Abstract ID: 255
Extended Range Forecast of Monsoon over India at Met-subdivision and Districts levels : Prospect of its applications in Agriculture
Lead Author: D R Pattanaik
IMD, India
Keywords: Extended Range Forecast , Climate Forecast System, Met-subdivision level forecast, District level forecast, Verification Scores. ,
Abstract: The performance of operational extended range forecast (ERF) issued by IMD at met-subdivision and district level is evaluated for the southwest monsoon during recent years 2019, 2020 and 2021. During 2019 monsoon season the transition of monsoon from above normal to below normal in July month is well captured in the ERF in smaller spatial scales of meteorological subdivision level forecasts, which is being used widely for Agromet advisory purpose upto two weeks.
The performance of operational extended range forecast (ERF) issued by IMD for recent two monsoon seasons (2020 and 2021) at 676 districts over India is also computed. The ERF prepared in the real time up to 4 weeks are evaluated at district level in terms of predicting above normal (AN), normal (NN) and below normal (BN) rainfall. Other statistical scores like Forecast Accuracy, Bias Score (BS), Probability of Detection (POD), False Alarm Ratio (FAR), Critical Success Index (CSI), Equitable Threat Score (ETS) etc. have been calculated for ‘AN’ ‘NN’ and ‘BN categories separately at district level. For 2020 monsoon season the mean percentage of ‘C’ (correct) category forecast for the total number of districts is found to be 50%, 41%, 35% and 31% respectively for week 1 to week 4 forecasts, whereas for 2021 it is correspondingly higher with 54%, 44%, 36% and 34% respectively. Over the four homogeneous regions the percentage of ‘C’ category forecasts for all 4 weeks is increased by 2% to 7% in 2021 monsoon season compared to that of 2020 monsoon season with highest improvement of 7% (44% to 51%) is noticed in week 2 forecast over the NW India. During the 2020 monsoon season, the Forecast Accuracy in terms of predicting ‘above normal’ normal and below normal categories show 65-68% for week 1 forecast and it gradually decreases to about 52-56% in week 4 forecasts, whereas, for 2021 monsoon season it is slightly higher by 2% to 4%. The Bias Score during both the seasons has a tendency to over predict the above normal categories (With BS > 1) and under predict the below normal category (With BS < 1), whereas the normal category indicates much higher BS values indicating over prediction.
The present analysis shows encouraging results at least for up to two weeks with relatively higher skill. Considering the complexity of the monsoon forecast in the extended rage time scale at like at district level the results show very encouraging results at least up to 2 weeks for various applications, although, it has the scope for further improvement.
Co-authors:
D R Pattanaik, India Meteorological Department, New Delhi
