CHARACTERIZATION OF DROUGHTS IN MID-CLIMATIC REGIONS OF INDIA

Abstract

Droughts may be characterized by their frequency, intensity, duration, and severity, which vary across the climatic spectrum. Among the regional climatic parameters governing occurrence and characteristics of droughts, the most common are precipitation and its distribution over the year or the length of wet season and temperature and, in turn, evapotranspiration. The mid-climatic regions ofIndia refer to the platue regions, the arid regions, and parts of northern plains. The study has been carried out with the major objectives as to define climatic regions particularly useful for drought characterization, to develop relationship between climatic parameters and the drought characteristics, to determine the governing morphological factors which influence the drought characteristics, to propose indices for assessment of hydroiogical drought severity using streamflow and to propose a method for estimation of water storage requirement to restrain adversity ofdroughts. In this study, a new climatic classification is proposed based on (a) the ratio of mean annual precipitation (Pa) to global terrestrial mean annual precipitation (Pg) (b) the ratio of mean annual potential evapotranspiration (Ep) to mean annual precipitation (Ep/Pa), and (c) length of wet season. The critical value of Pg (= 800 mm) was estimated using weighted averages of precipitable water over terrestrial regions. The middle of the climatic spectrum was defined as Pa/Pg = 1. The regions with less than average moisture and more than average moisture were defined as Pa/Pg <1and Pa/Pg >1, respectively. Based on the values of Pa/Pgand Ep/Pa, the climatic spectrum was divided into eight regions, i.e., super-arid, hyper-arid, arid, semiarid, sub-humid, humid, hyper-humid and super-humid. The relevant arid, semiarid and subhumid climatic regions were categorized as 5<Ep/Pa <12, 2<Ep/Pa <5and 0.75 <E„/Pa <2, respectively. These climatic regions have a wet season length of approximately two, three and four months, respectively. This form was derived specifically for drought characterization across the climatic regions, and notably, it closely matched with other existing classifications. In view of the range of Epand Pa in different parts of India, proposed classification appeared to be rational. For further analysis annual rainfall series during the period 1901-2006 for 110 stations located in different regions in India were analysed using percent annual rainfall departure to identify drought years and drought events. Potential evapotranspiration rates were computed using 30 years of daily meteorological data from various stations. The average drought return period was related with the Ep/Pa ratio using power and exponential regression models. Power type regression showed better correlation than did the logarithmic or exponential type regression. The relationships indicated the frequency of meteorological droughts to be greatly governed by the Ep/Pa ratio. Average drought return period varied from 2 to 3 years in arid regions (with 12 > Ep/Pa > 5), 3 to 5 years in semiarid regions (with 5 > Ep/Pa > 2) and 5 to 9 years in subhumid regions (with 2 > Ep/Pa > 3/4). Since, the mean annual deficit (Ep - Pa) may be zero or less than zero. Therefore, the relationship between the average drought return period and the ratio of mean annual deficit (Ep - Pa) to mean annual precipitation (Pa) was derived using an exponential regression. These relations revealed that the average drought frequency increases with increase in the Ep/Pa ratio. In arid and semiarid regions it decreases gradually from once in 2.5 years on an average to once in 5 years for a long range of Ep/Pa ratio, i.e. 2-10. However, in wet sites (i.e. sub humid and humid regions), it decreases sharply from once in 5 to once in 11 years for a short range of Ep/Pa ratio, i.e. 0.5-2. Both the relations clearly indicated drought frequency to decrease exponentially with decrease in the mean annual deficit (Ep - Pa). Similar relations of Ep/Pa ratio with duration and intensity of drought were derived. The incidence of drought events of 2 and 3 consecutive years was relatively more frequent in areas with Ep/Pa ratios of 4.5-7.0 and 7.5- 8.5, respectively. The analysis revealed that the areas with greater Ep/Pa ratio faced relatively more frequent severe intensity droughts than did the others. Thus, vulnerability to severe drought increases with increase in Ep/Pa ratio. These inferences were consistent with those for Australia, Brazil, United States, Ukrain & Kazakstan, Moracco etc. Since the relations closely followed the drought frequency behaviour in similar climatic regions, they are expected to be quite useful in systematic drought analysis for different climatic regions. Implication