Content uploaded by Chynna Sevilleno
Author content
All content in this area was uploaded by Chynna Sevilleno on Jan 29, 2019
Content may be subject to copyright.
A preview of the PDF is not available
Matina Flood Watch: A Predictive Modeling of Urban Flooding Implemented in a Mobile Application
Abstract
Globally, flooding has been predicted to increase in occurrences as sea levels proceed to rise due to global warming and climate change. The Philippines, situated near the Pacific Ring of Fire and the Pacific Ocean, receives numerous typhoons throughout the year, compelling the researchers to probe further into Disaster Risk Reduction mechanisms in Davao City. Matina Flood Watch was created to further improve flood forecasting in Davao City. In this study, the predictability of urban flooding in Matina using different machine learning algorithms was tested using data gathered from the Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA), flood records from Davao City Disaster Risk Reduction and Management Office (DCDRRMO), and real-time weather data from the APIs of OpenWeatherMap and Wunderground, which was then integrated into an android mobile application. The study compared different machine learning algorithms to determine which was the most suited for predicting flood levels and their corresponding areas. Random Tree achieved the best results with its model for flood water levels garnering an R2 score of 86.31% with a root mean squared error of 42.23% and its model for flooded Matina areas achieving an accuracy of 97.6% with a Matthew's Correlation Coefficient of 67.9%. The study has also demonstrated the need to develop DRRM mechanisms in Davao City and has successfully shown how the Matina Flood Watch application can aid disaster risk reduction efforts in the city.
ResearchGate has not been able to resolve any citations for this publication.
Executive Summary Satellite observations from 1993 to 2015 show that the tropical Western Pacific region, to the east of the Philippines, has experienced a sea level increase of 5-7 mm/yr, which is about twice the global average. The large increase is partly due to natural modes of ocean variability, such as El Nino Southern Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO), and partly due to an anthropogenic signal, which will continue regardless of the natural oscillations. Coastal tide-gauge records around the Philippines indicate a general pattern towards increased sea levels over the past 50 years. These tide gauges include also changes resulting from vertical motion of land and show different sea level trends for different locations around the Philippines (Fig. 1). There has been a particularly large increase in Manila (up to 1 metre), largely because of excessive groundwater extraction and land subsidence. Future sea level rise in the Philippines is projected to be slightly larger than global average sea level rise and this will increase the hazard posed by storm surges. The range of uncertainty in future sea level rise for the Philippines is also larger than the range of uncertainty in global sea level rise (Fig. 1). Due to the long term commitment to sea level rise, mean sea level is expected to increase throughout the 21 st century regardless of whether greenhouse gas emissions will be reduced. In the near-term, both the medium-low concentration greenhouse gas concentration scenario (RCP4.5) and the high-end scenario (RCP8.5) lead to similar increases in sea level. However, by the end of the century, the high end scenario adds approximately 20 cm to the projected mean sea level. The increase in the mean sea level do not differ substantially across the Philippines but in some coastal cities, such as Manila, land subsidence will make an additional contribution to sea level rise. An additional possible source of increase is a destabilisation of parts of the Antarctic ice sheet, which can add several tenths of a metre by 2100, but the probability of such an event and its extent are not known. The most severe impacts of sea level rise will likely manifest themselves through extreme events. Estimating the magnitude of these extreme events is beyond the scope of this study but higher mean sea levels will increase the likelihood of inundation for any given extreme event. c Crown Copyright 2016 1 Figure 1: Historical sea level change (SLC) from tide gauge observations and future projection. Plot shows the sum of SLC components discussed in this study for the global ocean (black line, with uncertainty range in grey) and for the coastal region of Legaspi City, Philippines (blue line, with uncertainty), under the RCP8.5 future scenario. Projections for the City of Legaspi, Philippines (3-5% higher than the global average). Annual mean tide-gauge measurements from Legaspi (blue line) and Cebu (green line) are also presented and a linear trend is added to the observations from Legaspi. The observed historical sea level trend from the tide gauge in Cebu is much smaller than at Legaspi, and in both cases a simple extrapolation of the trends clearly under estimate the projected values for the Philippines.
In the assessment of social impact caused by meteorological events, factors
of different natures need to be considered. Not only does hazard itself
determine the impact that a severe weather event has on society, but also
other features related to vulnerability and exposure.
