This transfromation is the opposite of selectNumberedTidalPeaks.
The input of the transformation is two scalar time series. The first one is non-equidistant and contains the times of the astronomical low tide. The values are not the levels, but the tide number. The second input is a regular, continuous time series that contains water levels. This can be a forecast, an observed or hindcast series.
The transformation will use the time stamps of the astronomical tide number input, and it will search for low tide values in the continuous time series around each of those times. The search window can be skewed around this astronomical extreme as the period before and after the extreme can be specified individually using minutesBeforeTidalExtreme and minutesAfterTidalExtreme.
Within this search window, the slope of the continuous timeseries is determined over the regression period (regressionPeriodInMinutes). This slope is determined looking forward for every timestep in the continuous series. A minimum or maximum is found if the slope is horizontal (thus the value of the angle is 0). If the slope value changes from positive to negative, it is a maximum, if it changes from negative to positive, it is a minimum.
...