Flood Frequency Analysis Techniques

This assignment delves into the field of regional flood frequency analysis, examining different methods employed to estimate flood probabilities in ungauged areas. It covers concepts like the region of influence approach, canonical correlation analysis, and L-moments, while referencing relevant studies and publications. The focus lies on understanding how these techniques are applied to assess flood risks and support water resource management.

Regional Flood Estimation Methods 1
REGIONAL FLOOD ESTIMATION METHODS
By (Name)
Name of the Class
Name of Professor/Tutor
School/University Name
Location (City and State)
The Date

Regional Flood Estimation Methods 2
Contents
Introduction.....................................................................................................4
Derivation of Methods Used in Regional Flood
Estimation……………………………………...4
Region of Influence Approach..........................................................................5
Canonical Correlation Analysis (CCA)..............................................................7
Regional Flood Frequency Analysis..................................................................8
Use of GIS and Remote Sensing Technology in Regional Flood Estimation.....9
Conclusion.....................................................................................................10
References.....................................................................................................12

Regional Flood Estimation Methods 3
Abstract
Regional flood estimation methods are very important in the determination
of expected flood events and time. This provides a very crucial flood warning
system that can be used to warn people of the impending danger of floods.
These methods provide planners and residents of a region with accurate and
reliable information regarding floods and their frequency of occurrences. In
this paper, a systematic review of pertinent literature was done to analyze
these methods. Most of the literature reviewed was obtained from Google
scholar and Google books.

Regional Flood Estimation Methods 4
Introduction
Floods are natural disasters that lead to loss of life and destruction of
property (Syngellakis, 2016). They might be caused by heavy downpours,
poor drainage, or the type of slopes in the area. Therefore floods pose a lot
of danger not only to the community but also to human life. Due to this
danger, different regional flood estimation methods have been developed
with an aim to better predict the occurrence of floods. Without proper
planning and management of floods, it can be a disaster, but if managed
well, it can only be a hazard. The public and planning departments in
government institutions require reliable and accurate estimates of large
floods to promote flood risk management structures and policies (Wohl,
2000, p. 334).
In this paper different regional planning methods will be analyzed. This
will be done via literature analysis of the understanding of different regional
flood estimation methods. Most of the papers analyzed are downloaded from
Google scholar and other credible sites and will form the basis for this
discussion.

Regional Flood Estimation Methods 5
All the different regional flood estimation methods will be analyzed for
their efficiency, accuracy, and reliability. This study will provide a frame work
for understanding flood estimation methods, which can be very useful for
developing policies related to floods forecasting and management.
Derivation of Methods Used in Regional Flood Estimation
The methods for regional flood estimation are divided into three parts:
the selection of the area to be studied, developing a technique for indexing
of flood values for the catchment chosen (based on the physical
characteristics of the catchment), and the development of a regional flood
frequency curve (to enable the estimation of the a flood of a set period from
index values obtained earlier) (Roy and Mistri, 2013). This systematic
approach provides a very good baseline for estimating floods.
Region of Influence Approach
This method focuses on the collection of data from stations in a well-
defined region. This methods is important in enhancing the estimation of at-
site quantiles. In this method, a region of influence is identified for all the
gauging stations which consist of a given set of gauged stations near the
selected station. In order to measure the proximity of each station, a p-
dimensional Euclidian distance space in which the attributes are variables
related to the identification of the stations which are similar in high flow
rates. The model equation for the distance is shown below (Burn, 1990):

Regional Flood Estimation Methods 6
Djk-- Euclidian distance from site j to k
P – Attributes used in measuring the distance
C j
i --standardized values used in the measurement of attribute i, for the site j
(Burn, 1990).
The value of the distance from the equation above gives a measure of how
each station is close to each other (Burn, 1990).
Next, is to identify the region of influence, by choosing a threshold
value that acts as a cut-off point for the distance measures (Burn, 1990). All
the stations whose distance is more than the threshold value are eliminated
from the region of influence. In this method, conventional regionalization
techniques are used to select for the choice of the cut-off value. Another
method of identifying the threshold value is to correlate the candidate
station with the sites near the cutoff value (Burn, 1990). This makes sure
that the stations selected are representative stations for the region of
influence.
A weighting function is used to show the relative significance of each
of the gauging stations in the region of influence in relation to the at-site
extreme flows. This function is depicted by the equation below (Burn, 1990):

