Available online at www.sciencedirect.

com

ScienceDirect
Tropical Cyclone Research and Review 10 (2021) 87e95
www.keaipublishing.com/tcrr

Flash flood modeling in the data-poor basin: A case study in

Matina River Basin
Rhonalyn V. Macalalad a,c, Shichao Xu b, Roy A. Badilla a, Socrates F. Paat a, Bema C. Tajones a,
Yangbo Chen b,*, Gerry Bagtasa c
a
Philippine Atmospheric, Geophysical and Astronomical Services Administration, Department of Science and Technology, Quezon City, Philippines
b
School of Geography and Planning, Sun Yat-Sen University, Guangzhou, China
c
Institute of Environmental Science and Meteorology, University of the Philippines, Quezon City, Philippines
Available online 22 June 2021

Abstract

Forecasting flooding hazards is a very effective non-engineering measure for flood control. Presently, the commonly used forecasting method
in simulating flash flood events is through a watershed hydrological model. Many Asia-Pacific countries like the Philippines are prone to
frequent hydrometeorological hazards such as tropical cyclones, resulting in frequent heavy rainfall events. However, most rivers in the many
basins lack water flow observation data, which makes it challenging to use lumped and data-driven models for flash flood forecasting. With the
continuous progress of remote sensing (RS) and geographic information system (GIS) techniques, the physically-based distributed hydrological
model (PBDHMs) has rapidly advanced. PBDHMs can directly determine the model parameters according to the underlying surface charac-
teristics from remotely-sensed data, which makes it possible for flood forecasting in areas with little to virtually no data. In this study, the Matina
River basin in Davao City was selected as a case study in simulating a small data-poor basin in the region. The Liuxihe model was used to
formulate a forecasting scheme and simulated the past flash flood events. The results show that there is a good correspondence between the past
heavy rainfall events and their corresponding simulated river discharges. The results conform to the hydrological regularities, which can be used
for flood forecasting and can serve as a baseline for the development of a flood forecasting system in the rivers of Davao City and elsewhere.
© 2021 The Shanghai Typhoon Institute of China Meteorological Administration. Publishing services by Elsevier B.V. on behalf of KeAi
Communication Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Typhoon; Flash flood; Liuxihe model

1. Introduction resulting in river flooding. Oftentimes, these heavy rainstorms

can cause rapid water level rise in streams causing flash floods
Typhoons and similar extreme hydrometeorological hazards especially in small-sized watersheds (Zhang et al., 2015).
are frequent in Asia-Pacific countries such as the Philippines, Because of their abrupt onset, flash floods can usually cause
resulting in heavy and frequent rainfall. These heavy and serious local damages to life and properties. In many areas
persistent rainfall events can cause rivers to rise and overflow where river flood control projects are limited, such hazards as
flooding may lead to huge disaster losses and have a serious
* Corresponding author. impact on the regional social and economic development.
E-mail address: [email protected] (Y. Chen). Flood forecasting is the most important non-engineering
Peer review under responsibility of Shanghai Typhoon Institute of China measurement for flood prevention and control (Xie et al.,
Meteorological Administration.
2018). Accurate flood forecasting can provide effective flood
warning information to people in flood-affected areas. A
population can take preventive countermeasures, espe-
Production and hosting by Elsevier on behalf of KeAi cially when large flooding events are imminent, and can be

https://doi.org/10.1016/j.tcrr.2021.06.003
2225-6032/© 2021 The Shanghai Typhoon Institute of China Meteorological Administration. Publishing services by Elsevier B.V. on behalf of KeAi Communi-
cation Co. Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

