Most small dams are not actively managed to affect floods, and may contain some storage to reduce flooding for those with more frequency.  For low frequency severe floods, these small dams may overflow without proper relief and design.  If they fail, they will add, and sometimes substantially to flood severity.  Larger dams are more likely to be managed, and with better forecasting should be able to affect flooding severity.  But if they fill too high, they too may extend flooding duration or extremes in failure or rapid drainage if management of flows made some miscalculations. 

In relation to the precipitation recurrence interval the dam won't affect their forecast, since the precipitation is related with the climate of the region. However, the dams will affect the behavior of the runoff of a watershed, since it cause the damping and delay of the flood wave. The extent of such impact, will vary with the size and type of the dam.

Back to the question. I think you could seperate your flow data into two categores:

• Daily annual maximum, before the dam
• Daily annual maximum, after the dam

Then you could fit a probabilistic function for each category and with this probabilistic function estimate the recurrence related with a flow.

Added to my initial response.  Besides effecting the timing of downstream flow, dams also influence the delivery of sediment downstream.  The timing, magnitude and duration of flow is what can be somewhat managed with water control and release structures.  Dams that fail to deliver at least the bankfull event on a regular basis, such as annually or every other year, may face encroachment of vegetation on the downstream channel, which can increase roughness and flooding potential, due to loss of channel capacity.  

Dams also capture sediment, and the downstream releases of flows from dams often have a clean water effect which can cause channel degradation and instability in some conditions.  It is likely that this clean water erosion effect in some circumstances, such as in relation to braided streams, may actually have some positive benefit in promoting erosion of the braided system and returning to a single thread system.  However, the newly entrained sediments will be delivered downstream.  

Understanding flooding frequency and severity of course depends on climate, rainfall, but there are many other site specific, watershed, and valley/stream type factors that can influence flooding frequency and severity relative to a specific location.   Most studies suggest that forested watersheds are best able to control flooding severity due to deep roots and higher evapotranspiration as compared to other land uses such as cultivated, range grasslands, and urban areas with high levels of impervious surfaces.  It is difficult to generalize to every circumstance, and that is why it is good when civil engineers, hydrologists, geologists, soil scientists, watershed managers, developers and communities can discuss the differences to expect when hydrologic modifications as dams are considered.

Even with well thought out flood frequency analysis and design, all must understand the limitations of available flood data often is less than a century, so projections include a level of uncertainty that may be easy to overlook.  But when a dam, or a series of dams may ultimately flood or even destroy your city, the level of uncertainty and need for dam maintenance and careful management must be factored in.  Since small dams seldom get the level of scrutiny, design and management, extreme flooding analysis may need to include an analysis of added effects of small dam failures contributing to the severity.  Forecast flooding in most cases is going to address the flooding of the specific storm, and its expected flood potential, but in extreme storm/flood conditions, it may not include or consider the dynamics that dams add, such as the untimely releases or failures from existing dams that may complicate and increase flood severity or duration.

Lest we forget, dams are not permanent water storage and flow management structures, and our ability to evaluate flood frequency and risk with precision has limits.  The addition of dams complicates the considerations of extreme events due to their ability to store water, alter flow regime with the added risk in failure.  More attention to factoring in uncertainty, considering climate change, future development, land use change, and channel adjustments to alterations in flow and sediment regime will help us realize these questions are complex and not easily answered.  We seldom have the undisturbed forest circumstance with 50 to 100 years of rainfall and stream flow data and with LiDAR elevation detail, watershed and stream surveys of channel geometry and morphology. Knowledge of past and existing land uses and practices affecting hydrologic response is typically incomplete.  Technical and public understanding of flood frequency and risk and flood zone mapping needs substantial improvement so informed understanding and responsible decisions can be made.

There may be some potential for mitigation of flooding effects to existing development for those who can afford retrofitting structures, altering dam management toward flood control emphasis or recognizing conditions of channel encroachment by vegetation or aggradation by sediment as they contribute to flooding frequency and severity.  As with most water resource management questions, there are trade-offs with needs for drinking, irrigation, recreation, power, fish and other aquatic live, navigation, etc.

I agree and appreciate the scholarly explanations by Dr. Francisco and Dr. Hansen. Basically dams are built to serve as water conservation and flood control structures. These dams are expected to control the floods in terms of severity and damages. It is therefore expected that the frequency and/or intensity of floods will be reduced due to the construction of dams in a given region.