1:45 PM - 3:15 PM
[AAS04-P12] Assimilation of Observational Data by an Autonomous Surface Vehicle Just Below the Typhoon with WRF-3DVAR
Keywords:Autonomous Surface Vehicle, Typhoon forecasting, Data assimilation, Weather Research and Forecasting Model, Three-dimensional variational data assimilation system, Typhoon observation
For the first attempt, we assimilate sea level pressure (SLP) and temperature (SLT) as the observations. The assimilation time was set at 12 UTC on Aug. 31, 2022, when the ASV was closest to the typhoon center. This time is coordinated to the time resolution of NCEP GDAS/FNL, where NCEP GDAS/FNL are available every 6 hours for the initial and boundary conditions. At this time, the ASV was located about 23 km from the typhoon center. The respective values to be assimilated are 951.226 hPa (SLP) and 302.257 K (SLT), and both obtained as averages every 5 minutes by a water-resistant Baro Troll (In-situ Inc.) mounted at the height of 50 cm above the mast of the ASV.
To obtain the first guess, WRF was run for several hours beforehand. To obtain the WRF forecast output and see the assimilation effect on the typhoon forecast, WRF was run for several days with the analysis from the WRFDA for 3DVAR. The procedure is as follows: run WRF for 6 hours from 06 UTC on Aug. 31 with initialization by NCEP GDAS/FNL for spin-up and obtain the first guess (step 1); run WRF-3DVAR on the first guess to assimilate the observation data and produce the analysis at 12 UTC on Aug. 31 (step 2); run WRF for 24 hours from 12 UTC on Aug. 31, with the analysis obtained in step 2 as the initial condition, and obtain the WRF forecast output at 12 UTC on Sep. 1 (step3); run WRF as in step 3 with the first guess obtained in step 1 as the initial condition, and obtain the WRF forecast output without assimilation (step4); compare the WRF forecast outputs obtained in step 3 and step 4 as data with assimilation (DA) and without assimilation (NO-DA), respectively (step 5).
As a result, the pressure decreased by about 20 hPa in DA compared to NO-DA. The assimilation effect is large because of the scarcity of data near the typhoon center, even though the input value was at a single point and a single time. As the next step, we will test the effects of multi-temporal data. In addition, we will perform various experiments using oceanographic data such as sea surface temperature (SST) and investigate the atmosphere-ocean interaction just below the typhoon. We will then evaluate the impact of the observations near the typhoon center by comparing the real-time forecasting obtained by using other observations around the world.