Details

Autor(en) / Beteiligte

• Mahajan, Seema
• Mazumdar, Himanshu S

Titel

A Regenerative Prediction Algorithm for Indian Rainfall Prediction

Ist Teil von

• International journal of engineering science and technology, 2013-11, Vol.5 (11), p.1832-1832

Erscheinungsjahr

2013

Quelle

EZB Electronic Journals Library

Beschreibungen/Notizen

• Rainfall forecasting is critical for the crop planning and water management strategies. The proposed study presents a novel approach for modelling time series precipitation data. The 51 years of Indian rainfall data is used for the development of the model. The authors use a nonlinear predictive code based on 11th order with 240 coefficients. Coefficients are optimized using a gradient descendent algorithm. Algorithm is tested using 40 years of rainfall training data. Prediction error tested outside training period is found less than 1% for few months. Prediction period is extended to one year by including progressive predicted values in input samples using regenerative feedback algorithm. This model is applied for different training and testing periods with average error of 2% to 10%.