Abstract:This study aims to determine trends in the long-term annual mean and monthly total precipitation series using nonparametric methods (i.e. the Mann-Kendall and Sen's T tests). The change per unit time in a time series having a linear trend was estimated by applying a simple non-parametric procedure, namely Sen's estimator of slope. Serial correlation structure in the data was accounted for determining the significance level of the results of the Mann-Kendall test. The data network used in this study, which is assumed to reflect regional hydroclimatic conditions, consists of 96 precipitation stations across Turkey. Monthly totals and annual means of the monthly totals are formed for each individual station, spanning from 1929 to 1993. In this case, a total of 13 precipitation variables at each station are subjected to trend detection analysis. In addition, regional average precipitation series are established for the same analysis purpose. The application of a trend detection framework resulted in the identification of some significant trends, especially in January, February, and September precipitations and in the annual means. A noticeable decrease in the annual mean precipitation was observed mostly in western and southern Turkey, as well as along the coasts of the Black Sea. Regional average series also displayed trends similar to those for individual stations.
This study aims to predict the daily precipitation from meteorological data from Turkey using the wavelet-neural network method, which combines two methods: discrete wavelet transform (DWT) and artificial neural networks (ANN). The wavelet-ANN model provides a good fit with the observed data, in particular for zero precipitation in the summer months, and for the peaks in the testing period. The results indicate that wavelet-ANN model estimations are significantly superior to those obtained by either a conventional ANN model or a multi linear regression model. In particular, the improvement provided by the new approach in estimating the peak values had a noticeably high positive effect on the performance evaluation criteria. Inclusion of the summed sub-series in the ANN input layer brings a new perspective to the discussions related to the physics involved in the ANN structure.Key words wavelet transforms; artificial neural networks; precipitation; Turkey; estimation Prévision de précipitation journalière à l'aide de réseaux neuronaux-ondelettes Résumé Cette étude cherche à prévoir les précipitations journalières à partir de données météorologiques Turques, à l'aide de la méthode de réseau neuronal-ondelettes qui combine deux méthodes: la transformée en ondelettes discrète (TOD) et les réseaux de neurones artificiels (RNA). Le modèle RNA-ondelettes produit un bon accord avec les données observées, en particulier pour l'absence de précipitation pendant les mois d'été, et pour les pics durant la période de test. Les résultats indiquent que les estimations du modèle RNA-ondelettes sont significativement supérieures à celles que l'on obtient avec un modèle RNA conventionnel ou avec un modèle de régression multilinéaire. En particulier, l'amélioration permise par la nouvelle approche pour l'estimation des valeurs de pic a un effet positif notable sur les critères de performance. L'inclusion des sous-séries sommées dans la couche d'entrées du RNA ouvre de nouvelles perspectives en termes de prise en compte de la physique dans la structure d'un RNA.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.