Academic Commons

Conference Objects

Calibration of Probabilistic Ensemble Forecasts for Indian Summer Monsoon Rainfall: A Non-Gaussian Approach

Acharya, Nachiketa

Due to the large variability of seasonal Indian summer monsoon rainfall (ISMR), the user community should be given probabilistic forecasts that convey the inherent uncertainty within the prediction. Probabilistic seasonal prediction can be done based on the general circulation models (GCM)’s outputs, however the output of these ensemble predictions cannot be used directly and requires further calibration in order to produce reliable forecasts. The common approach to make such probabilistic forecast is to calibrate the deterministic forecast from the model and then convert to probabilistic space using Gaussian distribution. Because neither observation for seasonal ISMR nor the GCM’s forecast follow a Gaussian distribution, we introduce a non-gaussian model viz., Extended Logistic Regression (ELR) for calibration method. ELR has been implemented on coupled GCMs of the North
American Multi-Model Ensemble (NMME) project over 1982-2010 following a leave-one-year-out cross-validation manner. The skill of the proposed method has been evaluated against observed rainfall and found to be a good reliable and well calibrated forecast.

Geographic Areas

Files

  • thumnail for CALIBRATION OF PROBABILISTIC ENSEMBLE FORECASTS FOR INDIAN SUMMER MONSOON RAINFALL- A NON-GAUSSIAN APPROACH.pdf CALIBRATION OF PROBABILISTIC ENSEMBLE FORECASTS FOR INDIAN SUMMER MONSOON RAINFALL- A NON-GAUSSIAN APPROACH.pdf application/pdf 269 KB Download File

Also Published In

Title
Proceedings of the 8th International Workshop on Climate Informatics (CI2018)
Publisher
National Center for Atmospheric Research
DOI
https://doi.org/10.5065/D6BZ64XQ

More About This Work

Academic Units
International Research Institute for Climate and Society
Published Here
October 2, 2020