Please use this identifier to cite or link to this item: http://localhost:8081/xmlui/handle/123456789/520
Full metadata record
DC FieldValueLanguage
dc.contributor.authorBasistha, Ashoke-
dc.date.accessioned2014-09-17T07:12:18Z-
dc.date.available2014-09-17T07:12:18Z-
dc.date.issued2009-
dc.identifierPh.Den_US
dc.identifier.urihttp://hdl.handle.net/123456789/520-
dc.guideGoel, N.K.-
dc.guideArya, D.S.-
dc.description.abstractClimate change has been attracting global attention at an ever-increasing rate over the last two decades. Given the far-reaching impacts of climate change on the status of lifeon earth as a whole, as understood from palaeo-climatic studies, the concern is justified. Studies on climate change bear crucial importance for India where the population pressure is intense and agrarian economy depends highly on rainfed agriculture. In the present study attempts have been made to systematically study climate change through analysis of observed data series of climatological parameters. This has been achieved through following sequential steps: (i) Development of software to carry out statistical tests for detecting trend and shift in time series, (ii) Sensitivity analysis of several popular methodologies of trend analysis, (iii) Analysis of Indian sub divisional rainfall and regional temperature series, (iv) Analysis of time series for climatological parameters from North Indian observatories, (v) Preparation of a continuous digital rainfall grid dataset with 1 km resolution by interpolating normal annual rainfall in Uttarakhand, and (vi) Analysis for change detection in rainfall and rainy day series from rain gauge stations in Uttarakhand. To carry out an extensive data analysis, the necessity of customised user-friendly software was felt. The present work is an attempt to create a user-friendly software incorporating tests used for trend and shift detection. It has been developed as a macro using Visual Basic Applications embedded into an MS Excel file. It is an interactive program that provides tool tips, options and warnings, and generates error messages for common errors. It has context sensitive help and displays mathematical formulae involved in each test. It incorporates Correlogram test. Mann-Kendall test with its suggested modifications, Pettitt-Mann-Whitney test and moving t test for detection of shift alongwith several other popular tests. In the studies of climate change and hydrological change analyses, mostly Mann- Kendall test has been used for detection of trend. However, it has been shown by many researchers that in the presence of autocorrelation it generates incorrect results. Hydrometeorological data series, particularly average temperature, relative humidity are generally autocorrelated in nature. Therefore, several modifications have been suggested in literature for its improvement. The tests which have been considered in this study are listed below: (i) Mann-Kendall test (original), (ii) Modified Mann-Kendall test, (iii) Pre-whitened Mann-Kendall test, (iv) Mann-Kendall test modified by effective sample size following Bayley and Hammersley, and (v) Mann-Kendall test with Trend-free pre-whitening. Limited number of sensitivity studies (power of test studies) in the presence of autocorrelation is available. Hence, in-depth study of alternative tests for handling such data was necessary. In this study, autoregressive process of First Order (ie. AR(1)), with values -0.9 to 0.9 in steps of 0.1, for which the improved methods were suggested, is dealt with, concentrating on a series length of 100. The power of tests to detect a trend that starts at time lags of 0 to 90 in steps of 10 has also been compared. For all the processes and lags, trend values were varied from -0.05 a to 0.05 a in steps of 0.01 a , measured in terms of standard deviation of the series concerned. The capability to detect shift as nonrandomness in the series has also been considered with shift introduced at time lags of 10- 10-90. For this purpose, synthetic series were generated by addition of shifts having magnitude of 3a to -3a in steps of -0.5 a, for all the lags and processes. For each combination of trend or shift, 2500 series were generated. As a result, 10.78 million series were analysed at 10% Significance level. The results of the analysis indicate that (i) none of the method considered are able to fully counteract the effect of autoregression in detecting presence of trend, particularly for series with high magnitudes of autocorrelation, (ii) non-detection of change by statistical tests does not imply absence of change as tests fail to detect (i) weak changes (ii) changes that start