Exploring hodograph-based techniques to estimate the velocity of right-moving supercells

Five supercell motion forecast algorithms are investigated with respect to their hodograph-analysis parameters. The algorithms employed are Maddox (1976), Colquhoun (1980), Davies and Johns (1993), Rasmussen and Blanchard (1998), and Bunkers et al. (2000). Another method derived from the data presented herein, called the “offset” scheme, is used to explore a simple relationship between the observed storm motions and the mean wind. The hodograph-analysis parameters used in these algorithms seem to have been chosen arbitrarily. These arbitrary parameters are essentially the top and bottom levels of the mean wind layer, and a deviation vector from the mean wind defined through that layer. More recently, the Bunkers and RB98 schemes have implemented the vertical wind shear vector, in association with numerical modeling studies. The current work explores the sensitivity of the algorithms to the aforementioned arbitrary parameters by systematically varying those parameters, using a dataset of 394 right-moving supercells, and associated proximity soundings. It has been found that, among other results, the Bunkers scheme is more sensitive to the depth of mean wind layer than it is to the depth of the vertical wind shear layer. It has also been shown that, when using M76-type schemes, a more accurate forecast is obtained by using deep non-pressure-weighted mean wind layers (i.e., greater than SFC-10 km). Indeed, all the forecast schemes show a strong tendency for the u-component of the predicted storm motion to be regulated by the depth of the non-pressure-weighted mean wind layer. The v-component, on the other hand, appears to be regulated by the deviation vector from that mean wind. The results demonstrate also the inherent difficulty in using an observed hodograph to predict supercell motion. Although the shear-relative schemes are shown to perform better on the average than schemes based on the mean wind alone, there are times in which they also result in large forecast errors. Finally, it is suggested that some modifications of the originally recommended hodograph-analysis parameters be made in order to attain the minimum average forecast error for each scheme.