Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Etats Unis → 1965 → Evaluation of alternative Markovian models for precipitation occurrence in Oregon

Oregon State University (1965)

Evaluation of alternative Markovian models for precipitation occurrence in Oregon

Adotevi-Akue, George Modesto

Titre : Evaluation of alternative Markovian models for precipitation occurrence in Oregon

Auteur : Adotevi-Akue, George Modesto

Université de soutenance : Oregon State University

Grade : Master of Science (MS) 1965

Résumé
Precipitation occurrence in three winter periods and three summer periods is examined for two Oregon stations : Seaside and Squaw Butte. The winter periods are January, February, and the interval January 15 to February 20. The summer periods are July, August, and the interval July 10 to August 15. The two intervals were selected as representing stable periods in the Natural Calendar of the Atmosphere proposed in the literature. Seaside is a coastal station with a marine climate, while Squaw Butte is in the high desert of eastern Oregon. At both stations winter is the rainy season and summer is dry. By first defining a wet day as any on which .01 inch or more of precipitation is measured, and then proceeding from the assumption that daily precipitation occurrence is a two-state Markov Chain of order 4, the hypotheses are tested that the order of dependence is 0, 1, 2, and 3 within the fourth order. At Seaside for the winter periods another threshold was employed, with wet days being those having .20 inch or more of precipitation. In no case did a Chi-square test for goodness of fit show a model of zero order to be suitable. For winter periods at Squaw Butte, a model of order as high as 3 was not clearly suitable. For all other tests, however, a Markovian model of an order within the range tested was judged suitable. The variation of threshold tested at Seaside in winter has no major significance for the periods examined. The tests as conducted neither confirm nor deny the validity of the concept of the Natural Calendar. As a result, however, recommendations for further tests of this validity are made.

Présentation

Version intégrale

Page publiée le 1er janvier 2019