evacuated to mitigate disasters. At present, the commonly used along the periphery of a highly urbanized city (Davao City)
forecasting method is to simulate and forecast flash floods by (CLUP, 2013) and is vulnerable to urban sprawl. Finally, (4)
using watershed hydrological models. Among them, a climate change projection shows more intense flooding is
commonly used watershed hydrological models are the expected in the future (Cabrera and Lee 2020). Many small
lumped-models (Crawford and Linsley, 1966; Burnash, 1995; towns in the Philippines have these similar characteristics. The
Sugawara, 1995; Zhao, 1977; Todini, 1996) which utilize increasing occurrences of river flooding in the MRB suggest
parameter-based approaches to quantify non-spatially varying the need for a flood forecasting system in a region with such
representation of hydrological processes. Also in recent de- data-poor or ungauged basins. From remotely sensed data, we
cades, studies have applied machine learning or artificial set up the Liuxihe model to formulate a forecasting scheme
neural network models to flood forecasting of river basins due and simulate the measured flood process. We present the
to their ability to deal with the nonlinearity problem intrinsic determination of acceptable model parameters and evaluate
between rainfall and runoff in hydrologic systems (Zhou et al., their performance in simulating flood events in the basin.
2019; Chang et al., 2020). The Physically-based distributed
hydrological models or PBDHMs are based on the physics of 2. Davao City and flooding
the hydrological processes and use physically-based equations
to describe these processes. Together with the continuous 2.1. Description of modeling site
progress of remote sensing and geographic information system
(GIS) techniques, further development and application of Davao City is a highly urbanized city located in the
PBDHMs have become possible (Chen, 2009). PBDHMs can southeastern part of Mindanao Island in the southern
fully consider the heterogeneity of the underlying surface of a Philippines (Dumayas, 2015; CLUP, 2013). It lies between 60
basin and the spatial distribution of rainfall in a basin by 580 to 70 340 N latitude and 1250 140 to 1250 400 E longitude. It
dividing the basin into refined units, thereby improving the has a total land area of 2443.61 km2 making it the largest city
accuracy of flood forecasting in the basin; thus, these (by land area) in the Philippines. Davao City is the third most
PBDHMs have been called the new generation of basin flood populous city in the country with a population of 1,632,991
forecasting models. Representative PBDHM models include with 670 per km2 density (Census of Population, 2015) or
the SHE model (Abbott et al., 1986a; 1986b), the VIC model 1.57% of the total population of the Philippines (CLUP, 2013).
(Liang et al., 1994), the Vflo model (Vieux and Vieux, 2002), The population of Davao city is increasing at a rate of 2.36% a
the CASC2D model (Julien et al., 1995), the WetSpa model year since the year 2000. As this growth rate continues, it is
(Wang et al., 1997) and the Liuxihe model (Chen et al., 2010), expected that its population will double in 32 years. The city's
among others. location is favorable for regional and international trade.
A common difficulty in flood forecasting in the Asia- Davao city serves as the main trade, commerce, and industry
Pacific countries is the lack of flow observation data for hub of the southern Philippines. It also has the largest econ-
most rivers within small basins. For this reason, lumped- omy in the southern Philippine region with an annual growth
models are at a disadvantage as they require a certain num- rate of 2.53% over a 15-year period (CLUP, 2013). Agricul-
ber of flood processes to calibrate the model parameters. Data- ture, industries such as manufacturing, food processing and
driven models (i.e. machine learning) also need a lot of data metalworks, and tourism are some of the main contributors to
for training to achieve better results. In the case of PBDHMs, its economy (Dumayas, 2015).
it can directly determine the model parameters according to Fig. 1a shows the location and topography of Davao City.
the characteristics of the underlying surface, which makes it The site is mostly mountainous to the west characterized by a
possible to forecast the flood in the area with little to virtually mountain range that extends to the southwest. This demarcates
no data. Previous studies show that existing observation data, the western boundary of the city. In the mountain range sits the
even if limited, can be used to optimize the model parameters highest peak in the Philippines called Mt. Apo (3, 144 m
to improve the simulation accuracy. However, even without above sea level), a semi-active volcano located at the south-
any data, it is still possible to determine the model parameters, western tip of the city. To the east is a large adjoining lowland
albeit larger modeling uncertainties should be paid attention to area characterized by coastal plains and gradually sloping
(Chen et al., 2010). valleys that extends to the Davao gulf. To the south are also
In this study, we aim to apply a PBDHM called the Liuxihe lowland areas extending towards the gulf. These areas are land
model (Chen et al., 2010) to a small river basin in the southern with 0e3 percent slopes. The entirety of Davao City's land is
Philippine region called the Matina river basin. The Matina drained towards the Gulf of Davao River and its tributaries
river basin, herein referred to as MRB, is a small watershed (CLUP, 2013).
located in Davao City which experiences frequent flooding Davao City has a total of eight (8) separate watersheds
and even flash floods (Cabrera and Lee 2018). The MRB was (Hearne et al., 2019) as shown in Fig. 1b. Among the water-
chosen as a representative basin because it has the following sheds, the Davao River Basin is the biggest. This river basin is
characteristics: (1) the basin is small and not much attention is one of the eighteen (18) major river basins in the country and
given to it. Hence, the river basin is ungauged. (2) Due to its is considered the 3rd largest river in the southern portion of
climate and topography, it is prone to flooding. Moreover, Mindanao Island. The main Davao river has a length of
flood event frequency is observed to be increasing. (3) It is 150 km. Davao River and its tributaries are the main drainage
88
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