late relative to the length of the time series. vi (iii) original Mann-Kendall test is the best for analysis of non-autocorrelated series, and Pre-whitened Mann Kendall test is suitable for series with low levels of autocorrelation. For statistical analysis of trend in observed meteorological data series, Pre - Whitened Mann Kendall test has been used because (i) results of the tests do not differ widely at low levels of autocorrelation which are commonly observed; and (ii) for autocorrelated series this yields better results being least influenced by autocorrelation. This has been employed for trend detection in sub divisional rainfall and regional temperature series of India, multi-parameter meteorological dataset of North India and rainfall series of Uttarakhand. A number of climate change studies over India have been reported in literature. The conclusions reported in literature are diversified in nature. Most of the studies have not considered the autocorrelation aspect. Hence, in the present work, studies have been conducted on an extensive database comprising of annual time series of each month and season, in addition to the annual values. Sub-divisional rainfall and regional temperature analysis: Monthly rainfall data from 30 homogeneous subdivisions (http://www.tropmet.res.in) and monthly temperature (maximum and minimum temperature - available, average temperature and temperature range - derived) data from seven temperature regions were analysed over the period 1902-2003. The analysis brings out that (i) there are indications of change in rainfall pattern over the century. It was observed that May is receiving more rains and August is receiving less, (ii) variability of temperature has increased over the recent period (1990-2003), evident by a loss of autocorrelation, (iii) rainfall trends are significant in a few subdivisions only, (iv) temperatures continued to increase over the century, however, rise in maximum temperature are more prominent. It is reflected through an increase in temperature range. (v) November and December witnessed most of the rise in minimum temperature, whilethe majority of falls were observed in January and June. vn (vi) the rise of maximum and minimum temperatures resulted in rise in monthly average temperature (i.e., in December, January and February). (vii) change in temperature is more prominent than change in rainfall. (viii) the patterns of change in rainfall and temperature are far from homogeneous, both in space and time; more so over the recent past. Analysis of North Indian climate data: In recent times, climate change in North India has been attracting interest particularly due to observations of increased aerosol (measured in terms of optical depth) during winter season. Therefore, trend detection studies have been further extended by considering several meteorological parameters: (i) Monthly rainfall, (ii) Monthly number of rainy days (days recording rainfall of 2.4 mm or more), (iii) Heaviest rainfall recorded in 24 hour for each month, (iv) Monthly average temperature, (v) Monthly mean maximum temperature, (vi) Monthly mean minimum temperature, (vii) Monthly temperature range, (viii) Monthly highest maximum temperature, (ix) Monthly lowest minimum temperature, (x) dew point temperature at 8:30 hours, and (xi) dew point temperature at 17:30 hours. All monthly, seasonal and annual series for the period 1902-2005 were studied. The results reveal that (i) over the century, decrease in monsoon rainfall and rainy days are evident in terms of statistical trend. The percentage change results indicate high spatial variability of rainfall, (ii) all the meteorological variables show increased variation over the more recent period (1990-2005). (iii) reduction of number of rainy days without decrease in total rainfall is indicative of increasing intensities, (iv) majority of rainfall series indicates absence of trend; changes (wherever present) in all three rainfall parameters (rainfall totals, number of rainy days and heaviest 24 hour rainfall) appear to be more of local nature. (v) over the century long period, increase in extreme rainfall (heaviest 24 hour event) is not major, (vi) the monthly patterns of change are most vivid, which somewhat get moderated at annual level. The trends lack homogeneity over the months and seasons. (vii) temperatures and humidity continued to rise in general. vm (viii) increase of highest maximum is modest as compared to increase of mean maximum temperature, (ix) lowest minimumtemperatures are generally rising. Spatio-temporal analysis of rainfall in Uttarakhand region: Studies on the Himalayan region have been very