Enhanced Observations of Rainfall Rates on Pacific Ocean TAO Buoys

As part of the transition of the Tropical Atmosphere Ocean (TAO) Array from NOAA's Pacific Marine Environmental Laboratory (PMEL) to the National Data Buoy Center (NDBC), obsolete sensors that were part of the original TAO Legacy buoys have been replaced by commercial off-the-shelf (COTS) sensors on the updated TAO Refresh buoys. A significant change in the new Refresh buoys is a shift in processing data from on board the buoy to a shore-side processing systems, rain data is one of the observations affected by this processing shift.

Historical observations collected by TAO Legacy buoys were transmitted once per day via the Argos satellite system. As a result, the rain data was limited to a single daily average of rainfall rate, standard deviation, and percent time raining. In contrast, the current system employed by the TAO Refresh buoys transmit 60 data messages per hour via the Iridium satellite system. The rainfall rates are then calculated shore-side using a similar method as historical observations. NDBC also utilizes existing quality control and the post-recovery data processing methods for the real-time data. To maintain consistency of observations and preserve data quality these real-time processes were compared to the existing post-recovery processes using recently recovered Refresh data. Initial results from the Refresh buoy deployed at 00° 170°W from May 2012 through November 2013 show a mean rain rate of 0.0172 mm/hour when processed through the real-time algorithm versus a mean rain rate of 0.0154 mm/hour from the post-recovery algorithm for a percent difference of 11%.

With the enhanced capabilities of the TAO Refresh array, NDBC is better positioned to support timely analyses of high frequency rainfall events through reduced latency of data and, at the same time, empower users of the high-resolution data to create their own custom filters and statistics to meet their specific needs. An added benefit of the Refresh technology is the reduced sampling error due to the temporal averaging. Data analysts are now able to pinpoint a sensor malfunction down to a specific time rather than a day.