of elevation, of distance from mountain/major hills and of distance from sea coast could be observed on drought intensity and drought persistence (i.e. duration). However no specific pattern could be established among them. Typically, it was found from the data that stations, falling in regions located at relatively short distance from sea (ds) and mountain, and having Pa~Ep, hardly faced drought events of severe/extreme intensity, and droughts of 2 or more consecutive years hardly persisted in such areas. Thus, the study revealed that, besides the climatic parameters, drought characteristics at a given location are also influenced by the distance from sea and mountain barrier as well. Stream flow drought severity analysis was carried out for Ken and Betwa rivers in central India using 22-42 years (1960-2001) 10-daily streamflow data ofnine discharge measurement sites. Variable truncation level approach was used to identify the streamflow drought events. Using a pooling approach based on inter-event critical duration and critical deficit volume ratio criterion, severity and duration of independent drought events were computed. The degree ofseverity ofan independent drought event was classified using a new drought severity index (DSIe) defined as a function of (i) the ratio of deficit flow volume to corresponding volume at the truncation level and (ii) the ratio ofduration ofdeficit flow to the maximum possible duration of the independent streamflow drought event. The proposed method appropriately described the hydrological droughts in studied basins. The results indicated that the drought events starting during August-November were likely to be more severe than those in the other months, the upper reaches ofriver course ofstudy basins suffered with greater degree of severity of streamflow drought than those in lower reaches, indicating the upper reaches to be more exposed to severe hydrological droughts. The derived duration and severity of streamflow drought events for Ken River system at Banda site were used for stochastic modeling by theory of runs (run-sum and run-length). The autocorrelation function was used to identify the timeindependence ofthe series. The parent distribution function was selected by adopting the L-moments ratio diagram based on the estimated values ofL-Cs and L-Cki and the parameters of the selected distribution function were estimated by L-moments and probability weighted moment (PWM) methods. The extreme severity of drought event, total deficit volume and maximum drought duration during the monsoon period were considered for fitting distribution functions for hydrological drought forecasting. The time-independent series of extreme severity of drought events, total deficit volume and maximum drought duration during the monsoon period fitted reasonably well with the Pearson Type-Ill (PT-III) and GEV distributions and might be used for forecasting drought severity and duration for different return periods in the study basin. in Further a methodology was proposed to assess surface water storage requirement using carryover factor for supporting usual activities during drought. Assuming that, if flow in the stream is maintained at truncation level during a drought period, it may support the sustenance of normal life pattern in the basin. The analysis was carried out for two major tributaries of the Ken river, Sonar & Bearma rivers. The study indicated that the hydrological droughts usually began in considered basins during August-October months. These events either got terminated during the period between September-December or, if not, then they usually continued till the onset of the next monsoon. Drought events starting during early monsoon months were found to be more severe than those starting during the late or post monsoon months. Analysis further showed the availability of sufficient water in the considered basins for drought mitigation through surface storage schemes. Considering maximum streamflow drought severity values, a methodology was formulated for estimation of volumetric requirement of artificial surface storages to protect the basins' usual activities during a drought. The results of this study can useful in water resources planning paticularly for water conservation and management. The study finally proposed a new classification of climatic regions to describe drought characteristics, relationships of meteorological drought characteristics with climatic parameters (ratios of Ep/Pa and (Ep-Pa)/Pa), a new streamflow drought severity index, and a method for estimation of surface water storage requirements for drought mitigation. The results of the study are expected to enhance the understanding for a systematic analysis of drought characteristics and help in evolving appropriate drought management strategies.