized as follows: in section 2 remote sensing
in drought assessment is described; in section 3 classification of drought severity is
presented; in section 4 the methodology is developed and section 5 delineates the
analysis and discussion of results.
2 Remote Sensing in Drought Assessment
In this section the remote sensing potential in drought assessment is presented along
with drought indices based on remote sensing.
172
2.1 Remote Sensing Potential
The application and utility of remote sensing to drought assessment is growing
rapidly, mainly due to the increasing number of pertinent satellite systems and their
capabilities as well as to the International Decade for Natural Disaster Reduction and
impacts of climatic variability and/or change. Remote sensing methodologies and
techniques can be employed in several aspects of drought, such as vulnerability and
damage assessment and warning. Exposure to drought can be controlled as well as an
effort can be undertaken to alleviate the effects of drought. There are three steps
which outline the elements of drought monitoring and, in general hazard monitoring
and management, and are briefly described as follows: prevention which involves
activities designed to provide permanent protection from hazards such as hazard and
land cover mapping and vulnerability assessment; preparedness, which involves
activities designed to minimize loss of life and damage including hazard warning;
and relief, which involves assistance and/or intervention during or after hazard. The
possible contribution of remote sensing could be focused on relief and, possibly,
preparedness or warning although in many cases remote sensing can make a valuable
contribution to disaster prevention, where frequency of observation is not such a
prohibitive limitation.
A major consideration for development of remote sensing for drought assessment
and disaster reduction is the extent to which operational users can rely on a continued
supply of data. There are two types of remote sensing systems for drought
assessment, namely meteorological and environmental (or resource) satellites.
Meteorological satellites are usually operational, since there is a commitment to
continually provide data. Besides weather forecasting, meteorological satellites have
found application in several other important hazard applications mainly due to the
high frequency of coverage and moderate resolution. The two meteorological
satellites, namely METEOSAT and NOAA/AVHRR, can contribute to operational
monitoring and assessment of drought. In addition, environmental satellites such as
LANDSAT, SPOT and recently Ikonos, WV2 with high to very high resolution, but
low frequency of coverage, can contribute to land-use classification and qualitative
features of drought and less to quantitative assessments.
Monitoring the extent of drought is best achieved in near arid areas by the extent
of vegetation. This can be done by multispectral visible imagery from polar orbiting
satellites. In particular, the Normalized Difference Vegetation Index (NDVI) of the
visible channels (Ch1 and Ch2) of NOAA/VHRR is effectively used (Kogan, 2002).
The technique can be calibrated against biomass and give good guidance on
extending drought affected areas. Soil moisture can be directly measured in the
microwave region and interpretation of Synthetic Aperture Radar (SAR) data may
provide some information on soil moisture.
2.2. Remotely Sensed Drought Indices
If drought is considered as a phenomenon, it is certainly an atmospheric
phenomenon. However, if drought is considered as a hazard, there is a tendency to
classify drought types into three categories, namely meteorological or climatological,
173
agricultural and hydrological drought and to include as a fourth class the
socioeconomic impacts of drought (Keyantash and Dracup, 2002). Table 1 shows a
list of satellite-based drought indices for the quantification of drought.
Table 1. Satellite-based Drought Indices
1. Normalized Difference Vegetation Index
2. Deviation NDVI index
3. Enhanced Vegetation Index
4. Vegetation Condition Index
5. Monthly Vegetation Condition Index
6. Temperature Condition Index
7. Vegetation Health Index
8. Normalised Difference Temperature Index
9. Crop Water Stress Index
10. Drought Severity Index
11. Temperature - Vegetation Dryness Index
12. Normalized Difference Water Index
13. Reconnaissance Drought Index
In this paper, RDI is used based on remote sensing. The Reconnaissance Drought
Index (RDI) is a new index, which is used for hydrorneteorological drought
estimation (Tsakiris and Vangelis, 2005). RDI is a physically-based and general
index and can be used in a variety of climatic conditions. Moreover, RDI provides
information for the water deficit in an area as it is based not only on precipitation, but
also on potential evapotranspiration. In order to assess and monitor drought, it is
necessary to detect several drought features. Moreover, remote sensing data and
methods can delineate the spatial and temporal variability of several drought features
in quantitative terms.
3 Classification of Drought Severity
The classification of drought severity is achieved through the combination of
several drought features using remote sensing data. A brief summary of the study
area and description of drought features follows.
