Skill of the extended range prediction (NERP) for Indian summer monsoon rainfall with NCMRWF global coupled modelling system

• 全文链接
• 请求原文

作   者：

Anitha Gera;  Ankur Gupta;  Ashis K. Mitra;  Nagarjuna Rao D.;  Imranali M. Momin;  Madhavan N. Rajeeavan;  Sean F. Milton;  Gill M. Martin;  Matthew J. Martin;  Jennifer Waters;  Daniel Lea; 

作者机构：

Met Office;  NCMRWF;  Reading;  Noida;  India;  Ministry of Earth Sciences;  GoI;  UK;  MoES; 

关键词：

global coupled model;  sub-seasonal prediction;  Indian monsoon rainfall;  extended range prediction; 

期刊名称：

Quarterly Journal of the Royal Meteorological Society

i s s n：

0035-9009

年卷期：

2022 年 148 卷 742 Pt.A 期

页   码：

480-498

页   码：

摘   要：

An extended range prediction system (NERP) based on the Unified global coupled modelling system has been implemented at National Centre for Medium Range Weather Forecasting (NCMRWF). Predictions of weekly anomalies for rainfall, winds and surface temperature are issued once a week for the subsequent 4?weeks. This study is an assessment of the skill of the NERP system over India during the summer monsoon period of June to September (JJAS). The week-1 and week-2 rainfall anomaly forecasts exhibit good prediction skill over the Indian region. The evaluations made on meteorologically homogeneous regions of India show moderate to good skill even in week-3 and week-4 forecasts for some regions. The variability of the monsoon seasonal cycle in the model is well represented along with the magnitude of summer monsoon rainfall. The model brings out the mean monsoon rainfall features - the peaks during JJAS occurring over the Western Ghats, the Arakan coast and the Bay of Bengal. The onset of the monsoon and the interannual variability of the northward propagation of the rainfall anomalies over the Indian longitudes match reasonably well with observations. The model has some wet bias over the equatorial Indian Ocean and mild dry bias over the Indian land region that develop during the 4-week period. The anomaly correlations of rainfall are significant over most of the subdivisions of India till week-3 and week-4. Probabilistic scores of reliability Receiver Operating Characteristic (ROC) and fair continuous ranked probability score (FCRPS) show that the model exhibits good skill over most regions till week-2 and moderate skill in weeks 3-4 over other regions. The estimates of forecast reliability are low, a likely consequence of the limited ensemble size. A case-study of year 2018 showed that the JJAS rainfall regime shifts from wet to dry and vice versa are captured well.