Current trends in the optimization of hydraulic flood simulations in ungauged steep rivers (presentation)

Abstract

Presentation for the conference paper of the same title presented at the 10th International Conference in Fluvial Hydraulics (River Flow 2020).
[ABSTRACT]
The numerical tools simulating flood events must be accurate, in order to provide
useful data, and computationally efficient, to facilitate informed decision-making during
a flood. One of the main limitations of modelling software at the present time is the CPU time needed to perform simulations over complex spatial domains. Recent studies show that the models lack the necessary implementations to take advantage of the parallelism available on current hardware, which hinders their use in interesting applications such as real-time flood forecasting, calibration and uncertainty analysis, or visualization and gamification of floods for improved risk communication. For instance, methods based on artificial intelligence have contributed highly to the state-of-the-art of flood simulations, providing better performance and cost-effective solutions than complex hydrodynamic fluid solvers. The current study provides an overview of strategies dedicated to the optimization of hydraulic flood simulations suitable for ungauged steep rivers, which are not extensively studied.

Full text

Current trends in the optimization of
hydraulic flood simulations in
ungauged steep rivers
Adina Moraru, Nils Rüther, Oddbjørn Bruland
10th International Conference on Fluvial Hydraulics (River Flow)
Delft (the Netherlands)
7-10 July 2020
World of Wild Waters

Norwegian University of Science and Technology 2
adina.moraru@ntnu.no
Contents
•Motivation
•Problem characterization and simplification
•Coding and fluid simulations
•Artificial Intelligence in flash flood analysis
•Discussion
•Concluding remarks

Norwegian University of Science and Technology 3
adina.moraru@ntnu.no
Why to optimize hydraulic flood simulations?
The numerical tools simulating flood
events must be accurate, in order to
provide useful data, and computationally
efficient, to facilitate informed decision-
making during a flood.
Complex hydrodynamics affecting steep
rivers call for more efficient hydrodynamic
fluid simulations. Another important
challenge in small catchments is the lack
of reliable historical hydrology and high-
resolution terrain data.
©A. Jørgensen Bruland (2017)

Norwegian University of Science and Technology 4
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
2. Combining coding with hydraulic modelling
3. Implementing Artificial Intelligence in flood analysis
González-Cao et al. (2019)

Norwegian University of Science and Technology 5
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
•Basing the study on optimal terrain data
•Defining the model dimensionality and scale
•Coupling GIS-Remote Sensing and numerical simulations
oIdentifying critical locations

Norwegian University of Science and Technology 6
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
•Basing the study on optimal terrain data
Castellarin et al. (2009)

Norwegian University of Science and Technology 7
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
•Basing the study on optimal terrain data
Castellarin et al. (2009)

Norwegian University of Science and Technology 8
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
•Defining the model’s dimensionality and scale
Ideal cross-sectional spacing:
0.5W* (complex) to 1W* (simpler morphologies)
Modified after Moraru et al. (2019)

Norwegian University of Science and Technology 9
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
•Defining the model’s dimensionality and scale
Modified after Moraru et al. (2019)

Norwegian University of Science and Technology 10
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
1. Characterizing the problem and simplifying it
•Coupling GIS-Remote Sensing and numerical simulations: identifying critical locations
Table 1. Some geomorphic indexes used as indicators for flood analysis.
Geomorphic
index
Estimation
Considered
values
References
Width ratio (Wr)
𝑊𝑟 =Waf
𝑊𝑏𝑓
Wr > 1 for
erosion
Moraru et al.
2018, Ruiz
-
Villanueva et al.
2018, Scorpio
et
al. 2018, and
references
therein
Relative section
ratio (Wxs)
𝑊𝑥𝑠
=Wn
𝑊(𝑛 − 1)
Wxs > 1.1 for
expansions,
Wxs < 0.9 for
contractions
Usman 2019
Width to depth
ratio (W/D)
𝑊/𝐷 = W
𝐷
W/D < 12 for V
-
shaped
channels
Usman 2019
Bend curvature
A, W, M, R,
ϕ, θ,
t
Dependent on
bend angle and
material
Murniningsih
2018

Norwegian University of Science and Technology 11
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
2. Combining coding with hydraulic modelling
•Using algorithms to increase efficiency
•Parallelizing CPU and/or GPU computing