3.1 Study Area
The region of Thessaly overtakes the central - Eastern department of continental
Greece and overtakes a total areal extent of 14036 Km2 (10.6% of total extent of the
174
country). The 36.0% of ground are in a plain, the 17.1% semi-mountain, while the
44.9% is mountainous. High mountains surround the plain of Thessaly, which
constitutes the bigger plain of the country that divides westwards to Eastern from the
river Pinios that is the third bigger river of country. Pinios River is springing the
western slopes of Pindos Mountains and outflows, after 216 km, in the Aegean Sea.
Its main tributaries are Titarisios, Enipeas, Kalentzis and Litheos. Surface and
groundwater resources are jointly used to cover rural, urban and industrial needs,
whilst on the same time they are essential to the preservation of the wetland
developed in the area. The main watershed in Thessaly water district is the Pinios
basin which covers 9500 km2. Thessaly is one of the thirteen hydrological districts of
the country and it is located in Central Greece (Figure 1). Thessaly plain is a drought-
prone area, which is also the main agricultural region of Greece.
Fig.1. Location and geographical map of Thessaly region .
At the western side of Thessaly the climate is continental; the winters are cold and
the summers are hot and the temperature difference between the two seasons is large.
At the eastern side of Thessaly the climate is typical Mediterranean. Summers in
Thessaly are usually very hot and dry, and in July and August temperatures can reach
40o C. Mean annual precipitation over the whole Thessaly region is about 700mm
and it is distributed unevenly in space and time. The mean annual precipitation varies
from about 400mm at the central plain area to more than 1850mm at the western
mountain peaks.
3.2 Drought Features
The major drought features are defined as follows:
175
Severity: severity or intensity of drought consists of the classification and escalation
of the phenomenon from mild to moderate, severe and extreme. The severity is
usually determined through drought indicators and indices, which include the above
mentioned classes.
Duration: duration of a drought episode is the time interval from the start and end
time usually in months. Since drought is a complex phenomenon and hazard the
assessment of start and end time is a complicated issue.
Onset: the beginning of drought is determined by the appearance of drought episode.
The beginning of droughts is assessed through indicators or indices reaching certain
threshold value.
End time: end time of drought episode signifies the termination of drought based
again on threshold values of indicators or indices.
Areal extent: areal extent of drought is considered the spatial coverage of the
phenomenon as it is quantified in classes by indicators or indices. Areal extent varies
in time and remote sensing has contributed significantly in the delineation of this
parameter.
Table 2. Drought categories based on RDI, VHI and PDSI
VHI
Drought Categories RDI Values PDSI Values
values
Incipient dry spell - - -0.5 to -0.99
Mild drought 0 to -0.99 <40.0 -1.0 to -1.99
Moderately Dry -1.00 to -1.49 <30.0 -2.0 to -2.99
Severely Dry -1.50 to -1.99 <20.0 -3.0 to -3.99
Extremely Dry <-2.00 <10.0 < -4.0
4 Methodology
The methodology covers the estimation of RDI based on remote sensing data and
techniques. Specifically, the methodology follows the described steps, which include
prepossessing of satellite data, calculation of air temperature, estimation of potential
evapotranspiration with the use of satellite data, rain map extraction and remotely
sensed estimation of RDI. A brief description follows.
4.1 Data base and Prepossessing of satellite data
For the RDI estimation the following data is used:
Ø Daily precipitation of Thessaly water district in 50 x 50 km2 spatial analysis
derived by ground measurements provided by the .Joint Research Center (JRC) of
EC, lspra, ltaly.
Ø Crop coefficients maps extracted by Corine Hellas 2000 for each month of the
176
year.
Ø Monthly maps of daytime sunshine duration for 39.39° Middle North Latitude of
Thessaly.
Ø A time series of ten-day Brightness temperature (BT) images extracted from
Channels 4 and 5 of NOAA/AVHRR for 20 consecutive hydrological years (October
1981 - September 2001) 8x8 km2 provided by NOAA. The variables which are
extracted from satellite data are Brightness Temperature (BT) and Normalized
Difference Vegetation Index (NDVI) on monthly time step. Next step is the
geometric and atmospheric correction of all images with the use of software Erdas
Imagine.
4.2 Calculation of air temperature
Air temperature maps are derived from LST satellite images based on regression
analysis between LST values and ground measurements of air temperature from
meteorological station of Larissa, which is located in the region. LST is calculated
with the use of BT and NDVI images on a pixel basis (Kanellou et al., 2008). The
derived empirical relationship between LST and air temperature (Tair) is given by:
Tair = 0.6143 - LST + 7.3674 R 2 » 0.82 (1)
4.3 Estimation of potential evapotranspiration with the use of satellite data
The RDI uses precipitation and potential evapotranspiration. In this paper
potential evapotranspiration is estimated with the use of Blaney-Criddle method. This
method is selected as it is appropriate for subtropical climates with dry and hot
summers such as Mediterranean region, since it has been applied in California,
instead of Thornthwaite method, which is more appropriate for climates with wet and
hot summers (e.g. East U.S.A). Blaney and Criddle (1950) estimated the monthly
potential evapotranspiration (ET m) in mm, by the equation (2):
ETm = k * [0.46T + 8.16]* p (2)
where T is the mean monthly air temperature, p is the monthly daytime sunshine
duration, which depends on the latitude of the area, and k is the crop coefficient,
different for each cultivation, vegetation type, season and land use.
