Observation series modeling by fourier and Nap Sinus adjustment a tool of assistance to the prediction of the temperature Tamanrasset Algeria

In a context of world climatic reheating, the zones in margin of the deserts today are particularly threatened. From there, was born the need to obtain with tools and models for predictions with the temperature in order to establish the future scenarios & assumptions and the climatic impact on its areas. The approach adopted in this work is of type statistical/stochastic which rests on the use of rough series, thus integrating the seasonal variations in the exercise of modeling. The analysis of the time series of temperatures of the area of Tamanrasset on more than 15 years made it possible to reveal that the latter was only the sum of a Gaussian white noise which one produces with a generator of Box-Muller, and of a signal modeling the seasonal fluctuations of the series of temperatures realizing two mathematical models, namely the model of Fourier and that of Nap Sinus of order 8, such as T (T) =Aj (T) + B (T). By exploiting a technic of crossed validation, we built two models over determinate durations, which we test on the remainder of temperature series. The Root Mean Square Error (RMSE), the scatter chart, the correlation coefficient, and finally the Nash coefficient will be used as criteria of evaluation of the adopted models. Generally and according to criteria´s of evaluations studied, it appear from both models a great aptitude to reproduce the seasonal cycles of the time series of temperatures with a RMSE about 2,13°C for the two models, whereas the coefficient of correlation varies to him from 0,92 to 0.95 of a model to another. Finally the coefficient of Nash oscillates around 0, 89. In the End, we developed a very powerful tool for simulation with a less development time and less hard, comprising only one good generator of white noise, and a good modeling, which is able to predict, and with reliability of long series of temperature in the desert area of Tamanrasset, Algeria.