The Impacts of WRF Model Tendency Errors on Tropical Cyclone Intensity Forecast

In this study, the nonlinear singular vector (NFSV) approach is utilized to identify the most disturbing tendency errors of the Weather Research and Forecasting (WRF) model for tropical cyclone (TC) intensity forecast. For the selected nine TC cases, the NFSV-tendency errors of the WRF model, including components of potential temperature or/and vapor, are revealed when the TC intensities are forecasted with 24-hrs leading time. The NFSV-tendency errors present barotropic positive anomalies around the location where the TC will finally move to, and concentrate in the mid atmosphere; furthermore, they are insensitive to the TC intensity and its following development. Especially, the NFSV-tendency errors, compared with initial errors, can cause much larger forecast errors, inferring that NFSV-tendency error is an important source of prediction uncertainty of TC intensity. Specifically, the NFSV-tendency errors enhance the convergence (divergence) in lower (upper) levels and induce intense vertical motion, which decreases the density of atmosphere rapidly, finally increasing the prediction uncertainties in temperature, wind, and moist accordingly. Moreover, the NFSV-tendency errors concentrate their large errors in one region, which may indicate that the tendency errors in this region make larger contributions to the forecast errors and provide ideas to improving TC intensity forecast skill by superimposing an external forcing term to correct the model.