Comprehensive Management Scheme (adapted from Yevjevich et al ., 1983; Grigg & Vlachos, 1990, 1993; Karavitis, 1992, 1999a). 

Contexts in source publication

Context 1

... The SPI 6 and 12 have been calculated country-wise. For that estimation, precipitation data from 46 stations were collected in collaboration with the National Meteorological Service of Greece (HNMS), the Ministry of Public Works (MPW) and the Public Power Corporation (PPC, S.A.), covering different time periods from 1947 to 2009 ( Figure 4). All the precipitation data were converted to monthly values. All the chosen precipitation stations exhibit good data quality, according to the main criterion for such a selection, namely the existence of minimal data gaps in the time series. The criterion also that the SPI values should be estimated from a time period of at least 30 years was fulfilled, since all the stations used provided such data time series (McKee et al ., 1993). No attempt to fill in any existing data gaps was made, since as stated above the gaps were minimal and it was considered that raw data may be more appropriate to represent natural drought conditions (extreme minimum values) rather than enforcing ‘ corrective ’ homogeneity (Karavitis et al ., 2011b). The monthly precipitation data were used as input for the SPI calculation tool (DMCSEE Project, 2009). The tool also has a built in capability to treat the data gaps in the precipitation time series and not to use them in the final computation. In the current effort, the SPIs for January and August for 1990, 2007 (drought years) and 2009 (normal year) were selected to be presented and compared. The geo-statistical method of kriging was chosen for the spatial distribution and intensity of the drought according to the SPI values, since precipitation is a natural parameter and changing the point original values may be acceptable (Karavitis et al ., 2011b, 2012a, b). The kriging method (ordinary kriging) and the models to be followed (hole effect, exponential and spherical) were examined and selected according to pertinent statistical parameters. The calculated parameters were the root mean square error and the average standard error, whose values should be as low as possible in order to accept the corresponding surface. From the assessed statistical parameters the surfaces that visualized more appropriately the SPI values were mostly those created by the hole effect model (Karavitis et al ., 2012a, b). Thus, the SPI values were spatially visualized using ordinary kriging (hole effect model) in an ArcGIS 10 environment and are presented for SPI 6 and SPI 12 in Figures 5 – 7. Additionally, it is pointed out that the SDVI may still be estimated based on SPI values, even if such values are indirectly extrapolated using the kriging process for any specific points or areas, where precipitation data (for the direct SPI calculation) are not available, as long as data on the remaining indicators are present. 2. An additional 28 extra locations (apart from the 46 station locations used) were also selected to serve primarily as sign posts and are presented in Figure 4. Fifteen of such locations are high mountain peaks (higher than 2,000 m a.s.l.), and the surrounding mostly barren areas, where the final SDVI estimate has axiomatically appointed a zero value since no significant economic or social aspects could be mea- surably affected by drought. The remaining 13 locations are key areas of pivotal importance, where crucial economic and social activities are present. In such areas, the above-mentioned extrapolated SPI estimates on step 1 were used for the SDVI calculation. It is believed that these premises may lead towards a more representative and suitable ‘ calibration ’ of the SDVI estimation, since the initial runs of the procedure without the forced values in the selected locations have produced some results deviating from the real conditions. This fact was particularly evident in the outcomes visualization, where the algorithm used had assigned vulnerability values in areas of no actual vulnerability to drought (e.g. mountain peaks) and viceversa. However, in regions where such knowledge is not available, the assigned values without the sign posts may suffice, since they may represent reasonable enough approximations. 3. Impacts present nascent difficulties that need to be taken into account independently of geography, climate and political boundaries (DMCSEE, 2012; Grigg, 2014). The fragmentation of mostly anecdotal reports does not lend itself to classify the drought impacts in economic, social and environmental categories. Instead, most reports are about agricultural impacts, and information about the other categories is dispersed (Grigg, 2014). Data on water demand, water supply, pertinent water infrastructure and drought impacts have been gathered for 59 of the 74 locations (not including the 15 mountain peaks). More specifically, data time series covering different time periods from 1982 to 2009 on supply network losses per water district, water availability and average water demand/consumption per capita, and supply and demand deficits have been collected from the Water Resources Management Plans of the River Basins in Greece (Special Secretariat for Water (SSW), 2013). These studies were produced using MIKE SHE for the hydrological conditions and MIKE HYDRO BASIN for the management options. These Decision Support Systems are products of the Danish Hydrologic Institute (DHI). Infrastructure information was based on the difference between the designed and the actual capacity of the country ’ s reservoirs (dams) as well as on the reported operational conditions of the water supply infrastructure at any given year including the drought ones (Ministry of Infrastructure, Transport and Networks (MITN), 2013). The data reflected various time periods from 1964 to 2012. Impact data have been also acquired from mass media archives, from reduction percentages of the agricultural production for the drought years and from archive information on various drought impacts and aspects of the corresponding local and national authorities and agencies, all dating from various time intervals between 1950 and 2012 (Karavitis, 1999b; DMCSEE, 2012; Hellenic Statistical Authority (HSA), 2013). Table 4 presents a sample of mass media drought data for Greece in 2007 and Table 5 illustrates part of the applied drought responses assessment based on Figure 2 and Table 2. All such data were used to produce the described scaled values according to Table 3. 4. The SDVI value per selected area and month has been calculated according to Equation (3). Then, the produced values were classified into the vulnerability classes demarcated in Table 3. Finally, the SDVI has been visualized using the Inverse Distance Weighting (IDW) in Geographical Information System (Figure 8). The outcomes for both the index performance and drought vulnerability in Greece are examined and analyzed in the following section. The IDW was chosen instead of kriging ...

