Product Usage
Converting Probabilistic Forecast to Deterministic
The probabilistic forecasts generated by NCWF2 can be converted to a deterministic (storm / no storm) forecast by choosing a probability threshold that must be exceeded before storms are consider a "likely" hazard to aviation at a specific location. However, as is eloquently stated by Wilks (2001) who says that the “Conversion from probability to categorical information (e.g., storm / no storm in this case) inevitably results in the loss of information, to the detriment of the forecast users. So, rather than choose this threshold a priori, NCWF2 allows the user to determine the threshold for action (e.g., closing routes or corner posts) based on local experience (how have predicted storm coverages impacted local air traffic in the past).The choice of probability threshold may depend on a number of factors such as air-traffic congestion, the mixture of aircraft type (e.g., private, commercial, regional) in the sector, and other mitigating circumstances.
From a forecast accuracy standpoint, it has been found that the 0.4 threshold has the best CSI/bias performance of all the probability levels at the national scale. The probabilistic forecast can be converted to deterministic by simply contouring at the desired probability level (toggle figure to see conversion from probabilistic to deterministic).
Probabilistic forecast of convection of VIP of 4 or greater (dark pinks = high likelihood of strong storms). Mouse over the image to see deterministic forecast derived
from the probability field using a 0.4 probability threshold (black contour). Shaded areas indicate regions of probability exceeding 0.75.
Implementation into Decision Making NEXT >