Maps of mean monthly crop coefficients for each vegetation type and land use in
500x 500 m2 pixel size, as well as maps of daytime sunshine duration (p) for each
monthly value for the Thessaly water district (39,39° North Latitude) are extracted in
a GIS environment (ArcMap 9 .1. software) (Kanellou et al., 2008).
The monthly crop coefficient and the maps of daytime sunshine duration are
combined with the air temperature maps for the whole data set in order to extract
177
Blaney-Criddle potential evapotranspiration for each month in the time series (1981-
2001).
Fig. 2. Crop coefficient map (August). Fig. 3. Blaney-Criddle ETp for June 1982.
4.4 Rain map extraction
For the estimation of RDI it is required to estimate monthly areal precipitation.
Rain maps over Thessaly on a monthly basis are provided by JRC, ISPRA. These
data cover Greece from 1975 to 2005 per 50x50 km2. From daily values of all time
series the monthly cumulative rain of each hydrological year from 1975 to 2005 is
calculated. Then rain maps produced every month using linear interpolation.
4.5 Remotely sensed estimation of RDI
Estimation of RDI is achieved with the use of models in software Erdas Imagine,
in which are used monthly temperature maps, crop coefficient (Kc) maps, sunlight
maps (p), potential evapotranspitration Blaney- Criddle (ΕΤp) maps and rain maps
(P). In this study, RDI is calculated on a monthly and annual basis
The calculation of the index starts with the estimation of a k coefficient (Tsakiris
and Vangelis, 2005), as it is given by the equation:
j =k (3)
åP
j =1
j
a k = j =k
å PET
j =1
j
where Pj and PETj are the precipitation and potential evapotranspiration,
respectively, of the j-th month of the hydrological year. The hydrological year for the
Mediterranean region starts in October, hence for October k=l.
178
RDIn is the Normalised RDI, which is given by:
ak (4)
RDI n (k ) = -1
ak
The Standardised RDI (RDlst) is given by:
yk - yk (5)
RDI st (k ) = ^
sk
^
where yk is the ln ak, y k is its arithmetic mean and s k is its standard deviation.
5 Results and Discussion
The results include quantification of drought through RDI estimation on a monthly
basis (1981-2001) using remote sensing data, as well as extraction of several drought
features. The analysis results are presented in tables 3 and 4 and figures 4 and 5. In
particular, Table 3 presents the drought periods, their duration along with the start
and end times based on RDI monthly estimates. This table indicates that during the
20-year period there are eight drought episodes lasting one hydrological year, with
minor exceptions. Moreover, Table 4 presents the areal extent of the monthly RDI
values including all the drought classes in terms of number of pixels for each drought
episode.
From Tables 3 and 4 it can be stated that the eight drought episodes are classified
in two severity classes, namely mild to moderate drought (five cases) and severe to
extreme drought (three cases). The next step consists of plotting the values, which are
less than -2.0 of Table 4 in two separate figures, one for the mild class and another
for the extreme class, respectively, along with a fitted curve for each plot.
