Precipitation Recurrence Intervals for Model My Watershed's Site Storm Model

Model My Watershed's Site Storm Model allows users to specify the amount of precipitation to be applied in the model. Typically, a user will focus on modeling a quantity of rainfall that corresponds with an annual exceedance probability. For example, a user may be interested in modeling the 24 hour precipitation that has a 50% probability of occurring in any given year (meaning that the amount of precipitation recurs only every 2 years) or a 2% probability of occurring in any given year (the amount of precipitation recurs only every 50 years). Sometimes these precipitation exceedance probabilities are referred to as the “2 year storm” or the “50 year storm”, respectively. This Resource shares the results of an analysis that can guide use of the Site Storm Model, particularly for areas of interest in the mid-Atlantic region of PA, NJ, and NY.

The HydroShare Resource "Modeling Future Climate for Model My Watershed" (Ensign, S. 2020. Modeling Future Climate for Model My Watershed, HydroShare, https://doi.org/10.4211/hs.60058ceda8334e68be141516c5b8de3f) demonstrates an algorithm for generating sequences of stochastically-selected precipitation events and inter-event durations that replicate the observed frequency and magnitude of annual weather patterns at a location. This stochastic weather-generating algorithm was used to predict a 20 year time series of daily precipitation using the predicted increase in precipitation from down-scaled global climate models. These 20 year time series predict 2080-2100 weather at eleven weather observing stations which are used by Model My Watershed's Watershed Multi-Year Model for areas of interest within the Delaware River Basin.

The depth-duration-frequency curves presented in this Resource were derived from this stochastic weather-generating algorithm at the eleven weather observing stations described above. For each weather observing station, ten iterations of 20 year stochastic weather were generated for each of the Representative Concentration Pathways (RCP) 4.5 and 8.5. From each of sets of 10 time series, the series with the lowest and highest total precipitation over the 20 year period were selected for further analysis. Following the methods outline in Maimone et al. 2019 and AlHassoun 2011, the frequency factors for a Gumbel type I extreme value distribution were used to generate depth-duration-frequency predictions at 2, 5, 10, 25, 50, 100, 200, 500, and 1000 year intervals for the low precipitation and high precipitation series. Because the RCP 4.5 and 8.5 stochastic precipitation time series were generated independently, the RCP 8.5 predictions were not a simple increase over the RCP 4.5 predictions.

For comparison with these predictions of future weather conditions at the end of the century, the historic precipitation frequency at the 11 locations of interest was downloaded from the NOAA Hydrometeorological Design Studies Center (https://hdsc.nws.noaa.gov/hdsc/pfds/index.html). The data requested included precipitation depth, metric units, Annual Maximum Series, the 24 hour estimates and upper and lower 90% confidence intervals of Annual Exceedence Probabilities (1/year) from 1/2 to 1/1000. These data are plotted along with the future projections.

There are two files in this resource for each weather observation location used by Model My Watershed's Watershed Multi-Year Model for areas of interest inside the Delaware River Basin. There is a table of exceedance probabilities listing precipitation in millimeters and a corresponding figure.