Norwegian University of Science and Technology 12
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
2. Combining coding with hydraulic modelling
•Using algorithms to increase efficiency
Watanabe et al. (2018)
Dependence of calculated water surface level on iteration number

Norwegian University of Science and Technology 13
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
2. Combining coding with hydraulic modelling
•Parallelizing CPU and/or GPU computing
Crespo et al. (2011)
Flow diagram showing the differences on the CPU and GPU implementation

Norwegian University of Science and Technology 14
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
3. Implementing Artificial Intelligence in combination with:
•Field data collection
•Hydraulic simulations
•Image processing
•Internet of Things (IoT)

Norwegian University of Science and Technology 15
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
3. Implementing Artificial Intelligence in combination with:
•Field data collection
Khan et al. (2019)
Hourly field data collection for multi-modal sensing
Block diagram of multi-modal sensing
Field data
collection Data labeling Neural
Network
Criteria
classification
Estimation
GUI/Master
station
Recommendations
for risk management

Norwegian University of Science and Technology 16
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
3. Implementing Artificial Intelligence in combination with:
•Hydraulic simulations
Berkhahn et al. (2019) Courtesy by S.W. Son (2020)
Flowchart scheme of a forecast model with an ensemble neural network ANN surrogate model of the physical based model for
the initial stage of a flash flood

Norwegian University of Science and Technology 17
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
3. Implementing Artificial Intelligence in combination with:
•Image processing
Akshya & Priyadarsini (2019) Histogram of Visual Word
Samples of aerial images of a city
Aerial images of flooded areas Aerial images of non-flooded areas

Norwegian University of Science and Technology 18
adina.moraru@ntnu.no
How to optimize hydraulic simulations?
3. Implementing Artificial Intelligence in combination with:
•Internet of Things (IoT)
Kartika et al. (2019)
System’s block diagram
IoT Platform
Machine
Learning User
Excel
Prediction result on Android application

Norwegian University of Science and Technology 19
adina.moraru@ntnu.no
What to consider when selecting an optimization
technique for ungauged steep rivers?
The following factors play an important role in the uncertainty of hydraulic models:
•Small rivers are more sensitive to extreme floods than large rivers
•Small ungauged rivers are not exhaustively studied and high-resolution topographic
and historical hydrologic data are seldom available
•Norwegian topography is complex (very steep gradients, rivers with tailwaters in
fjords) and data coming from nearby gauging stations, if existing, is hard to extrapolate
•Thorough sediment transport-related data is not always available, even though this
affects flow velocity considerably

Norwegian University of Science and Technology 20
adina.moraru@ntnu.no
What requirements should the optimization
technique meet?
•Focusing the effort on critical locations
and
•A) Optimizing the code of the numerical
model
or
•B) Simulating in parallel (High Performance
Computing platforms)
or
•C) Using Artificial Intelligence and transfer
learning
The technique
should achieve
60%-90% of the
floodplain coverage
presented by the
national emergency
authorities and be
cost-effective

Norwegian University of Science and Technology 21
adina.moraru@ntnu.no
What optimization technique works best in
ungauged steep rivers?
A) Optimizing the code of the numerical model
•Advantages: similar improvement in numerical efficiency could be achieved
through code development as through parallelization
•Disadvantages: the code’s areas might not run efficiently in parallel (e.g.
structures) or must run sequentially (e.g. calculations of Momentum and
Continuity)

Norwegian University of Science and Technology 22
adina.moraru@ntnu.no
What optimization technique works best in
ungauged steep rivers?
B) Simulating in parallel (High Performance Computing platforms)
•Advantages: highest optimization rates obtained out of the three techniques
•Disadvantages: unaffordable in small communities

Norwegian University of Science and Technology 23
adina.moraru@ntnu.no
What optimization technique works best in
ungauged steep rivers?
C) Using Artificial Intelligence and transfer learning

Norwegian University of Science and Technology 24
adina.moraru@ntnu.no
What optimization technique works best in
ungauged steep rivers?
C) Using Artificial Intelligence and transfer learning
•Advantages: multiple machine learning methods available able to emulate physical
models satisfactorily, potential use of transfer learning to extrapolate flood analysis
in a river with reliable data available to data-scarce rivers
•Disadvantages: uncertainty from the training and validation data (need of intensive
and data-reliable training)

