Font Size: a A A

The Application Of LOWESS Fit To Isotope Stratigraphy

Posted on:2008-01-20Degree:MasterType:Thesis
Country:ChinaCandidate:Y HuangFull Text:PDF
GTID:2120360215469382Subject:Applied Mathematics
Abstract/Summary:PDF Full Text Request
LOWESS(Locally Weighted Regression Scatterplot Smoother, also called LOESS)is a robustness, nonparametric static method for fitting smooth curve to scatter plot data. Cleveland sets this up in 1979 at the first time. This paper utilized (87)Sr/(86)Sr,δβC isotope data in the upper Yangtze platform,though this math method, established strontium, carbon isotope—stratum thickness curve, obtained (87)Sr/(86)Sr—stratum thickness data in the upper Yangtze platform,and offered important base information for demarcating and contrast stratum age in this area. This variation trend of the curve is quite consistent with that of the contemporaneous curve set up by the samples from North America, Europe, Africa and Asia, showing the global correlation significance of isotopic compositions of strontium of marine carbonate.This paper involved four main parts. The first part explained the need of studying the variation trend of the (87)Sr/(86)Sr,δ(13)C curve and researched the feasibility of these curve,looking back the actuality for Strontium Isotope Stratigraphy (SIS for short) be used to date marine sediments.The second part is a heart content in this paper,pointed out the advantage and the disadvantage of parameter regression and nonparametric regression,putted forward the study thoughtfulness in this paper is LOWESS, and used LOWESS fitting the curve which is stratum stratum thickness (X)—(87)Sr/(86)Sr (Y) curve.Particularly, introduced mathematics method that based this curve, discussed feasibility for this method. The mostly factor which influence this curve is the choice for window and polynomial degree. This paper come out different LOWESS curves in different window and different polynomial degree, compared those curves, and come out best LOWESS curve. This paper also used LOWESS fitting the other curve which is stratum stratum thickness(X)—δ13C(Y)curve.The third part is another heart content, digitized curve which come out in Howarth et al. (2001), in order to gaining numeric age—87Sr/86Sr data; and utilized curve which come out in LOWESS, interpolated at intervals of 0.00001 in 87Sr/86Sr in order to gaining stratum thickness—87Sr/86Sr data; compared this two data, gained numeric age—stratum thickness data in the upper Yangtze platform.In the fourth part, we discuss collectivity evolvement trend for Strontium Isotope in the upper Yangtze platform though stratum stratum thickness(X)—87Sr/86Sr(Y)curve.Some creative conclusions in this paper below: relatively general using LOWESS for dating marine sediments for the first time. This paper summed up basic content and basic method of LOWESS, and expounded theory foundation in dating marine sediments.The theoretical researches are main parts of this paper, but some practical applications are also involved in it. So it would have some valuable references in dating marine sediments, and make benefit attempt for solving difficult problem in dating marine sediments.
Keywords/Search Tags:nonparametric regression, LOWESS, scatter plot smoothing, strontium and carbon isotope, dating marine sediments
PDF Full Text Request
Related items