Developing a parameter to detect imminent TC formation based on large-scale flow properties

Projections of future climate and associated changes in tropical cyclone (TC) behaviour are subject to considerable uncertainty. The uncertainty can be loosely divided into two categories: (1) a climate model's ability to suitably project into the future the climate and transient (synoptic-scale) weather features that determine the TC life-cycle, and (2) the climate models ability to resolve or at least adequately represent smaller scale processes fundamental to the TC life-cycle. Progressive improvements to climate modelling systems reduce the first category of uncertainty. Downscaling and statistical techniques can help reduce resolution issues responsible for the second category of uncertainty. However, both introduce new sets of largely unquantifiable uncertainty, which brings into question our ability to monitor TC behaviour in climate models.

A possible solution to the TC monitoring problem is to identify the key large-scale flow properties that influence TC behaviour, i.e., properties that are resolvable by coarse resolution models or identifiable in coarse resolution reanalysis data. A TC formation parameter is introduced that has been designed to identify imminent TC formation in such data. If successful the parameter will be directly applicable to climate model data, and will identify TC formation, location and decay, with less uncertainty than other methods. The proposed formation parameter (FP) identifies locations of imminent formation, unlike existing genesis parameters that identify larger-scale regions favourable to formation.

Recent TC genesis studies have identified the importance of curvature vorticity and humidity throughout the lower- to middle-troposphere in the immediate vicinity of TC genesis. Inspired by these results the FP in its current form requires enhanced curvature vorticity at 850 and 500 hPa, and high relative humidity (RH) at 950 and 700 hPa. Similar to existing genesis parameters the FP also includes an upper limit on the 850 to 200 hPa wind shear. The curvature vorticity term is the product of a normalized Okubo-Weiss parameter and absolute vorticity (hereafter OWZ). The normalized Okubo-Weiss has a value of one for solid body rotation, zero for pure shear vorticity, and zero for pure shear and stretching deformation. The curvature vorticity term is also multiplied by the sign of f to give positive values for cyclonic curvature vorticity in both hemispheres.

The FP has been tuned and tested on daily ERA-interim reanalysis data for a region in the western Pacific (30° S to 30° N, 120° E to 140° W), and verified against the IBTrACS TC database. A threshold approach is applied to the FP ingredients to identify potential TC formation. The FP is considered favourable at every grid point in which: (i) OWZ exceeds 5 exp-5 1/s at 850 hPa, (ii) OWZ exceeds 4 exp-5 1/s at 500 hPa, (iii) RH exceeds 90 % at 950 hPa, (iv) RH exceeds 70 % at 700 hPa, and (v) The 850 – 200 hPa wind shear magnitude < 12.5 m/s. When two or more neighbouring grid points are “favourable” for three consecutive 0000 UTC reanalyses (i.e., at least 48 hours) the circulation is classified as a TC.

Preliminary testing on a period of 44 months of data has yielded an 82% success rate. Successes include predicted TCs that appear in the IBTrACS database, and non-prediction of non-developing circulations in the database. Failures include false alarms and missed predictions. Of the failures 8% were border-line cases (misses for systems that were only just TCs, and false alarms that predicted a favourable FP for only the three consecutive reanalyses). Unfortunately, successful prediction of non-developers that do not appear in the database cannot be included as successes. These figures suggest the FP will prove to be a useful product for identifying TC formation in global climate models. Some adjustment will be made to the five criteria above for tracking TCs once developed and for identifying changes in intensity. The preliminary testing has shown the FP may also be of use as a TC genesis forecasting tool.