Norwegian University of Science and Technology 25
adina.moraru@ntnu.no
So what?
•The original code of a numerical model is not always available for optimization
•Optimizing the original code of a numerical model should be considered carefully
and always bearing in mind the nature of the problem at hand and the limitations of
the optimization technique
•High Performance Computing platforms are very recommended if the logistics
permit it, yet they are not extensively used due to their high cost
•Artificial Intelligence (AI) has been widely used in hydrologic models, yet there are
fewer studies on hydraulic modelling and AI
•Transfer learning must be carefully used in ungauged steep rivers, as the complex
dynamics of such rivers difficult the extrapolation of the training to other cases

Norwegian University of Science and Technology 26
adina.moraru@ntnu.no
Thank you for your
attention!
Questions/suggestions?

Norwegian University of Science and Technology 27
adina.moraru@ntnu.no
References
•González-Cao, J., García-Feal, O., Fernández-Nóvoa, D., Domínguez-Alonso, J. M., & Gómez-Gesteira, M. (2019). Towards an Automatic Early Warning
System of Flood Hazards based on Precipitation Forecast: The case of the Miño River (NW Spain). Natural Hazards and Earth System Sciences
Discussions, 2(June), 1–25. https://doi.org/10.5194/nhess-2019-200
•Castellarin, A., Di Baldassarre, G., Bates, P. D., & Brath, A. (2009). Optimal Cross-Sectional Spacing in Preissmann Scheme 1D. Journal of Hydraulic
Engineering, 135(2), 96–105. https://doi.org/10.1061/ASCE0733-94292009135:296
•Moraru, A., Usman, K. R., Bruland, O., & Alfredsen, K. (2019). River idealization for identification of critical locations insteep rivers using 2D
hydrodynamic modelling and GIS. 22nd Northern Research Basins Workshop and Symposium, 144–153. https://doi.org/10.13140/RG.2.2.13276.64647
•Watanabe, A., Kojima, T., Mikami, T., Matsunobu, K., Suzuta, H., & Tomizawa, S. (2018). Interpolation of water surface profile in unsteady open channel
flow using the adjoint method based on 2D shallow water equations. In A. Paquier & N. Rivière (Eds.), River Flow 2018 - Ninth International Conference
on Fluvial Hydraulics (Vol. 40, p. 06030). E3S Web of Conferences. https://doi.org/10.1051/e3sconf/20184006030
•Crespo, A. C., Dominguez, J. M., Barreiro, A., Gómez-Gesteira, M., & Rogers, B. D. (2011). GPUs, a new tool of acceleration in CFD: Efficiency and
reliability on smoothed particle hydrodynamics methods. PLoS ONE, 6(6). https://doi.org/10.1371/journal.pone.0020685
•Khan, T. A., Alam, M., Shahid, Z., Ahmed, S. F., & Mazliham, M. (2019). Artificial Intelligence based Multi-modal sensing for flash flood investigation. 5th
International Conference on Engineering Technologies and Applied Sciences, ICETAS 2018, 1–6. https://doi.org/10.1109/ICETAS.2018.8629147
•Berkhahn, S., Fuchs, L., & Neuweiler, I. (2019). An ensemble neural network model for real-time prediction of urban floods. Journal of Hydrology, 575,
743–754. https://doi.org/10.1016/j.jhydrol.2019.05.066
•Akshya, J., & Priyadarsini, P. L. K. (2019). A hybrid machine learning approach for classifying aerial images of flood-hit areas. ICCIDS 2019 - 2nd
International Conference on Computational Intelligence in Data Science, Proceedings, 1–5. https://doi.org/10.1109/ICCIDS.2019.8862138
•Kartika, N. K. E., Murti, M. A., & Setianingsih, C. (2019). Floods prediction using radial basis function (RBF) based on internet of things (IoT). In IEEE
(Ed.), Proceedings - 2019 IEEE International Conference on Industry 4.0, Artificial Intelligence, and Communications Technology, IAICT 2019 (pp. 125–
128). https://doi.org/10.1109/ICIAICT.2019.8784839