Comparison among the methods to synthesize the hydrological data

AuthorAhmed, Nasir Uddin
Call NumberAIT Thesis no. 851
Subject(s)Hydrology
Hydrography
NoteA thesis submitted in partial fulfillment of the requirement for the Degree of Master of Engineering at the Asian Institute of Technology, Bangkok, Thailand.
PublisherAsian Institute of Technology
AbstractThe purpose of this study is to compare the data generating models to synthesize streamflow data from the available historical data of three rivers, namely, the Mekong at Vientiane, the Ping at Bhumibol damsite and the Nan at Sirikit damsite by three data generating models, namely, the Thomas and Fiering model which incorporates serial correlation between consecutive monthly flows, the independent normal distribution model and the lognormal distribution model and later, by two modified models, namely, the modified normal distribution model and the modified lognormal distribution model. The historical data of the three rivers concerned are found to be trend free and that there are no periodicity other than one year, as these studies for the said rivers have already been done by previous researchers. The capacity requirements for water supply and flood control were taken as the criteria for judgement of the models and the comparison was done with the capacity requirements calculated from the historical data. In every case, the capacity requirement was calculated numerically from the line-series (historical and generated) with different assumed constant demand levels for water supply and for flood control. The reliability of the fitness of the data generated from the models with the historical one were checked by one kind of significance test which may be termed as \'the direct significance test, in which the capacity requirements from the historical data were checked within the significance level of one standard deviation from the mean of the capacity requirements calculated for the different sets of generated data from different models. The results showed that for water supply purpose, for most practical range of assumed water demand levels, the Thomas and Fiering model gave statistically and hence practically unacceptable over-estimation of the capacity requirements than those derived from the historical data and the normal model gave over-estimation of the capacity requirements in very low demand levels but the lognormal model gave very much under-estimation of the capacity requirements than those from the historical data. For the flood control purpose, the Thomas and Fiering model and the lognormal model could get the results from the historic data with one standard deviation range of their space requirements from the generated data. And the normal model gave very much under-estimation of space requirements in comparison to those from the historical data as well as those from the other generating models. The conclusion was drawn in this manner that, for the purpose of water supply, the Thomas and Fiering model gave statistically and hence practically over-estimation of the capacity requirements than those from the historical data for most of the practical range of demand levels and the capacity requirements from the normal model and lognormal model were very much impractical. For flood control purpose, though the space requirements from the historical data were within one standard deviation of the space requirements from the Thomas and Fiering model and the lognormal model, these models could not be recommended for practical usage because the co-efficient of variance were found to be very high. And, the normal model also gave statically and practically very much under-estimation of the space requirements.
Year1975
TypeThesis
SchoolStudent Research Before 1980
DepartmentOther Field of Studies (No Department)
Academic Program/FoSThesis (Year <=1979)
Chairperson(s)Okamoto, Masami
Examination Committee(s)Suphat Vongvissesomjai ;Selvalingam, Selvadore
DegreeThesis (M. Eng.) - Asian Institute of Technology, 1975


Usage Metrics
View Detail2
Read PDF0
Download PDF0