Deep Learning Experiments for Tropical Cyclone Intensity Forecasts

Reducing tropical cyclone (TC) intensity forecast errors is a challenging task that has interested the operational forecasting and research community for decades. To address this, we developed a deep learning (DL)-based Multilayer Perceptron (MLP) TC intensity prediction model. The model was trained using the global Statistical Hurricane Intensity Prediction Scheme (SHIPS) predictors to forecast the change in TC maximum wind speed for the Atlantic Basin. In the first experiment, a 24-hour forecast period was considered. To overcome sample size limitations, we adopted a Leave One Year Out (LOYO) testing scheme, where a model is trained using data from all years except one and then evaluated on the year that is left out. When tested on 2010–2018 operational data using the LOYO scheme, the MLP outperformed other statistical-dynamical models by 9-20%. Additional independent tests in 2019 and 2020 were conducted to simulate real-time operational forecasts, where the MLP model again outperformed the statistical-dynamical models by 5-22% and achieved comparable results as HWFI. The MLP model also correctly predicted more rapid intensification events than all the four operational TC intensity models compared. In the second experiment, we developed a lightweight MLP for 6-hour intensity predictions. When coupled with a synthetic TC track model, the lightweight MLP generated realistic TC intensity distribution in the Atlantic Basin. Therefore, the MLP-based approach has the potential to improve operational TC intensity forecasts, and will also be a viable option for generating synthetic TCs for climate studies.

Adjoint-Derived Impact of Assimilated Observations on Tropical Cyclone Intensity Forecasts of Hurricane Joaquin (2015) and Hurricane Matthew (2016)

The adjoint-derived observation impact method is used as a diagnostic to derive the impact of assimilated observations on a metric representing the forecast intensity of a tropical cyclone (TC). Storm-centered composites of observation impact and the model background state are computed across 6-hourly analysis/forecast cycles to compute the composite observation impact throughout the life cycle of Hurricane Joaquin (2015) to evaluate the impact of in situ wind and temperature observations in the upper and lower troposphere, as well as the impact of brightness temperature and precipitable water observations, on intensity forecasts with forecast lengths from 12 to 48 h. The compositing across analysis/forecast cycles allows for the exploration of consistent relationships between the synoptic-scale state of the initial conditions and the impact of observations that are interpreted as flow-dependent interactions between model background bias and correction by assimilated observations on the TC intensity forecast. The track of Hurricane Matthew (2016), with an extended period of time near the coasts of Florida, Georgia, and the Carolinas, allows for a comparison of the impact of aircraft reconnaissance observations with the impact of nearby overland rawinsonde observations available within the same radius of the TC.

The Role of Inner-Core Moisture in Tropical Cyclone Predictability and Practical Forecast Skill

Errors in tropical cyclone intensity forecasts are dominated by initial-condition errors out to at least a few days. Initialization errors are usually thought of in terms of position and intensity, but here it is shown that growth of intensity error is at least as sensitive to the specification of inner-core moisture as to that of the wind field. Implications of this finding for tropical cyclone observational strategies and for overall predictability of storm intensity are discussed.

On the Predictability and Error Sources of Tropical Cyclone Intensity Forecasts

The skill of tropical cyclone intensity forecasts has improved slowly since such forecasts became routine, even though track forecast skill has increased markedly over the same period. In deciding whether or how best to improve intensity forecasts, it is useful to estimate fundamental predictability limits as well as sources of intensity error. Toward that end, the authors estimate rates of error growth in a “perfect model” framework in which the same model is used to explore the sensitivities of tropical cyclone intensity to perturbations in the initial storm intensity and large-scale environment. These are compared to estimates made in previous studies and to intensity error growth in real-time forecasts made using the same model, in which model error also plays an important role. The authors find that error growth over approximately the first few days in the perfect model framework is dominated by errors in initial intensity, after which errors in forecasting the track and large-scale kinematic environment become more pronounced. Errors owing solely to misgauging initial intensity are particularly large for storms about to undergo rapid intensification and are systematically larger when initial intensity is underestimated compared to overestimating initial intensity by the same amount. There remains an appreciable gap between actual and realistically achievable forecast skill, which this study suggests can best be closed by improved models, better observations, and superior data assimilation techniques.

