Open AccessAutomating Time Series Analysis to Predict/Forecast Rainfall in AGUELMAM SIDI ALI Watershed in Morocco
Author(s) -
Ridouane Chalh,
Zohra Bakkoury,
Driss Ouazar,
Moulay Driss Hasnaoui,
Addi Ait-Mlouk
Publication year - 2019
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.i7561.078919
Subject(s) - akaike information criterion , univariate , time series , autocorrelation , environmental science , water resources , meteorology , estimation , autoregressive integrated moving average , precipitation , climatology , productivity , watershed , statistics , econometrics , computer science , mathematics , geography , multivariate statistics , machine learning , engineering , ecology , macroeconomics , systems engineering , geology , economics , biology
Moroccan economy is largely based upon rainfall, use of water resources and crop productivity, for that it’s considered as an agricultural country. It’s more required and more important for any farmer to forecast rainfall prediction in order to analyze crop productivity. Predicting the atmosphere or forecasting the state of the weather is considered as challenge for scientific research. The prediction of rainfall monthly or/and seasonal time scales is the application of science and technology to invent and to schedule the agriculture strategies. Recently different research articles achieve to forecast and/or predict rainfall monthly or seasonal time scales using different techniques. The methodology followed in this work, be focused on automating time series analysis to forecast / predict precipitation daily, monthly or seasonal in Aguelmam Sidi Ali basin in Morocco for last 32 years ago from 1975 to 2007. We first have to study the rainfall data theoretically using the simplest form statistical analysis, which is the univariate analysis, as long as only one variable is involved in our case study. To get the selected and suitable model of time series to automate, we used different autocorrelation methods based on various criterion such as: Akaike Information Criterion (AIC), estimation of parameters using Yule-Walker (YW) and Maximum Likelihood Estimation (MLE). The results of our experiment show that it is possible using our system to obtain accurate rainfall prediction, with a more details and with a very fast way. It shows also that it’s possible to predict for next months or next years. To minimize the risk of floods and natural disasters within a basin in general and within the Aguelmam Sidi Ali basin in particular, accurate and timely rainfall forecasting is required.