Table 3. Duration of drought episodes (in months in Thessaly)
Drougth Years Start End Duration
Oct 1984 - Oct 1985 Oct 1984 Oct 1985 13
Oct 1987 - Oct 1988 Oct 1987 Oct 1988 13
Sep 1989 - Oct 1990 Sep 1989 Oct 1990 13
179
Oct 1991 - Sep 1992 Oct 1991 Sep 1992 12
Oct 1992 -Oct 1993 Oct 1992 Oct 1993 13
Oct 1996 - Sep 1997 Oct 1996 Sep 1997 12
Oct 1999 - Sep 2000 Oct 1999 Sep 2000 12
Oct 2000 - Sep 2001 Oct 2000 Sep 2001 12
Table 4. Monthly Areal Extent of Drought years for the period 1981-2001. The values are in
number of pixels (each pixel= 8x8 km2)
Years 1984- 1987- 1989- 1991- 1992- 1996- 1999- 2000-
1985 1988 1990 1992 1993 1997 2000 2001
Month
Oct 207.704 0.039 75.214 81.906 111.175 2.589 25.984 16.675
Nov 84.984 0.234 167.91 163.812 128.343 207.769 8.398 207.691
Dec 95.527 66.21 23.019 205.765 203.183 45.351 74.156 205.273
Jan 82.093 86.312 207.714 207.648 199.039 72.425 199.3 40.957
Feb 2.441 0.238 207.789 207.496 45.347 199.359 22.148 164.46
Mar 0.265 0.218 207.789 203.925 199.718 201.363 207.417 207.617
Apr 167.148 127.707 159.589 0.222 206.437 0.41 203.73 4.847
May 125.921 196.152 2.089 12.472 0.105 207.753 185.269 14.039
Jun 205.144 59.792 161.484 8.48 114.875 16.73 12.289 133.984
Jul 128.741 60.468 40.132 8.386 201.535 146.554 171.257 0.089
Aug 114.992 130.746 0.25 203.839 191.9609 0.128 193.574 49.066
Sep 123.332 83.679 77.0234 204.066 81.285 203.824 75.566 199.304
Total 1338.27 811.8 1330.007 1508.023 1683.007 1304.261 1379.093 1244.007
The results are shown in Figures 4 (extreme class) and 5 (mild class), respectively.
It is interesting to notice that in Figure 4 the fitted curve starts at the beginning of the
hydrological year with at least 50 pixels and the areal extent is also increasing
throughout the year. On the other hand, in Figure 5 the fitted curve starts in spring
with even less than five pixels and the areal extent is also increasing throughout the
180
year, but not reaching high values. It is, thus, evidence in Thessaly that droughts can
be classified in two classes, namely mild to moderate and severe to extreme,
respectively, and that the former starts in spring, whereas the latter starts in October.
This finding can certainly be used for prognostic assessment purposes. It simply
requires more cases for verification.
Fig. 4. Cumulative Areal Extent (Number of pixels 8X8 km2) of extreme drought (<-2.0)
during drought episodes based on remotely sensed RDI
Fig. 5. Cumulative Areal Extent (Number of pixels 8X8 km2) of mild drought (<-2.0) during
drought episodes based on remotely sensed RDI
181
6 Conclusions
In this paper drought quantification is conducted through the estimation of remotely
sensed monthly RDI (1981-2001) in Thessaly, central Greece. Moreover, the remote
sensing potential is explored by assessing several drought features towards drought
severity classification. The results are encouraging since drought episodes are
classified in two distinct classes with prognostic ability for each class.
Acknowledgements: This research was funded by Pleiades, Smart and Hydrosense
EC projects. The conventional meteorological data was provided by the National
Meteorological Service of Greece. The precipitation maps were provided by the Joint
Research Center (JRC) of EC, Ispra, Italy. The satellite data was provided by NOAA.
References
1. Blaney, H.F. and W.D. Criddle (1950). Determining water requirements in
irrigated areas from climatological and irrigation data. USDA Soil Conservation
Service, Technical Paper, No. 96, 48p.
2. Heim, R.R. Jr. (2002). A review of twentieth- century drought indices used in the
United States. Bulletin of the American Meteorological Society, Vol. 83(8), pp.
1149-1165.
3. Kanellou, E., C. Domenikiotis, E. Tsiros and N.R. Dalezios (2008). Satellite-
based Drought Estimation in Thessaly. European Water 23/24:11-122.
4. Keyantash, J. and J. A. Dracup (2002). The quantification of drought: An
evaluation of drought indices. Bulletin of American Meteorological Society, pp.
1167-1180.
5. Kogan, F.N. (2002). World droughts in the new millennium from AVHRR-based
Vegetation Health Indices. EOS Transaction, American Geophysics Union, No.
83 (48), pp. 562-563.
6. Loukas A., L. Vasiliades and N.R. Dalczios. 2002. Hydroclimatic Variability of
Regional Droughts in Greece Using the Palmer Moisture Anomaly Index. Nordic
Hydrology: 33 (5): 425-442.
7. Rossi, G., 2000. Drought mitigation measures: a comprehensive framework. In
Drought and Drought Mitigation in Europe. J. Voght and F. Somma (eds),
Kluwer Academic Publishers. Dordrecht.
8. Thenkabail, P. S., M. S. D. N. Gamage and V. U.Smakhtin (2004). The use of
remote sensing data for drought assessment and monitoring in southwest Asia.
Research Report, International Water Management Institute, No. 85, pp. 1-25.
9. Tsakiris, G. and H. Vangelis (2005). Establishing a drought index incorporating
evapotraspiration. European Water, 9/10, pp. 3-11.
182