limited, primarily due to lack of availability of data. In the region. India Meteorological Department maintains observatories / raingauge stations, which are inadequate in number for the mountainous terrain. Data from 44 stations were used in this study. A review of literature reveals that studies related to spatial distribution were carried out in other parts of the Himalayan range, but the spatial variability of rainfall over the state of Uttarakhand is still unexplored. Continuous rainfall data in grid format are required to run models for hydrological and agricultural research as well as water resources planning and management. The present work attempted to prepare a normal annual rainfall map in Himalayan region of India lying in Uttarakhand state at 1km spatial resolution which currently is not available. A comparative analysis of interpolation techniques like Inverse Distance Weighted. Polynomial, Splines, Ordinary Kriging and Universal Kriging shows that Universal Kriging with hole-effect model and natural logarithmic transformation with constant trend having Root Mean Square Error (RMSE) of 328.7 mm is the best choice. This is followed by Ordinary Kriging (RMSE 329.1 mm), Splines (RMSE 392.4 mm). Inverse Distance Weighted (RMSE 409.8 mm) and Polynomial Interpolation (RMSE 418.5 mm). Cross validation of the results shows the largest over prediction at Tehri rainfall station (62.5%) and largest under prediction at Nainital station (-36.5%). Attempts have been made to explore changes in rainfall pattern over Uttarakhand during twentieth century using data of 79 years (1902-1980) from 30 raingauge stations maintained by India Meteorological Department. Pre-Whitened Mann-Kendall test was applied to detect trend and Pettitt-Mann-Whitney test was employed to detect possible shift. Implication of the change was discussed in terms of percentage change over mean. The results show that the most probable year of change in annual as well as monsoon rainfall in the region is 1964. There was an increasing trend upto 1964 (corroborating with All India and nearby plains), followed by a decreasing trend in 1965-80 (exclusive to this region). In the entire region, changes are most conspicuous over the Shivaliks and southern part of the Lesser Himalayas. ix The major findings of the research are summarised as: (i) The prevalent modifications of Mann-Kendall test are insufficientto counteract the effect of autocorrelation, thereby yielding incorrect results for highly autocorrelated series; original Mann-Kendall test is best for analysing nonautocorrelated series, whereas Pre-Whitened Mann-Kendall test is suitable choice for series with autocorrelation levels normally encountered in nature; (ii) There are indications of change in seasonal pattern over the country; the patterns are heterogeneous over the months as well as over the regions; (iii) Warming winter average temperatures and cooling in early summer, with highly variable spatial patterns of decrease in monsoon rainfall are evident in North India; (iv) Universal Kriging method with Hole-Effect model and natural logarithmic transformation incorporating constant trend is the best suitable method for interpolation of normal annual rainfall over Uttarakhand, and (v) The decrease in monsoon rainfall observed over Uttarakhand (particularly, the Shivaliks and southern part of the Lesser Himalayas) during 1965-1980 indicates a divergent behaviour from the adjoining plains. This work revealed the necessity of future studies in the following directions: (i) improving trend detection in presence of autocorrelation, probably through modification of Pre-Whitened Mann-Kendall test, (ii) undertaking detailed climatological analysis which considers the linkage between the change recorded in different climatological parameters and their possible causes, (iii) analysing discharge data to investigate climate change effects, (iv) interpolating Uttarakhand rainfall in monthly and seasonal scale, preferably incorporating effects of topography, and (v) analysing satellite data (NDVI series) to assess the impact of climate change on agriculture.en_US
dc.language.isoen.en_US
dc.subjectCLIMATE CHANGEen_US
dc.subjectSPATIO-TEMPORAL ANALYSISen_US
dc.subjectREGIONAL TEMPERATURE ANALAYSISen_US
dc.subjectMETEOROLOGYen_US
dc.titleCLIMATE CHANGE STUDIES: A SPATIO-TEMPORAL ANALYSIS FOR PART OF INDIAen_US
dc.typeDoctoral Thesisen_US
dc.accession.numberG20526en_US
Appears in Collections:DOCTORAL THESES (Hydrology)

Files in This Item:
File Description SizeFormat 
CLIMATE CHANGES STUDIES A SPATIO-TEMPORAL ANALYSIS FOR PART OF INDIA.pdf23.55 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.