Author
|
Submitted to: Precipitation Conference Proceedings
Publication Type: Proceedings Publication Acceptance Date: 1/16/2001 Publication Date: N/A Citation: N/A Interpretive Summary: Rainfall data for watershed models have used 24-hr data. When analyzing these data, the data are often assumed to have occurred from 2400-2400, even though they were not measured during this time. Often the time distributions of storm totals are needed for engineering design and other purposes and a fixed precipitation pattern (design storm) is often used to distribute the "daily" precipitation amounts. However, design storms do not capture the wide variability of rainfall intensities found in natural storms. Furthermore, storm rainfall spanning midnight is not considered when "daily" totals are distributed. Data that have short durations and are measured to a .01 inch resolution are needed for using infiltration formulas and for short-time increment intensity-duration- requency analysis, but are generally unavailable. A storm generator has been developed that simulates storm occurrence, storm duration, storm depth and within-storm precipitation intensities. The synthesized outputs have varying storm durations and depths (durations not constrained by 24-hr times) and have intensities with a time resolution of the order of minutes. The storm generator synthesizes the month, day, year, hour and minute of the beginning and ending of individual storms, storm depth and the within-storm intensities. The model is described and supporting studies of the storm generator are discussed. The model has potential utility for risk analysis for a variety of uses such as for crop pests and diseases, erosion modeling, intensity-duration- requency analysis, flood analysis, drought studies, international agricultural trade and engineering design. Federal, state and university scientists, consultants and regulatory agencies will find it useful. Technical Abstract: Precipitation data for models have used data collected over a 24-hr period. This could be 2400-2400, 0800-0800, or some other observation time. When using these data, they are often assumed to have occurred from 2400-2400, even though they were not measured during this period. Furthermore, often the time distribution of storms is needed for design and other purposes, and a fixed precipitation pattern ("design storm") is often used to distribute the "daily" precipitation amounts. However, design storms do not capture the wide variability of rainfall intensities found in natural storms. Furthermore, storm rainfall spanning midnight is not considered when "daily" totals are distributed. There are scanty short-time increment precipitation data (of the order of minutes) having fine depth resolution (.01 inch). Such data are needed for using advanced infiltration formulas and for short-time increment intensity-duration-frequency analysis. A storm generator has been developed at the ARS experimental watershed facility at Coshocton, Ohio that stochastically simulates storm occurrence, storm duration, storm depth, and within-storm intensities. The model is statistical, and does not include storm physics. The synthesized outputs have varying storm durations and depths (not constrained by artificial 24 hr times), and intensities with a time resolution of the order of minutes. The storm generator synthesizes month, day, year, hour, and minute of the beginning and end of individual storms ("storm" occurrence and duration), storm depth, and within-storm intensities. The fundamental elements of the model are described. The model is undergoing testing and further development. Supporting studies of the storm generator are discussed. The model has potential utility for risk analysis for a variety of uses. |
