Upgrades to the UW-CIMSS AMSU-based tropical cyclone intensity algorithm

Accurate analysis of a Tropical Cyclones (TC) current intensity is a critical part of assessing short-term intensity changes. The Dvorak Technique (DT) has traditionally been the primary method for estimating TC current intensity. However the DT at times may provide conflicting estimates that can vary by more than 30 knots between analysis agencies. Since 2001 CIMSS has been using data from the Advanced Microwave Sounding Unit (AMSU) to estimate TC current intensity. These estimates have been used by TC warning agencies around the world to help augment and in some cases override values provided by the DT. The algorithm consists of a least-squares regression that matches AMSU-A brightness temperatures (Tb) with measured Minimum Sea-Level Pressures (MSLP) from reconnaissance aircraft. Several changes have been made to the algorithm during the last 2 years to improve both the number and the quality of estimates. Seven years of AMSU data from 1998-2005 have been used to fine-tune the algorithm for a variety of storm intensities and structures. Sources of error due to sub-sampling and precipitation contamination have been addressed as well as the issue of estimate confidence.

One challenge with interpretation of the AMSU estimates has been in assessing the estimate confidence. Due to various sources of error and uncertainty some estimates are more accurate than others. For the 2005 season a confidence indicator was developed to assist end users in determining how much weight to give the estimate. The confidence indicators (Good, Fair and Poor) were developed by evaluating the algorithm's skill in a variety of scan geometries and parameter inputs (radius of maximum winds, position, latitude, etc). The confidence includes a range of pressure and wind values that bracket the estimate and scale with TC intensity.

Because of the resolution of the AMSU-A instrument (48 km at nadir to > 100 km at the limb) many TC warm cores are sub-sampled. In order to correct for sub-sampling the CIMSS algorithm employs a statistically based bias correction that compares the AMSU-A Field of View (FOV) resolution to the storms estimated radius of maximum winds (RMW). In the past the RMW parameter came primarily from the TC warning centers. The warning centers provide the RMW through the ATCF messages updated every six hours. However, IR imagery can provide an accurate estimate of the RMW in cases when a well-defined eye is present in the imagery. An IR-based RMW estimate has been made a part of the Advanced Objective Dvorak Technique (AODT). In cases when a clear eye scene is present the AMSU algorithm uses the RMW value from the AODT. The advantage of this method is that the RMW value can be matched to the AMSU overpass time. This also permits more frequent updates of the RMW.

Strong and deep precipitation cores can the have the effect of reducing the AMSU-A measured Tb resulting in a negative impact on the TC intensity estimate. A precipitation correction developed by Robert Wacker (Major, USAF) uses information contained in AMSU-A channels 2 and 15 to address this problem. The result is improved intensity estimates in the presence of heavy precipitation. The most dramatic improvements are seen in weaker storms. The warm core of these storms tends to be centered at lower altitudes near the mean height of channel 7 (~ 250 hPa), which is one of the primary channels used to produce the estimates and more subject to precipitation effects than higher channels.

Results of these changes along with performance of the algorithm in the 2004-2005 seasons will be presented.