Fig. 1. Davao City's geographical a) location, topography, and b) watersheds.

system of the entire Davao City. It drains an area of over mean total annual rainfall of 1759.1 mm. Fig. 2a shows a
1,700 km2. Other smaller watersheds located within the city significant linear increasing trend of 13.8 mm/year in the
are the Talomo, Matina, Lipadas, Lasang, Sibulan, Tuganay, annual rainfall of Davao city from 1990 to 2020. The average
and Bunawan watersheds. Among these basins, Talomo, monthly mean temperature is above 28  C. The minimum
Lipadas, and Matina rivers are fully enclosed within the city temperature recorded was 22.4  C in February 1990 while the
territory. maximum temperature was observed in April 2016 with a
This study focuses on the Matina River Basin, which is one value of 35  C. Fig. 2b shows the annual mean maximum and
of the eight major river basins in Davao City having an area of minimum temperature in the city. In the last three decades,
78.79 square kilometers and accounts for 3.22% of the area of there are no significant trends in the observed maximum and
Davao City. This basin is located to the southwest of the main minimum temperatures.
city of Davao and is affected by the urban sprawl from the city Tropical cyclones induce many extreme weather events in
to its periphery. The main river of the Matina basin flows from the Philippines that lead to flooding (Macalalad et al., 2021).
Biao Escuela and traverses through barangays (villages) of However, rainfall contribution from tropical cyclones is min-
Matina, Pangi, and Matina Crossing before draining in the imal in the southern Philippines where Davao city is located
estuary located at the Matina Aplaya village in the gulf (IDIS, (Bagtasa 2017). In spite of this, Davao still experiences
2016). The Matina river basin, hereafter referred to as MRB, is frequent flooding events. In fact, 30% of the population in the
mainly made up of primary forests and large secondary forests region is at a high risk of pluvial flooding (Cabrera and Lee,
on mountainous terrain. Its relatively uniform temperature and 2020). Floodings in the region are usually caused by exces-
rainfall all year round also make it an ideal location for sive run-off from rivers due to intense rainfall from convective
agriculture. The highest elevation in the MRB is 258 m above clouds embedded in the inter-tropical convergence zone
sea level and its location downstream of the city makes it (ITCZ), Low-Pressure Area (LPA) from passing cyclonic
prone to flooding. flows, converging winds from strong winter monsoon surges,
strong easterly trade winds, as well as rainfall from the outer
2.2. Climate and flooding events rainbands of passing tropical cyclones (Bagtasa 2020). The
low-lying downstream areas prone to flooding are located
The climate of Davao city is classified as climate type IV in along the riverbanks where a large number of informal set-
the Modified Coronas' Climate classification, a climate sub- tlements reside, exacerbating the potential hazards of flooding.
typing classification based on monthly rainfall variability used Flooding in the city usually affects 72 barangays (villages)
in the Philippines (PAGASA, 2014, http://bagong.pagasa.dost. mostly close to the central business district. According to the
gov.ph/information/climate-philippines). Climate type IV is City Disaster and Risk Reduction Management Office
characterized by a climate that experiences no dry season as (CDRRMO), a total of 13.77 km2 of the 138.74 km2 built-up
the rainfall is evenly distributed throughout the year with areas in Davao are considered highly susceptible to flooding.
monthly rainfall ranging from 108 mm to 186.7 mm and a With this, 146,930 households are at risk predominantly from

