storm_delineation: Storm delineation methods

storm_delineationR Documentation

Storm delineation methods

Description

These two functions assign determine whether each time point in a time series of rainfall totals is in a storm event or not. The cso_storm function implements an algorithm that has been used to determine storm events for CSO program reporting and analysis. The pot_storm method implements a more general purpose peak over threshold method along with a minimum inter-event time. Any rain value over the threshold is part of a storm and time stamps less than iet distance from each other are considered part of the same storm.

Usage

cso_storm(x, time, threshold = 0.1, iet = 10)

pot_storm(x, time, threshold, iet)

Arguments

x

a vector of rain amounts

time

a vector of time stamps, used to check order if present. If this argument is not used, the vector of rain amounts x is assumed to be ordered

threshold

a scalar representing the method threshold - this parameter has different interpretations for each delineation method

iet

a scalar representing the minimum inter-event time duration - the parameter has essentially the same meaning across methods, but the restriction is enforced in very different ways and so the results will differ across methods

Details

The cso_stor storm detection algorithm is documented in OneNote and a pdf document maintained by the ASM team. Please contact them for access to the most recent version. Jason Law has also documented the algorithm in a memo with more detailed mathematical description of the algorithm. The algorithm relies on start points that occur when it is currently raining, a forward looking rolling sum is over the threshold, and the previous time points within the IET (i.e., (i-iet+1):(i-1) time points where the current point is i. A storm continues if the previous value is a storm and the current value is not a stopping point. The current value is a stopping point if, a backward sum is not over the threshold and the forward


jasonelaw/BESdata documentation built on May 18, 2024, 2:01 p.m.