OBJECTIVE ASSESSMENT OF INDICES AND VULNERABILITY TO DROUGHT

Abstract

Drought is a normal feature of climate and affects more people than any other natural erally linked to deficit precipitation A hazard on the earth. The occurrences of droughts are gen low soil moisture, deficit river flows and depletion of groundwater as compared to their corresponding normal values in space and time. In general, drought is defined as an extended period - a season, a year or more - of deficient rainfall compared to the statistical multi-year average for a region that injuriously affects the usual crops and causes temporary scarcity of water for human and livestock consumption. Deficiency in rainfall is the primary cause of drought and commonly referred to as meteorological drought.The continuation of meteorological drought leads to deficiency of soil moisture causing agricultural drought. Subsequent reduction in river flows and depletion of groundwater below their corresponding 40 normal values is referred to as hydrological drought. Reasonable assessment of drought characteristics viz, frequency, duration and severity - both in space and time - are needed for identification of vulnerable areas and planning of themitigation strategieSto cope with drought event. In the present study, an attempt has been made for the assessment of meteorological, hydrological and agricultural drought characteristics using various drought indices in Ken river basin, located in drought prone region of central India. The effect of length of data on assessed drought characteristics is also examined using long term rainfall and stream flow data. An integrated drought vulnerability index (IDVI) is also proposed for assessment of degree of vulnerability to drought using meteorological, hydrological, physiographic and social factor in spatial and temporal domain. The methodologies applied and major findings from the study are summarized below. ASSESSMENT OF REGIONAL RAINFALL ChARACTERISTICS The regional rainfall characteristics of eight districts have been assessed using 102 years area weighted monthly rainfall data from year 1901 to 2002. The district wise rainfall departure analysis has been carried out with respect to the long term annual average rainfallof the corresponding district. The analysis reveals that the annual rainfall deficiency in various districts during first half (1901 to 1950) of the study period ranges from 31% (in Katni District in the year 1941) to 45 % (in Hamirpur and Sagar District in the year 1913) with an average deficiency of 37% for the Ken Basin as a whole. However, during the latter half of the study period (i.e. 1951 to 2002), the deficiency varied between 41% (in Katni District in the year 0 1993) to 53% (in Banda District in the year 1991) with an average deficiency of 44% for the Ken Basin as a whole.Rajnfall departure analysis further revealed that themagnitude of deficit rainfall in July and August months during severe drought years has increased during the latter half of the study period. The analysis of annual rainfall for Ken basin as a whole identified total fifteen drought years during the entire period of record, indicating the frequency of drought as once in seven years. However, separately analysed data during first half and second half of the study period indicated that the frequency of drought during first half is once in ten years, which has doubled to once in five years during the second half. The probability analysis of annual and seasonal rainfall reveals that the probability of receiving normal annual and seasonal rainfall in various districts in the basin range from lowest value of43 percent in Raisen district to highest value of 58% in Hamiur district for annual rainfall and 44 percent to 56 percent in Raisen and Hamirpur districts, respectively for seasonal rainfall. It has also been observed that the probability of occurrence of mean annual rainfall (i.e. 1165 mm) has decreased from 52 per cent in the first half of the study period to 46 per cent in the second half of the study period. Trends in annual, seasonal and monthly rainfall for the 8 districts under study have been investigated using Mann-Kendall ranic test and Sen's slope estimator. The trend test conducted for the entire time series of rainfall data indicate no significant trend in annual or seasonal rainfall. However, separately analysed rainfall data series during 1901 to 1950 and 1951 to 2002 indicated the entirely opposite trends with an indication of strongly decreasing trend during the latter period in contrast to increasing trend during the former period i.e. from 1901 - 1950. The decreasing trend in rainfall, increase in rainfall deficiency and number of drought eventsduring 1951 —2002 could be attributed to climatic anomalies in the basin. ASSESSMENT OF METEOROLOGICAL DROUGHT CHARACTERISTICS The meteorological drought characteristics of various districts falling under Ken river )asin have been assessed using six drought indices (DIs) namely, Effective Drought Index EDI), Standardized Precipitation Index (SPI), China Z-Index (CZI). Statistical Z-Score. ainfall Departure (RD) and Rainfall Deciles based Drought Index (RDDI) using 5 time steps 1-,3-,6-,9-, and 12-months for SPI, CZI, Z-Score, RD and RDDI and single time step for DI. The Pearson product moment correlation coefficient computed between various Dlsfor ach time step, indicated that these indices are highly correlated with each other for the same ime step. EDI is found to be best correlated with 9-monthly time steps of other Dis. On the )ther hand, it is observed that RDDI is least suitable for drought characteristics assessment in Ken river basin. The analysis also indicated that despite of good correlation amongst various frought indices, significant variation in the identified drought characteristics i.e. number of lv drought months, severity, time of onset and termination exists. The analysis of a historical drought event revealed that 1-month time step may produce erroneous result for the study basin; therefore, the drought characteristics assessment should be done for higher time step also. The analysis further reveals that the EDI has performed better compared to the other indices in timely identification of drought events and quantification of their severity. Overall the EDI has emerged as one of the preferred choices for assessment of drought characteristics for the study basin. EFFECT OF LENGTH OF DATAON ASSESSMENT OF METEOROLOGICAL DROUGHT CHARACTERISTICS The effect of length of data on assessment of drought characteristics has been examined using 100 years (1901 - 2000) rainfall data of two districts i.e. Sagar and Anantpur located in two different agro-climatic zones of India. The datahas been divided into datasets of different rA lengths ranging from minimum 30 years to 100 years with a moving window of 5 years, thereby generating 120 datasetsof different lengths and periods. Earlier identified most suitable drought indices namely, SPI-9 and EDI have been used for assessment of drought characteristics in Sagar and Anantpur districts using all 120 datasets for each and thus, generating 120 time series of both DIs for