Tropical cyclone intensity estimation using temporal analysis and spatial features in satellite data

Tropical cyclones (TCs) are becoming an increasing threat to life/property and hence are critical to understand them with more certainty. Accurate estimation and prediction of TC's intensity is a paramount. Developing new automated techniques to estimate the TC intensity and to overcome the existing errors in estimation still is a challenge. We have developed and tested automated method to estimate TC intensity based on only the existing historical data. The Hurricane Satellite data (HURSAT–B1) is used to develop the algorithm. Algorithm development focuses on the North Atlantic from 1995-2005. Temporal information provides a priori estimates of storm intensity prior to any satellite analysis. The temporal analysis uses the age of the cyclone, 12 and 24 hours prior intensities as predictors of the expected intensity. The algorithm is trained on 80% of the storms and verified with the remaining 20% of data. Instead of regression techniques, the 10 closest analogs (determined using a K-nearest-neighbor (K-NN) algorithm) are averaged to estimate the intensity. Such an estimate has a 9.6 knot RMS error (50% of points are within 5 knots). The spatial analysis uses spatial feature analysis to correlate features with storm intensity ranges. The present spatial technique analyzes 71% of storms at the correct Saffir-Simpson category. The next step in algorithm development is to combine the temporal analysis with the satellite image analysis.