Evaluation of Atmosphere and Ocean Initial Condition Uncertainty and Stochastic Exchange Coefficients on Ensemble Tropical Cyclone Intensity Forecasts

Tropical cyclone (TC) intensity forecasts are impacted by errors in atmosphere and ocean initial conditions and the model formulation, which motivates using an ensemble approach. This study evaluates the impact of uncertainty in atmospheric and oceanic initial conditions, as well as stochastic representations of the drag Cd and enthalphy Ck exchange coefficients on ensemble Advanced Hurricane WRF (AHW) TC intensity forecasts of multiple Atlantic TCs from 2008 to 2011. Each ensemble experiment is characterized by different combinations of either deterministic or ensemble atmospheric and/or oceanic initial conditions, as well as fixed or stochastic representations of Cd or Ck. Among those experiments with a single uncertainty source, atmospheric uncertainty produces the largest standard deviation in TC intensity. While ocean uncertainty leads to continuous growth in ensemble standard deviation, the ensemble standard deviation in the experiments with Cd and Ck uncertainty levels off by 48 h. Combining atmospheric and oceanic uncertainty leads to larger intensity standard deviation than atmosphere or ocean uncertainty alone and preferentially adds variability outside of the TC core. By contrast, combining Cd or Ck uncertainty with any other source leads to negligible increases in standard deviation, which is mainly due to the lack of spatial correlation in the exchange coefficient perturbations. All of the ensemble experiments are deficient in ensemble standard deviation; however, the experiments with combinations of uncertainty sources generally have an ensemble standard deviation closer to the ensemble-mean errors.

Advances in Tropical Cyclone Intensity Forecasts

NOAA established the 10-year Hurricane Forecast Improvement Project (HFIP) to accelerate the improvement of forecasts and warnings of tropical cyclones and to enhance mitigation and preparedness by increased confidence in those forecasts. Specific goals include reducing track and intensity errors by 20% in 5 years and 50% in 10 years and extending the useful range of hurricane forecasts to 7 days. Under HFIP, there have been significant improvements to NOAA's operational hurricane prediction model resulting in increased accuracy in the numerical guidance for tropical cyclone intensity predictions. This paper documents many of the improvements that have been accomplished over the last 5 years, as well as some future research directions that are being pursued.

The Impact of Targeted Dropwindsonde Observations on Tropical Cyclone Intensity Forecasts of Four Weak Systems during PREDICT

The value of assimilating targeted dropwindsonde observations meant to improve tropical cyclone intensity forecasts is evaluated using data collected during the Pre-Depression Investigation of Cloud-Systems in the Tropics (PREDICT) field project and a cycling ensemble Kalman filter. For each of the four initialization times studied, four different sets of Weather Research and Forecasting Model (WRF) ensemble forecasts are produced: one without any dropwindsonde data, one with all dropwindsonde data assimilated, one where a small subset of “targeted” dropwindsondes are identified using the ensemble-based sensitivity method, and a set of randomly selected dropwindsondes. For all four cases, the assimilation of dropwindsondes leads to an improved intensity forecast, with the targeted dropwindsonde experiment recovering at least 80% of the difference between the experiment where all dropwindsondes and no dropwindsondes are assimilated. By contrast, assimilating randomly selected dropwindsondes leads to a smaller impact in three of the four cases. In general, zonal and meridional wind observations at or below 700 hPa have the largest impact on the forecast due to the large sensitivity of the intensity forecast to the horizontal wind components at these levels and relatively large ensemble standard deviation relative to the assumed observation errors.