89
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

Fig. 2. Annual-mean precipitation, maximum and minimum temperature from 1990 to 2020.

the villages of Bucana, Ma-a, Matina Pangi, Tigatto, Matina last decade. It is hypothesized that this increase is due to two
Crossing, and Matina Aplaya. reasons: (1) the southwest-ward migration of the western
From 2000 to 2020, there were one hundred ninety-four North Pacific subtropical high that is causing tropical cyclones
(194) flood occurrences recorded in the city by the to move in a more southward track (Bagtasa 2017), and (2) the
CDRRMO. These events affected a total of 39,316 fam- warping of the Madden-Julian Oscillation in the Indo-Pacific
ilies, 2492 structures, 46 deaths, and accumulated damages warm pool region (Roxy et al., 2019), which in turn in-
amounting to approximately $1.8 million dollars (CLUP, duces heavier rainfall in the southern Philippine region
2013). The more notable flood events occurred on June 28, (Bagtasa 2020).
2011 (flash flood) due to the effects of the active convection
associated with the ITCZ. The flooding event on January 20, 3. Methodology
2013 (flash flood) caused by continuous excessive rainfall
from the effect of a monsoon surge extending to Mindanao, 3.1. The liuxihe hydrologic model
and on December 22, 2017 due to rainfall carried by Typhoon
Vinta (CLUP, 2013). Fig. 3 below shows the annual summary In the present study, we used the Liuxihe model for the
of flood incidents that occurred in the city from 2000 to 2020. modeling of Matina watershed flood simulations. The Liuxihe
There is an apparent increase in flood event frequency in the model is a PBDHM developed by Chen et al. (2010) of the
Sun Yat-sen University, China. The model segregates the basin
into grid cells, after which, it calculates the evapotranspira-
tion and production flow on the cell scale, it then converges
cell-by-cell to the outlet of the basin through the conflu-
ence network. The model confluence is divided into slope
confluences and river confluences using the kinematic wave
approximation and diffusive wave approximation, respec-
tively, for their calculations. At the same time, the Liuxihe
model proposes an automatic optimization method for model
parameters based on the Particle Swarm Optimization (PSO)
algorithm (Eberhart and Shi, 2001). In practical applications,
only a representative measured flood is needed to optimize the
model parameters, which can greatly improve the performance
of the model. The use of the refined confluence calculation
Fig. 3. Flood events in Davao City, 2000e2020. methods and the high-efficiency parameter optimization
90
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

technology have enabled the Liuxihe model to achieve good up). In addition, there are three types of soils including Gleyic
results in forecasting floods in small and medium-sized rivers Cambisols (CN4478), Eutric Nitosols (CN4546), and Ochric
and in reservoir inflow forecasting in China (Xu et al., 2012; Andosols (CN4548).
Chen et al., 2016, 2017; Li et al., 2017, 2019, Chen et al.,
2017a,b, Wang and Chen, 2019). 3.3. Rainfall data

