Assimilation of CYGNSS Ocean Surface Winds in HWRF

Subtitle: Preparing for post-launch assessment of impact on hurricane analysis and forecasts

In this poster, we will describe a number of activities underway to prepare for the near real-time evaluation of CYGNSS winds on hurricane analysis and forecasting after launch in November 2016. We are preparing operational or near-operational configurations of the Hurricane Weather Research and Forecasting (HWRF) model to assess the impact of CYGNSS data in near real-time during the 2017 hurricane season.

For a new observation type like CYGNSS, there are numerous steps to prepare for near real-time testing.  We are using simulated CYGNSS wind speed retrievals to examine the:

• sensitivity of impact to different DA techiques (e.g., 3dVAR, EnKF) 
• sensitivity of impact to the observable assimilated (e.g., scalar wind speed, derived wind vectors)
• sensitivity to impact of the frequency of assimilation (e.g., hourly, 3-hourly or 6-hourly DA cycling)
• observation pre-processing and quality control (e.g., superobbing, observation error estimation)

The simulated data were generated by the CYGNSS Science Team's end-to-end simulator (E2ES) which samples David Nolan's mesoscale hurricane nature run (Nolan et al., 2013). We are modifying the 2015 operational HWRF assimilation and forecast model to include CYGNSS observation preprocessing that applies a variational analysis method to derive vector winds for assimilation as conventional ship observations with GSI.  We will also share the plans for near real-time impact experiments and evaluation that will be conducted after launch.

References:

Nolan, D. S., R. Atlas, K. T. Bhatia, and L. R. Bucci (2013), Development and validation of a hurricane nature run using the joint OSSE nature run and the WRF model, J. Adv. Model. Earth Syst., 5, 382–405, doi:10.1002/jame.20031.