each district. The severity values of both DIs (SPI-9 and EDI) computed using different length of record (shorter than 100 years) with that of 100 years has been found to be highly correlated (r > 0.9) for the corresponding period.However,there is large difference in severity values computed using datasets of variouslength and period with that of 100 years data length. Also, there has been large variation in number of months identified under various severity categories of drought using datasets of different length and period. The variation in drought severity class for 30 years dataset as compared to 100 years dataset may be as high as 30.83% for SPI-9 and 36.67% for EDI. It has been found that the variation in severity values and number of inconsistent drought event decreases with the increase of data length.This indicated that the longer length of data provides more reliable estimates of drought severity. The values of various statistical and gamma distribution parameters (i.e. Average, Standard Deviation, Skewness, Alpha and Beta) have also been analysed for all 120 datasets. It is observed from the analysis that the difference in values of these parameters, computed using shorter length (less than 100 years) in comparison to 100 years data,decreases with the increase in length of dataset. It is observed from the analysis that relative variation in the values of different statistical parameters with change in length of dataset is dissimilar and length of data needed to obtain stable value of a given parameter varies for both districts. V EFFECT OF THE LENGTH OF STREAMFLOW RECORD ON HYDROLOGICAL DROUGHT ASSESSMENT The effect of length of streamfiow record on truncation level (TL) is assessed using 50 it years (1960 to 2009)10-day streamfiow data of two gauging stations, namely Banda and Nandghat located in Ken river basin and Sheonath river basin, respectively. The datasets of both locationsare divided into datasets of different length with a moving window of 5 years and minimum 10 years data length, thereby generating total 45 datasets for each gauging station. For assessment of TL,the flow duration curves (FDC) have been prepared for each dataset. 75% dependable flow has been obtained as the TL from streamfiow records of various lengths. The truncation level values for each 10-day period obtained from datasets of different record length indicated significant variations in TL values. The deviation in the values of TL obtained from shorter length of data sets with respect to 50 years data length has been obtained. It has been noticed that as the length of data increases the deviation in the values of TL decreases. However, with 30 years of streamfiow record, the deviation is within acceptable limits. For determining the streamfiow drought characteristics i.e. time of onset, termination, duration and severity of independent drought events,a pooling procedure has been applied. The drought characteristics identified using various lengths of record are also compared to estimate the minimum length of data required. This analysis suggested that at least 30 years of streamfiow record is required for reasonable estimation of streamfiow drought characteristics in both the river basins. ASSESSMENT OF AGRICULTURAL DROUGHT •1 The agricultural drought characteristics have been assessed using the soil water estimates obtained from the Soil Water Assessment Tool (SWAT)which is a process based distributed hydrological model. The SWAT model has been calibrated and validated using the observed streamfiow data of a period from 1984 to 1997 of four gauging stations located in Ken river basin. The high model efficiency between observed and simulated streamfiow at all four gauging sites indicated the very good model fit. For application of the SWAT Model, the entire Ken basin has been discretized into 104 sub-basins with 7942 Hydrologic Response Units (HRUs). The HRUs are hydrologically unique areas with unique slope, soil and land use demarcated by SWAT. The SWAT simulated soil water stored in soil profile at the end of each month for all 104 sub-basins has beenused for computation of soil moisture deficit index (SMDI). The SMDI maps have been prepared for each month. The SMDI maps thus prepared clearly distinguished the dry and wetperiods, which concur favourably with the meteorological vi dry and wet conditions and documented records. The basin area has been classified into different drought severity classes depending upon the SMDI values.Thus, monthly maps of SMDI for a given district can provide reliable information for decision making and formulation of drought mitigation plan. Also, the estimates of SMDI have been utilized for integrated assessment of vulnerability to drought in space and time. ASSESSMENT OF VULNERABILITY TO DROUGHT The vulnerability to drought at a given place and time depends on various physiographic, hydrologic, climatic, social and other factors like abilities to cope with water shortages. The Ken river basin exhibits different physiographic, hydrologic and meteorological conditions with varied exposure to demand and availability of water throughout the basin. In this study various physiographic, climatic, hydrologic and social factors viz, land use, irrigation support, soil type, soil depth, slope, distance from river reach, elevation zones, rainfall departure (RD), soil moisture deficit index (SMDI) and population density have been considered for assessment of vulnerability to drought. Here physiographic factors and social factors are static and semi-static factors, respectively.Whereas, RD and SMDI are dynamic factors and changes with time and space. The different weight has been assigned to different sub-classes of these factors depending on their relative influence on vulnerability to drought, termed as differential weighing scheme. The weights assigned to sub classes of all factors pertaining to a particular HRU are integrated using a simple scheme of addition of weights. The composite weight value scored by selected HRU divided by the sum of maximum weight values of each factor has been defined as Integrated Drought Vulnerability Index (IDVI). The 4.- maps of various physiographic factors and hydrologic factors for typical drought years have been overlaid using Geographical Information System (GIS) to generate composite map of vulnerability to drought in space and time. On the basis of values of IDVI estimates, the basin areas are categorized under different vulnerability classes i.e. Non-Vulnerable, Least Vulnerable, Moderately Vulnerable, Severely Vulnerable and Critically Vulnerable. The vulnerable patches thus identified for typical drought years are compared with documented drought events in the basin and qualitatively validated on the basis of records of crop yield in the basin. The results have been compared with existing uniform weighing scheme (assigning equal weights to sub-classes of various factors). It has been found that proposed differential weighing scheme provided more reliable assessment of vulnerability to drought in space and time for Ken basin for the typical drought events in the years of 1988 and 1993 as compared to uniform weighing scheme.