Context 2

... deficiency created usually by human activities. Finally, desertification is prin- cipally a man-made phenomenon altering significantly the ecological regime. It has been suggested that all the above terms and definitions associated with dryness may be considered as a part of a larger process named: ‘ xerasia ’ (Figure 1). The boundaries among these four categories are gradual depicting their interdependencies and complex nature signifying, for example, that a drought may also have not only natural but also some anthropogenic connections. Nevertheless, whatever the term and the overall context, drought should be associated with its impacts at a given locale. Such association including special technological, economic and societal traits may estimate the area ’ s vulnerability to various ‘ drought ’ manifestations. Drought has short- and long-range (usually cumulative) impacts in virtually all types of activities related to water, economy and society. Two methodological approaches may be underlined in order to study and assess drought impacts. In the first one, the impact approach, a climatic event (drought) operates on a certain ‘ exposure unit ’ (activity) producing an impact. This is a cause and effect approach. The second one, the interaction approach, suggests that various processes (physical, economic or societal) may influence the ‘ exposure unit ’ and the impacts are embedded and interrelated to the ‘ exposure unit ’ . In other words, environment, policies, economy and society combined negatively on a given activity may create a crisis. In recent decades, the interaction approach started to be considered as more realistic by presenting the impacts as ‘ orders of interactions ’ (Wilhite et al ., 1987; Karavitis, 1992; Grigg, 1996). In this regard, a classic first broad categorization of impacts was in a series of first-, second- and third-order impacts (Changnon & Easterling, 1989). First-order impacts are associated with changes related to the hydrologic cycle (i.e. precipitation, runoff, stream flow and groundwater). Second-order ones usually influence human activities such as agriculture, industry, urban users and transportation. Finally, third-order impacts may be understood as adaptations to first- and second-order impacts (i.e. income losses, adjustments in life style and rationing). Such impact dis- tinctions are extremely important so as to produce a drought management methodology. At the same time, drought impacts should be categorized according to a concise and comprehensive framework. Thus, any classification scheme becomes crucial, since it may lead towards potential drought responses in an implementable decision-making process. In general, the impact ’ s magnitude on an area is affected by the density of human activities, needs, demands, level of socio-economic structure and the environmental linkages (Eriyagama et al ., 2009). The 2012 drought in the USA produced severe hydrological, economic, environmental and social impacts and may be classified as the worst since the 1930s Dust Bowl (Grigg, 2014). During the 1989 – 1990 great- est drought on record for Greece, the impacts were devastating, as losses escalated to about 1.5 billion (10 9 ) USD in 1990 prices (Karavitis, 1992, 1999b). Wu et al . (2011) provide a short but quite explanatory description of drought impacts emphasizing the great losses, mostly economic, that occur during such an event, while Ding et al . (2010) provide a more detailed one of the drought economic impacts. Impacts trigger the societal responses to drought. The more holistic the responses, the more effective the drought mitigation may be. Hence, integrated water resources management (IWRM) should be used as the general context for comprehensive drought management approaches. The articulation of marks and threshold drought conditions can measure progress, performance and products of such management approaches (Grigg, 1996, 2008; Karavitis, 1999a; Vlachos & Braga, 2001). The major challenge for any drought mitigation policy in order to confront the impacts may be the development of comprehensive and effective drought management schemes. Such schemes should be based on proactive strategies incorporating pre-drought planning, drought responses and post-drought activities (Grigg & Vlachos, 1990, 1993; Karavitis, 1992, 1999a; Karavitis et al ., 2012a, b). If impacts are anticipated, then a responses plan may be set in advance. The core of a scheme for a drought responses plan may be composed from short- and long- range responses. Short-term responses should be initiated and terminated according to the drought duration, while long-term ones should be designed and implemented in advance of a drought event. Thus, impacts should be anticipated both spatially and temporally and initiate management interventions on certain vulnerability thresholds. In other words responses may also lead to impact classification. An image of a comprehensive management scheme is presented