Regional Flood Estimation Methods 7
Where:
WFjk --weighting for station k in the region of influence for site j
THL-- parameter
n is a constant
When the region of influence has been determined for each site, it is
now possible to predict the extreme flow rates at each site in relation to all
the information from the other stations that are in the same region of
influence. This gives a better representation of the flow rates, and enhance
flood estimation in that given region (BURN, 1990).
This method has been touted to be very efficient in regional flood
estimation. It is not only efficient but also provide very accurate flood
forecast. The method is also very flexible since it allows the inclusion of
information from surrounding stations in the same region of influence. The
method is also very versatile in that it can be combined with other different
extreme flow rates estimators to provide better results. This is because it is
easy to vary the threshold distance for the region of influence and the
attributes to be used in the measurement of similarity for the stations to be
added in the region of influence, and the weighting function used for

Regional Flood Estimation Methods 8
reflecting the importance of all the stations in the region (Tasker et al.,
1996).
Canonical Correlation Analysis
Another method for regional flood estimation is the use of canonical
correlation analysis. This method has not been widely used but is slowly
gaining popularity in the field of hydrology (Ouarda et al., 2001). When two
sets of variables are represented by flood peaks and watershed
characteristics, their correlation structures can be investigated using
canonical correlation analysis. This method is very important in multivariate
statistics since it provides a framework for factorial discriminant analysis
correspondent analysis and multivariate analysis. It provides a method to
establish the interaction between two groups of variables, through the
identification of linear combinations between the first group and the second
group.
The first attempt to use CCA in hydrology was made by Wong (1963)
and Snyder (9162). Other contributors were, Wallis (1967), Matalas and
Reiher (9167). Torranin attempted to apply the method of CCA in 1972 in
coastal monthly precipitation forecasts. This shows that this method has a
long history in the application of regional flood forecasting. In 1990 Cavadias
initiated the use of CCA in the estimation of maximum annual flood
distribution in Canada. This was a pioneering work that ushered the use of
CCN in regional flood estimation.

Regional Flood Estimation Methods 9
In a hydrological system, flood statistics and catchment attributes are
related by a multiple regression models. This multiple regression model has
residuals that are interpolated spatially using a kriging method, which is
used to minimize biases. In 2004, Ouarda and Chokmani came up with a
kriging procedure method in a physiographical space, which was a
multidimensional space defined by the catchment characteristics. Then they
constructed a physiographical space that represented the distance between
catchments based on their similarity—this was based on their catchment
attributes (Schumann, 2011, p. 110). This enabled them to map hydrological
catchment areas based on their characteristics for regional flood estimation
methods.
According to Kumar and Chatterjee, (2006), CCA can be very useful in
finding homogenous zones or sub regions in the hydrological systems for
reliable, and accurate regional flood estimation—it is efficient, accurate, and
saves a lot of time. Even as this method is advantageous it assumes
similarity of hydrological basins, which naturally is not the case, most
hydrological basins are not similar (Beran et al., 1990, p. 171). This might
introduce an error in the method, which might lead to inaccurate results.
Regional Flood Frequency Analysis
Regional flood frequency analysis was developed by Smith in 1989.
This method was based on a model that related to large quintiles, which is
modeled by a Pareto distribution; that is generalized. In 1991, Arnell and

Regional Flood Estimation Methods 10
Gabrielle developed this method further by incorporating two components:
generalized extreme value and extreme value distributions. They were able
to show that when a large region is divided into sub-regions more precise
estimates can be achieved. Subsequently, Farquharson et al. in 1992 used
the regional frequency curves through a GEV distribution, to map 162
stations in Africa. This shows the power of the method for regional flood
estimation (Hamed and Rao, 1999, p. 60).
More so, it is used to estimate the expected flood quantile of
magnitude Qt at a given project location. The return period T is used to
estimate the rarity of the flooding event. This method also allows for the
forecast of flood quantile estimates in a given site; in relation to the flood
data recorded in other gauging sites found in the same hydrological region
(Cunnane, 1988). That is, if one of the sites does not have flood data, it can
be estimated using other stations in the neighborhood.
Some Regional flood frequency analysis assume that a given region is
homogenous: that all the gauging stations' characteristics are homogeneous.
This allows for estimation of flood volumes using other stations. One of this
methods is the index flood method. This homogeneity allows for highly
accurate estimate’s that are even more accurate compared to at-site
estimation.
Other methods of regional flood frequency analysis do not require
homogeneity of the stations. Some of this methods are the joint multivariate

Regional Flood Estimation Methods 11
estimation method and Bayesian method. However, even though
homogeneity is not required in this methods, it increases the accuracy of the
estimates.
This method mostly relies on regional regression models to estimate
quantiles using physiographic basin characteristics. However, according to
Wohl, 2000 (p. 334) the reliance of hydrographs for this regression models
poses serious challenges since the distribution of the critical inflows and
critical duration is not clear. This ambiguity puts into question the accuracy
and reliability of the method.
Use of GIS and Remote Sensing Technology in Regional Flood Estimation
Most of the conventional means used for flood monitoring and
estimation, fail to record or estimate extreme flooding events (Sanyal and
Lu, 2004). However, remote sensing techniques in collaboration with
geographical information systems (GIS) have the capability to monitor this
extreme events. This makes them be a better method for regional flood
estimation than all the other methods discussed above.
According to Dzurik and Theriaque, (1996, p. 257) most wetlands cover
large areas that are not accessible via conventional means, which makes GIS
and remote sensing to be a very good tool for flood forecasting in this areas.
This remote sensing techniques cover large areas, even the inaccessible
areas, and in collaboration with GIS makes a very good tool for data flood
analysis in the watersheds. GIS tool used in this case provides a digital

