MOS guidance using a neural network for the rainfall forecast over India

Journal of earth system science

Sridevi Ch^1¹  Ashok Kumar^1²  Durai V R^1³ 

[1] India Institute of Tropical Meteorology, Pune 411 017, India.^2;India Meteorological Department, New Delhi 110 003, India.^1;India Meteorological Department, Pune 411 005, India.^3

关键词: MOS;    neural network;    T-1534;    regular grid;    monsoon season;    Indian window;   

DOI  :  

学科分类：天文学(综合)

来源: Indian Academy of Sciences

【 摘 要 】

In the present study, a model output statistics (MOS) guidance model was developed by using the neural network technique for a bias-corrected rainfall forecast. The model was developed over the Indian window (0–40$^{\circ}$N and 60–100$^{\circ}$E) by using the observed and global forecast system (GFS) T-1534 model output (up to 5 days) at a 0.125$^{\circ} \times$ 0.125$^{\circ}$ regular grid during the summer monsoon (June–September) 2016. The skill of the developed MOS model forecast against the observed 0.125$^{\circ} \times$ 0.125$^{\circ}$ grid rainfall data is obtained for the summer monsoon (June–September) 2017. The skill of the MOS model rainfall forecast is found to show good improvement over the T-1534 model’s direct forecast over the Indian window. In general, the T-1534 model’s direct forecast shows high skill but the forecast obtained by using the MOS model shows better skill than the direct model’s forecast, although a major improvement is seen for the Day 1 forecast at the national level. So the skill of the bias-corrected rainfall forecast by using the MOS guidance and the T-1534 model output is high and has the potential of being used as an operational forecast over the Indian region.

【 授权许可】

CC BY