in Figure 2. Given such considerations, a drought responses plan should be classified in the following parts (Yevjevich et al ., 1983; Grigg & Vlachos, 1990, 1993; Grigg, 1996; Karavitis 1992, 1999a, 2012): Supply augmentation measures . Such measures should examine all the potential water supply resources for the area. They should already be in place before a drought (base and emergency supply). Perhaps with the exception of water purchases, systems supply augmentation should be avoided during the drought as a crisis management action. The existing system designed after long-range planning should be capable of operating under drought conditions according to contingency plans; it should also be well maintained and improved in order to minimize the losses: Demand management/reduction measures . These responses should aim towards water consumption patterns according to conservation principles. The long-term measures should be in place according to proactive planning (legal measures, zoning/land use, landscape changes, agricultural changes such as changing to less water consumptive crops, irrigation scheduling, etc.). The short-term measures should be initiated during and terminated after the drought (water restric- tions, reduction of uses, pricing, etc.). The implementation and enforcement framework for demand reduction measures should also be in place (economic, legal and institutional). All in all, such measures should be implemented orderly and timely according to contingency plans; and finally Impact minimization . Such responses should concentrate on anticipatory strategies, relief and recovery measures. The framework for such responses should already be in place (economic, legal and adminis- trative). Spread of drought risk, damage recovery and compensation should be some of the measures considered, according to a drought master plan. In order for any responses to be applied, existing problems unquestionably must be resolved about the onset, the areal extent and the severity of a drought. In this quest drought index methods may be used. These methods characterize a drought according to a specific index. Thus, a drought index should primarily be an objective measure of the system status that may help in identifying the onset, increasing or decreasing severity and termination of a drought. Nevertheless, no single indicator or index alone may precisely describe the onset and severity of the event. Numerous climate and water supply indices are in use to present the severity of drought conditions. In the literature different indices have been discussed and applied, with the SPI (McKee et al ., 1993), the Palmer Drought Severity Index (Palmer, 1965) and the Crop Moisture Index (Palmer, 1968) being three most usually applied. Although none of the major indices is inherently superior to the rest in all circumstances, some indices are better suited than others for certain uses (Karavitis et al ., 2012a, b). All in all, the type of index, local conditions, data availability and validity usually lead to the index selection (Nardo et al ., 2005; Singh et al ., 2009; Skondras et al ., 2011; Rogge, 2012). However, apart from those event-related indices, a few more complex ones were developed, some of which refer to the vulnerability concept (Briguglio & Galea, 2003; Pratt et al ., 2004; Fussel, 2010; Ganase & Teelucksingh, 2011). The following indices serve as examples (Kaly et al ., 2004; Skondras et al ., 2011; Ganase & Teelucksingh, 2011): The Composite Human Vulnerability Index – by the Indian Institute of Technology in Bombay; The Key Indicators for Global Vulnerability Mapping – by the United Nations Environment Programme; The Coral Reef ‘ Vulnerability Index ’ of Exposure to Climate Change – by Green- peace; The Environmental Vulnerability Index – by the South Pacific Geoscience Commission; and the Climate Vulnerability Index – by Sullivan & Huntingford (2009). Increasingly, the term ‘ vulnerability ’ appears in the environmental change literature (Adger, 2006; Gallopin, 2006; Janssen, 2006; Janssen & Ostrom, 2006). It is associated with the evolution in environmental studies from impact analysis, to crisis assessment, to vulnerability evaluation reflecting the large number of critical variables involved, cumulative consequences and multi-dimensional sources of threats, hazards and unanticipated consequences. Even more vulnerability has been tied to security in all its forms, such as food security, economic security, environmental security and political security, all the way to individual security (Adger & Kelly, 1999; Adger, 2000, 2006). This dynamic evolution coincides also with the transformation from simple, linear models, to more complex, potentially circular, feedback, heterarchical and non-linear approaches. When combined with ‘ volatility ’ it becomes the current potent theme of expanding time-scale units of analysis and assessment. Nevertheless, the term vulnerability as well as its relative ones – ...