Regional Flood Estimation Methods 12
representation of the watershed characteristics, which can be used in
hydrological modeling.
Some of the characteristics of the watershed represented by GIS are
natural ground cover, imperviousness, stream networks, and the delineation
of the watershed (Dzurik and Theriaque, 1996, p. 257). These components
when incorporated into the GIS tool can be used in flood forecasting and
floodplain management. Soil moisture data collected by GIS and remote
sensing techniques can also be useful in flood estimation model(Lijiao Lou et
al., 2014, p. 82).
Even in as much as this method is advantageous, high-level training is
required for the use of GIS and remote sensing (Dijk and Bos, 2013, p. 36).
This makes it not to be accessible to most people. Also, the process is time-
consuming and requires a lot of resources; this makes it an expensive
endeavor compared to other methods discussed above. Since the methods
rely on satellite imageries, it is also susceptible to atmospheric weather
conditions such as cloudiness, and the methods for removing such are also
time-consuming (Wallis J.R, 1988, p. 171). Overall, given the availability of
resources, it can be a very reliable method that would produce accurate
results for even inaccessible areas.
Conclusion
All the methods mentioned are reliable; however, it depends on how
they are employed and used. Overall, this literature review was able to

Regional Flood Estimation Methods 13
establish that GIS and remote sensing was more reliable when large areas
are to be considered, while canonical correlation analysis was least used of
the methods.
References
Beran, M., Water, I.I.C, 1990. Regionalization in hydrology. International
Association of Hydrological Sciences.

Regional Flood Estimation Methods 14
BURN, D.H., 1990. An appraisal of the “region of influence” approach to flood
frequency analysis. Hydrological Sciences Journal, 35(2), pp.149–165.
Cunnane, C., 1988. Methods and merits of regional flood frequency analysis.
Journal of Hydrology, 100(1), pp.269–290.
Dijk, A. van and Bos, M.G., 2013. GIS and Remote Sensing Techniques in
Land- and Water-management. Springer Science & Business Media.
Dzurik, A.A. and Theriaque, D.A., 1996. Water Resources Planning. Rowman
& Littlefield.
Hamed, K. and Rao, A.R., 1999. Flood Frequency Analysis. CRC Press.
Kumar, R. and Chatterjee, C., 2006. Closure to “Regional Flood Frequency
Analysis Using L-Moments for North Brahmaputra Region of India” by
Rakesh Kumar and Chandranath Chatterjee. Journal of Hydrologic
Engineering, 11(4), pp.380–382.
Lijiao Lou, Baojiang Liu, and Mengjie Jin, 2014. 2014 International
Conference on Information GIS and Resource Management. DEStech
Publications, Inc. Available from: https://books.google.co.ke/books?
Ouarda, T.B.M.J., Girard, C., Cavadias, G.S., and Bobée, B., 2001. Regional
flood frequency estimation with canonical correlation analysis. Journal
of Hydrology, 254(1), pp.157–173.
Roy, S. and Mistri, B., 2013. Estimation of Peak Flood Discharge for an
Ungauged River: A Case Study of the Kunur River, West Bengal
[Online]. Available from:

Regional Flood Estimation Methods 15
https://www.hindawi.com/archive/2013/214140/ [Accessed 25 August
2017].
Sanyal, J. and Lu, X.X., 2004. Application of Remote Sensing in Flood
Management with Special Reference to Monsoon Asia: A Review.
Natural Hazards, 33(2), pp.283–301.
Schumann, A.H., 2011. Flood Risk Assessment and Management: How to
Specify Hydrological Loads, Their Consequences and Uncertainties.
Springer Science & Business Media.
Syngellakis, S., 2016. Management of Natural Disasters. WIT Press.
Tasker, G.D., Hodge, S.A., and Barks, C.S., 1996. Region of influence
regression for estimating the 50‐year flood at ungaged sites. JAWRA
Journal of the American Water Resources Association, 32(1), pp.163–
170.
Wallis J.R, 1988. Environmental Software. Computational Mechanics
Publications. Available from:https://books.google.co.ke/books?
Wohl, E.E., 2000. Inland Flood Hazards: Human, Riparian, and Aquatic
Communities. Cambridge University Press.