13.4. Development of IDF and DDF Curves

• Long-term rainfall records from reliable meteorological stations.
• Hourly or sub-hourly rainfall data for short durations (e.g., 5-min, 15-min).
• Annual maximum series or partial duration series are used.

To develop Intensity-Duration-Frequency (IDF) and Depth-Duration-Frequency (DDF) curves, we need specific data. First, we require long-term rainfall records collected from dependable meteorological stations that have measured rainfall consistently over many years. This helps ensure that the data reflects various weather patterns and events.

Next, we need short-duration rainfall data, ideally hourly or even shorter time intervals such as 5-minute or 15-minute intervals. This granularity is important for capturing the intensity of rainfall during brief, intense events. Finally, we utilize two specific types of rainfall series: the annual maximum series, which records the highest rainfall amounts in each year, and the partial duration series, which captures significant rainfall events irrespective of the year. Together, these datasets form the foundation for accurately constructing IDF and DDF curves.

Think of building a library of rainfall records as collecting stories from a neighborhood over the years. You want the most memorable stories that capture intense events (like a flood), which would be like our annual maximum series, as well as stories that include many interesting moments, even if they aren't the absolute highest (like the partial duration series). The goal is to have a complete picture of varying rainfall events, which will help us when we build drainage systems as if we were creating solutions inspired by those stories of weather events.

• Gumbel Distribution (most commonly used)
• Log-Pearson Type III
• General Extreme Value (GEV) Distribution The data is statistically fitted to these distributions to derive rainfall values for different return periods.

Once we have the required data, the next step is to conduct frequency analysis to understand how often certain levels of rainfall occur. This is typically done using statistical distributions. The three main distributions used in this context are the Gumbel Distribution, Log-Pearson Type III, and General Extreme Value (GEV) Distribution.

The Gumbel Distribution is particularly popular for modeling extreme values, such as heavy rainfall events. The Log-Pearson Type III distribution helps assess flood risks and is useful for estimating return periods based on longer records. The GEV Distribution also assists in this effort by providing a framework for analyzing extremes of different distributions. Each of these statistical distributions helps us fit our rainfall data so we can predict how frequently specific rainfall intensities will occur over designated return periods, which are essential for designing infrastructure.

Imagine you're trying to understand how often it rains heavily in your town. You might take an average of past events to predict future rain. This is like how we analyze rainfall data using different statistical methods, similar to using various lenses to view the same painting to see different details. Each lens, or distribution, gives us insights into when we might expect another heavy rain, just as different perspectives can help us appreciate the overall picture.

--