عنوان

مطالعه مدل مفص - آنتروپی ومفصل سه بعدی وچهار بعدی ‮‭C-vine‬ و -‮‭vine D‬در شبیه سازی داده های هیدرولوژیکی

شماره

‭۱۸۵۵۶پ‬

زبان متن نوشتاري يا گفتاري و مانند آن

per

عنوان اصلي

مطالعه مدل مفص - آنتروپی ومفصل سه بعدی وچهار بعدی ‮‭C-vine‬ و -‮‭vine D‬در شبیه سازی داده های هیدرولوژیکی

نام نخستين پديدآور

/مریم شفائی

نام ناشر، پخش کننده و غيره

: کشاورزی

تاریخ نشرو بخش و غیره

، ‮‭۱۳۹۶‬

نام توليد کننده

، راشدی

متن يادداشت

چاپی

جزئيات پايان نامه و نوع درجه آن

دکتری

نظم درجات

مهندسی آب گرایش مهندسی منابع آب

زمان اعطا مدرک

‮‭۱۳۹۶/۱۱/۱۰‬

کسي که مدرک را اعطا کرده

تبریز

متن يادداشت

Hydrological phenomena such as rainfall and flood include important features that are dependent each other. Frequency analysis and simulation of these hydrological phenomena, taking into account the dependence between these basic features, will lead to more accurate simulation of these phenomena. The joint functions based on the Sklar theorem, allow simultaneous use from multiple univariate distributions with different statistical distributions to construct a multivariate distribution (which also takes into account the dependence between variables). One of the major problems with the use of copula functions is the complexity of the problems with dimensions of greater than two. The vine copulas have provided more flexibility in the problems with dimensions of higher than two, among the multivariable structures that have been presented so far. So that, the vine copulas break down multivariate density functions into two-dimensional copulas (pair-copula), and thus high-dimensional copulas will include two-dimensional copulas. Each of the vine-copulas (C-vine and D-vine) provides a different decomposition structure. Thus, decomposition of the multivariate distribution functions to the pair-copulas provides the flexible method for modeling the high-dimensional distribution functions. Also, since the vinecopulas use from conditional dependence between the different characteristics of the hydrological phenomenon in their structures, thus, in this method, the effects of the important features of hydrological phenomena are considered and leads to the more accurate simulation of phenomena. In this thesis, simulation the important characteristics of rainfall and flood was investigated using C-vine and D-vine structures. For this purpose, the hourly rainfall statistics of the rain gauge station in Chahar Mahal and Bakhtiyari province during the statistical period (2008-2015) were investigated. As in the three-dimensional andfour-dimensional vine models of rainfall phenomena, will be modeled by four important characteristics of the rainfall event: wet period duration, maximum rainfall intensity, total rainfall depth and time intervals between rainfall events (dry period). In addition, in another section of the thesis, simulation of the important characteristics of flood events was considered., In simulation of vine structures, a long-term statistical period is required to find the real correlation between variables and consequently, to simulate more accurate variables. Given that the interior basins in Iran country have not such long statistics So, the flood statistics of the Ripeta station located in the Tiber river basin in Rome, with the statistical period (2000-1921), were used. The basic time, time to peak, peak discharge and total flood volume variables were extracted and used for modeling the flood characteristics. One of the major goals of this thesis is investigation the effect of permutations of variables in the D-vine and C-Vain structures. For this purpose, all three-dimensional and four-dimensional structures using permutations of variables in the nodes of the vine structures were developed. In the next step, two-dimensional Archimedes and Elliptical families were evaluated to find the best fit for pair-copulas of each structure. Then, according to different evaluation criteria, the most suitable copula family (for each pair-copula) was selected and in the next step, the parameters of the two-dimensional copulas (pairs-copulas) of different vine structures were estimated with the simultaneous application of the sequential and the maximum log- likelihood methods. Finally, each of the important characteristics of flood and precipitation was simulated using C-vine and D-Vain three-variable and four-variable models. The results of the simulations showed that the simulation of the precipitation characteristics with the four-dimensional D-vine method works more accurately than the four-dimensional, and also the four-dimensional D-vine has a better performance than a three-dimensional D-vin. In the simulation of the fourdimensional D-vine, only one variable was simulated weakly, and in other cases, all variables were modeled with high accurate. In addition, the results showed that three-dimensional D-vine has a higher simulation ability than the three-dimensional C-vine. However, in some cases, the cited two copulas act in the same way. Also, after simulation the flood characteristics using vine structures, results indicated the relative superiority of the four-dimensional D-vine copula to the four-dimensional C-vine copula in the simulation of flood characteristics. In addition, comparison the results of the simulation obtained from the three-dimensional D-vine and Cvine structures shows that the three-dimensional D-vine structure has a much higher ability when compared to the three-dimensional C-vine structure. In another section of this thesis, the accuracy of vine structures in extraction of three-dimensional and four-dimensional distributeddistributions was investigated. Studies showed that the joint distributions obtained from the three-dimensional C-vine and D-vin structures have good overlap and match with the threedimensional experimental copulas

مستند نام اشخاص تاييد نشده

شفائی، مریم

مستند نام اشخاص تاييد نشده

فاخری فرد، احمد، استاد راهنما

مستند نام اشخاص تاييد نشده

دین پژوه، یعقوب، استاد راهنما

يادداشت عمومي

سیاه و سفید

وضعیت فهرست نویسی

نمایه‌سازی قبلی