A probabilistic climatology‐based analogue intensity forecast scheme for tropical cyclones. (6th July 2016)

• Record Type: Journal Article
• Authors: Chen, Peiyan; Yu, Hui; Brown, Barbara; Chen, Guoming; Wan, Rijing
• Is Part Of: Quarterly journal of the Royal Meteorological Society. Volume 142:Number 699(2016)
• Page Start: 2386
• Page End: 2397
• Publication Date: 2016-07-06
• Subjects: tropical cyclone -- intensity -- probabilistic forecast -- climatology; Meteorology -- Periodicals; 551.5
• DOI: 10.1002/qj.2831
• Languages: English
• ISSNs: 0035-9009

Abstract

Abstract : In this article, we develop an objective probabilistic climatology‐based intensity forecast (PCIF) scheme for tropical cyclones (TCs) over the northwestern Pacific region based on data from all TCs that occurred in the region between 1981 and 2010. The PCIF scheme has a forecast period of 120 h at 6‐hourly intervals and was developed using eight persistence predictors: the latitude and longitude of the TC centre, the initial minimum central pressure (PMIN) and its change during the last 12 h, the TC movement direction within the last 24 h, the Julian day of the initial time, the life span of the TC between the initial time and its first recording in the observations, and the underlying surface conditions at the TC centre. The PCIF was independently tested for all TCs over the northwestern Pacific from 2011 to 2013. The mean absolute errors of its ensemble mean predictions for PMIN were 8.0, 12.5, 15.2, 17.0 and 17.2 hPa at 24, 48, 72, 96 and 120 h, respectively. Comparisons with the Tropical Cyclone Statistical Prediction, a climatological baseline approach used by the China Meteorological Administration to assess the skill of TC intensity forecasts, indicated the PCIF has significant prediction skill for lead times of 12–48 h. The PCIF probabilistic intensity predictions had better performance for severe or dying TCs at most lead times, and for most categories at short lead times. The PCIF was also found to perform better than the output of the European Centre for Medium‐Range Weather Forecasts Ensemble Prediction System in the probabilistic prediction of intensity at lead times shorter than 72 h.