تنبؤ احصائي لسرع الرياح لمحطات محددة في العراق == Statistical Forecasting of Wind Speed for Selected Stations in Iraq
Author name:
سرى ثامر ناصر
Supervisor name:
بدور ياسين حمود العامري | محمد احمد صالح ابو الطيب
General topic:
Meteorology
Specific topic:
Atmospheric Sciences
Degree:
Master
University:
Mustansiriyah University - College Of Science - Department Of Atmospheric Sciences
Language:
English
University location:
Baghdad
First pages:
32T81 - p.pdf
Abstract:
Wind speed is one of the most meteorological variables which has been changing rapidly with time. Due to the properties of wind speed such as non - stationarity, high fluctuations and irregularity. accurate forecasting becomes a challenge because, The wind speed is easily influenced by many factors, such as latitude, altitude, terrain, pressure, temperature. Therefore, The forecasting of the wind speed is important for many environmental, economic and meteorological applications. Due to the difficulties of the physical models and their non - availability, The Box - Jenkins ARIMA model has been used .Because, it is suitable tool to predict the wind speed and can also be used as an effective tool to get the missing data which existing within the time series of monthly mean wind speed. When Box - Jenkins ARIMA model has used, The ARIMA (15,1,10) model has determined as a suitable model for the monthly mean wind speed for studied stations in Iraqi after passing through the four stages of the model that include Identification of The Time Series, Estimation of The Model Parameters, Diagnostic Check of The Residuals and Model Adequacy and finally, The Forecasting. Many of the evaluation criteria have been used to check the accuracy of the best proposed prediction models by lowest values of each MAE and RMSE and MAPE and the highest values of the correlation coefficient (R2) and correlation coefficient (r), which confirmed the preference of ARIMA(15,1,10) for the monthly mean of wind speed but, This model could not applied on the daily mean of wind speed due to difficult in chose one best - fit model for all time series in all regions of Iraq. Because, The large variance of the wind speed during day and night and has not appeared clear correlation in all stations as shown in monthly data. The best solution is to take the difference of the non - stationarity time series from one lag to the next and then analyze this series after reaching to stationarity. The time series of daily wind speed need to take the difference once for the Baghdad station (d = 1) and twice for the Kirkuk station (d = 2) and three times for the Nasiriya station (d = 3) for reaching to stationarity. When ARIMA(15,1,10) model has been used, All the residuals were found within the confidence intervals and The error rate of using MAPE measure has been found near ( 4% ) for all studied stations in Iraq .
