科研之友
微信
新浪微博
Facebook
分享链接
摘要
The satellite-based Dvorak technique (DVKT) is the most widely available and readily used tool for operationally estimating the maximum wind speeds associated with tropical cyclones. The DVKT itself produces internally consistent results, is reproducible, and has shown practical accuracy given the high cost of in situ or airborne observations. For these reasons, the DVKT has been used in a reasonably uniform manner globally for approximately 20 years. Despite the nearly universal use of this technique, relatively few systematic verifications of the DVKT have been conducted. This study, which makes use of 20 yr of subjectively determined DVKT-based intensity estimates and best-track intensity estimates influenced by aircraft observations (i.e., +/- 2 h) in the Atlantic basin, seeks to 1) identify the factors (intensity, intensity trends, radius of outer closed isobar, storm speed, and latitude) that bias the DVKT-based intensity estimates, 2) quantify those biases as well as the general error characteristics associated with this technique, and 3) provide guidance for better use of the operational DVKT intensity estimates. Results show that the biases associated with the DVKT-based intensity estimates are a function of intensity (i.e., maximum sustained wind speed), 12-h intensity trend, latitude, and translation speed and size measured by the radius of the outer closed isobar. Root-mean-square errors (RMSE), however, are shown to be primarily a function of intensity, with the best signal-to-noise (intensity to RMSE) ratio occurring in an intensity range of 90-125 kl (46-64 m s(-1)). The knowledge of how these factors affect intensity estimates, which is quantified in this paper, can be used to better calibrate Dvorak intensity estimates for tropical cyclone forecast operations, postseason best-track analysis, and climatological reanalysis efforts. As a demonstration of this capability, the bias corrections developed in the Atlantic basin are also tested using a limited east Pacific basin sample, showing that biases and errors can be significantly reduced.
- 出版日期2010-10