The requests of data related to insurance claims received in meteorological
services proved to be a good indicator of the social impact that a weather
event causes, according to studies carried out by the Social Impact Research
Group, created within the framework of the MEDEX project. Taking these
requests as proxy data, diverse aspects connected to the impact of heavy rain
events have been studied.
The rainfall intensity, in conjunction with the population density, has
established itself as one of the key factors in social impact studies. One of
the conclusions we obtained is that various thresholds of rainfall should be
applied for areas of varying populations. In this study, the role of rainfall
intensity has been analysed for a highly populated urban area like Barcelona.
A period without significant population changes has been selected for the
study to minimise the effects linked to vulnerability and exposure
modifications. First, correlations between rainfall recorded in different
time intervals and requests were carried out. Afterwards, a method to include
the intensity factor in the social impact index was suggested based on return
periods given by intensity–duration–frequency (IDF) curves.
Flood risk assessment is generally studied using flood simulation models; however, flood risk managers often simplify the computational process; this is called a “simplification strategy”. This study investigates the appropriateness of the “simplification strategy” when used as a flood risk assessment tool for areas prone to flash flooding. The 2004 Boscastle, UK, flash flood was selected as a case study. Three different model structures were considered in this study, including: (1) a shock-capturing model, (2) a regular ADI-type flood model and (3) a diffusion wave model, i.e. a zero-inertia approach. The key findings from this paper strongly suggest that applying the “simplification strategy” is only appropriate for flood simulations with a mild slope and over relatively smooth terrains, whereas in areas susceptible to flash flooding (i.e. steep catchments), following this strategy can lead to significantly erroneous predictions of the main parameters—particularly the peak water levels and the inundation extent. For flood risk assessment of urban areas, where the emergence of flash flooding is possible, it is shown to be necessary to incorporate shock-capturing algorithms in the solution procedure, since these algorithms prevent the formation of spurious oscillations and provide a more realistic simulation of the flood levels.
The basic parameters of extra-tropical cyclones in the northern Baltic are examined in relation to extreme sea level events at Estonian coastal stations between 1948 and 2010. The hypothesis that extreme sea level events might be caused not by one intense extra-tropical cyclone, as suggested by earlier researchers, but by the temporal clustering of cyclones in a certain trajectory corridor, is tested. More detailed analysis of atmospheric conditions at the time of the two most extreme cases support this concept: the sequence of 5 cyclones building up the extreme sea level within about 10 days was very similar in structure and periodicity.
A flood is an extremely dangerous disaster that can wipe away an entire city, coastline, and rural area. The flood can cause wide destrotion to property and life that has the supreme corrosive force and can be highly damaging. In order to decrease the damages caused by the flood, an Artificial Neural Network (ANN) model has been established to predict flood in Sungai Isap, Kuantan, Pahang, Malaysia. This model is able to initiate the same brain thinking process and avoid the influence of the predict judgment. In this paper, presentation and comparison that using Bayesian Regularization (BR) back-propagation, Levenberg-Marquardt (LM) back-propagation and Gradient Descent (GD) back-propagation algorithms will be organized and carry out the result flood prediction. The predicted result of the Bayesian Regularization indicates a satisfactory performance. The conclusions also indicate that Bayesian Regularization is more versatile than Levenberg-Marquart and Gradient Descent with that can be backup or a practical tool for flood prediction. Temperature, precipitation, dew point, humidity, sea level pressure, visibility, wind, and river level data collected from January 2013 until May 2015 in the city of Sungai Isap, Kuantan is used for training, validation, and testing of the network model. The comparison is shown on the basis of mean square error (MSE) and regression (R). The prediction by training function Bayesian Regularization back-propagation found to be more suitable to predict flood model.
Flood water level prediction has become subject matter around the world because it can cause damaging threat to human life and property. Therefore an accurate flood water level prediction is vital in order to alert residents nearby flood location of incoming flood events. However, since flood water level fluctuates highly nonlinear, it is a very difficult task to predict flood water level accurately. Hence, as nonlinear model and well known as a very effective solution for handling nonlinear problems, ANN was chosen in this study. This paper proposed a 1 hour ahead flood water level prediction modeling using Multilayer Perceptron Neural Network. Results shows significant improvement from the original MLPNN model when the improved model is introduced.
