In a previous post, I wrote why we need to run ensemble analysis to get a feel of the uncertain of a model, so that we can be more certain about the risks we are facing.
In practice, running years of years of historical storms is not practical and therefore, running an ensemble of representative storms in most cases is the best way to estimate the risks.
In this article we will explain how Atlas 14 is prepared, and that will help us to decide which rainfall profiles to select for our ensemble. Most of this article is based on Jimmy’s blog post below,
NOAA Atlas 14 shows a range of typical storms with different duration and temporal distribution derived from the analysis of historical rainfall data for various regions of the United State. An example is shown below, all the observed storms were classified into groups by,
- Duration: how long the the storm lasted
- Region: for Texas there are 3 regions as shown in the map
- Quartile: the 1st, 2nd, 3rd and 4th quartiles correspond to whether the largest percentage of total rainfall occurs in the first, second, third or fourth quarter of the storm duration
Once all the storms are classified, a probability temporal distribution can be developed from the data for each of the cell in the table above. The result is a group of cumulative percentage curves for different probabilities. For example, the 10% curve indicates only 10% of the rainfall falls in the area left and above the curve (red area), which means, only 10% of the storms dump more rainfall in the first 30% of the duration than the 10% curve.
If you are interested in the details of how these storms were developed, refer to the technical references from the NOAA website.
