Despite marked progress over the past several decades, convective stormnowcasting remains a challenge because most nowcasting systems are based onlinear extrapolation of radar reflectivity without much consideration for othermeteorological fields. The variational Doppler radar analysis system (VDRAS) isan advanced convective-scale analysis system capable of providing analysis of3-D wind, temperature, and humidity by assimilating Doppler radar observations.Although potentially useful, it is still an open question as to how to usethese fields to improve nowcasting. In this study, we present results from ourfirst attempt at developing a Support Vector Machine (SVM) Box-based nOWcasting(SBOW) method under the machine learning framework using VDRAS analysis data.The key design points of SBOW are as follows: 1) The study domain is dividedinto many position-fixed small boxes and the nowcasting problem is transformedinto one question, i.e., will a radar echo > 35 dBZ appear in a box in 30minutes? 2) Box-based temporal and spatial features, which include time trendsand surrounding environmental information, are elaborately constructed, and 3)The box-based constructed features are used to first train the SVM classifier,and then the trained classifier is used to make predictions. Compared withcomplicated and expensive expert systems, the above design of SBOW allows thesystem to be small, compact, straightforward, and easy to maintain and expandat low cost. The experimental results show that, although no complicatedtracking algorithm is used, SBOW can predict the storm movement trend and stormgrowth with reasonable skill.
展开▼
