Geology Reference
In-Depth Information
presented SB here in a simple case study, focusing into its capability to learn
extreme points from the given training data and then to validate with other models,
the actual modeling capabilities are immense, as the model can
fit the data with
GPD. We hope SB will gain considerable attention from researchers from envi-
ronmental science and applied engineering in the near future.
8.4 Conclusions
Extreme hydrological events have aroused signi
cant interest in hydrology, as such
events have the potential to cause human and economic losses. Meanwhile, pre-
dictive uncertainty and imprecise peak (extreme
flow) estimations are continuing to
be major perplexing concerns in the
field of hydrology. This chapter presented
mathematical details of a new state-of-the-art rare event analysis method called SB,
and introduced it to the
field of hydrology through a case study. In the case study,
the capabilities of SB to predict rare events above several threshold values is
compared with arti
cial neural networks and SVMs. The study has also explored
the capability of the Gamma Test, utilizing the daily data from the Beas catchment
in conjunction with mathematical models such as SVM, SB, and ANNs, and
identified suitable inputs and training data lengths from the study region. This
chapter advocates that conjunctive application of an extreme value theorem with the
machine learning concept (SVM classi
cation in this case) is a very useful tool in
extreme value modeling in applied engineering. In this case study, SB has proved
its capability to offer better predictive accuracy to
find the peaks over a given
threshold, which has signi
cance in such catchments with high variability in dis-
charge volume.
References
1. Agalbjorn S, KonHar N, Jones AJ et al (1997) A note on the gamma test. Neural Comput Appl
5(3):131
-
133
2. Chang CC, Lin CJ (2001) LIBSVM: a library for support vector machines.
http://www.csie.
3. Chang CC, Lin CJ (2011) LIBSVM: a library for support vector machines. ACM Trans Intell
Syst Technol 2:27:1
-
27:27
4. Durrant PJ (2001) Win Gamma: a non-linear data analysis and modelling tool with
applications to flood prediction. PhD thesis, Department of Computer Science, Cardiff
University, Wales, UK
5. Embrechts P, Kl
ppelberg C, Mikosch T et al (2003) Modelling extremal events for insurance
and finance, 4th edn. Springer, Berlin
6. Evans D, Jones AJ (2002) A proof of the gamma test. Proc R Soc Ser A 458(2027):2759
ü
2799
7. Jain SK, Tyagi T, Singh V et al (2010) Simulation of runoff and sediment yield for a
Himalayan watershed using SWAT model. J Water Resour Prot 2:267
-
281
8. Joachims T (1998) Making large-scale svm learning practical. MIT Press, Cambridge
-
