I have hourly water levels of a river station which was subjected to 45 floods in a 30-year period. Please recommend the best method (POT or annual maximum) for flood frequency analysis. The data shows a slightly decreasing trend, may be due to lowering the riverbed level due to sand mining. What is the statistical test suitable to check the significance of the trend and how to proceed the analysis if the trend is higher than the acceptable value?
Annual (Maximum) series or AM also called block (annual) maxima and Peak over Threshold (or POT) are the two frequently modelling techniques widely accepted in the flood frequency analysis or FFA. In such circumstances, the event-based or AM series usually defining the random distributions at an annual scale for each targeted or study site by pointing the maximum streamflow discharge value i.e., flood peak. On the other side, POT based flood sampling includes all the peak values above the pre-defined threshold values, not just by considering a single peak like the AM series. Therefore, such restrictions in the AM based flood modelling sometime might be problematic, if the second largest peak in the same years is larger than the other year samples events.
On another side, POT based event sampling could demand for time independency between the two-consecutive peak within the same observation year, which usually a complicated effort during the selection of flood peak.
In my early training, the POT was called partial duration series. I learned the annual maxima was the typical approach, but for instances of frequent severe events, the partial duration series insures that secondary flood peaks of significance are not disregarded. I would consider both approaches and compare results.
Most stream morphology adjusts to the bankfull event, which is typically every 1.5 years, but this can vary some with circumstances. Channels that aggrading, such as braided streams, flood more frequently and bankfull indicator determination ia probably difficult due to instability and frequent adjustment. Sand mining instream typically causes upstream degradation due to change in base level, and the downstream area can have excessive bank or channel erosion due to the clean water effect as materials are settling into the mined depression areas. So If mining is occurring in the gauged stream reach or vicinity, it could alter the stage discharge relationship. Most gauging stations have staff gauges such as installed on bridge piers or stable location, and regular stream gauging with meters or doppler meters would be used to check for consistency and adjust the stage discharge relationship as needed. It is always good to secure total station or similar cross sections with reliable benchmarks on elevation, as stream morphology can adjust with conditions. Sand mining, especially if outside of the annual input rates can especially cause adjustments. Dredging materials too close to stream banks causes excessive vibrations and even immediate bank failure which can be visible or measured with bank pins.
If for instance the stream gauge was within a braided or aggraded section, I could easily see 45 floods in a 30 year period, and the sand mining may be used to help the stream maintain a primary channel for navigation, etc. Under these type conditions, the stage discharge relationship may not be stable, requiring frequent adjustment for reasonably accurate flood estimates. Rosgen showed us how to install both bank pins and channel chains to check when the banks were adjusting, and situations when the channels degraded during the storm or flood event, and then may or may not refill in the after flood sediment loading. If there are other streams being gauged in the same general physiographic area and rain/climate zone, you may look for any stations with longer records such as 100 year, and not have some of the mining issues to consider. If nothing else, check for aerial photos over the 30 years, and look for channel adjustments. On the Broad River, I noted from aerial photos, that the sand mining downstream areas showed sand bars which were truncated and eroded on the downstream portion of the point bars, and the undercutting of the banks probably contributed to bank trees tipping or rotating toward and sometimes into the channel. Matt Kondolff in California published on the clean water effect in sand mining that you might want to look up.
I normally use RAINBOW software for frequency analysis. It uses Weibull. If you need more help just contact me.
- Badges
- Science topic
- Similar topics
- Electrochemistry
- Battery