3.2. Land surface data Fig. 6 shows the location of the 4 rainfall stations around
the MRB. These stations are located at Pangi (PG), Mintal
The data comprising the physical characteristics of a basin (MT), Calinan (CN), and PshsSmc (PS). These are automatic
that are required for the construction of the Liuxihe model rain gauges installed and maintained by the Department of
include the DEM(Digital Elevation Model), land use/cover Science and Technology (the Philippines) e Advanced Sci-
map, and soil map. Among the various data, the DEM data is ence and Technology Institute. The Thiessen polygon method
the basis of the Liuxihe model structure. The DEM data was used to interpolate the hourly-rainfall data of the MRB,
quality directly affects the slope, basin area, river length, and and the corresponding interpolation weights are set at 0.475,
other significant geographic parameters of the basin. Hence, 0.424, 0.097, 0.004 for the PG, MT, CN, and PS rain gauge
DEM is the most important data used in model construction. stations, respectively.
This study selected the Shuttle Radar Topography Mission-3
(SRTM3) DEM data from NASA (National Aeronautics and 4. Model implementation and discussion
Space Administration) (https://www.nasa.gov/) and are map-
ped by the US Space Shuttle Endeavour Interferometric Radar In this study, we simulate two flooding scenarios in the
with a spatial resolution of 90 m. Fig. 4 shows the SRTM3 MRB. We set up the Liuxihe hydro model using the SRTM3
DEM data for the MRB. From this data, the basin has a size DEM, land use, soil, and rainfall data presented in the previous
of 65.49 km2, a minimum, maximum, and mean elevation section. We utilized the following derived values as initial
of 4 m, 258 m, and 131.86 m, respectively, and a minimum, parameters to run the Liuxihe Model. These parameters
maximum, and mean slope of 00, 24.620, and 3.360, represent the physical characteristics of each unit of the hy-
respectively. drologic model. First, flow direction was determined using the
The land use/cover type data are from the USGS(U.S. method of Jenson and Domingue (1988) where the DEM was
Geological Survey) land use type database (http://landcover. initially conditioned to remove “depressions” that hinder flow
usgs.gov/), and the soil type data are from the FAO(Food in the model. The flow path is then calculated by determining
and Agriculture Organization) soil type database (http://www. the steady decrease in DEM elevation values for the hori-
isric.org). The downloaded land use and soil type data were at zontally, vertically, or diagonally adjacent cells of each grid.
a resolution of 1000 m  1000 m and, therefore, were The channel and slope units were divided with a cumulative
resampled to a resolution of 90 m to match the DEM data. The flow threshold set at 600. The river was then extracted using
land use and soil types of the SRTM 3 Liuxihe model in the the Strahler method (Strahler, 1957) based on remotely sensed
MRB are shown in Fig. 5. From the data, there are three types images obtained from Google Earth. The rivers consist of
of land use in the basin, including cultivated and managed order 2 channel segments in 5 river nodes. Then, the river
areas, Mosaic areas (Cropland/Tree Cover/Other natural channels were divided into 9 virtual river sections. The cross-
vegetation), and artificial surfaces and associated areas (built- sectional width, side slope, and bottom slope of each virtual

Fig. 4. DEM of the MBR created with the SRTM3 database.

91
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

Fig. 6. The rainfall stations and corresponding rainfall interpolation of MRB.

river channel were then estimated. The result of the divisions

of unit classification, channel nodes, and virtual reaches is
shown in Fig. 7.
Second, the land use/land cover type parameters include the
slope roughness and evaporation coefficient summarized in
Table 1. Prior test simulations show that the model is not
sensitive to changes in the evaporation coefficient. Here, we
set the evaporation coefficients uniformly at 0.7. The slope
roughness adopts the recommended values used in previous
literature (Wang et al., 1997; O'Callaghan and Mark, 1984).
Third, Table 2 shows the initial values of soil parameters
which include the saturated water content, field capacity,
saturated hydraulic conductivity, soil thickness, wilting point,
and soil characteristics. The soil property is uniformly set at
2.5, and the remaining parameters are calculated by the soil
hydraulic characteristic calculator of Arya and Paris (1981).
Finally, on the meteorological parameters, rainfall is spatially
disaggregated using the Thiessen polygon method as described
in section 3.3. In addition, based on prior model testings, the
evaporation capacity of all units of the basin is assumed and
set to 5 mm/day.

4.1. Flood simulations

In this study, two flood events in the MRB were selected for
simulation, a severe thunderstorm event on May 3, 2017, and a
Fig. 5. Terrain property data of MRB from the SRTM 3 database. (a) Land-use persistent rainfall event on August 23, 2018. These events will
types; (b) soil types.

92
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

Table 3
Eigenvalues of simulation results. (Rainfall values are Basin-wide mean).
Flood event 20170503 20180823
Rainfall peak (mm) 12.524 7.505
(Basin-wide mean)
Average rainfall (mm) 1.73 1.574
Quantum of rainfall of PG station (mm) 38.76 43.225
Quantum of rainfall of MT station (mm) 16.366 0.509
Quantum of rainfall of CN station (mm) 7.081 0.3298
Quantum of rainfall of PS station (mm) 0.395 0
Flood peak (m3/s) 21.246 25.061
Runoff coefficient 0.433 0.540
Rainfall and flood peak time difference (min) ~80 ~50

of rain fell in the Davao region based on the GPM(Global

Precipitation Measurement)-derived rainfall (https://www.
nasa.gov/mission_pages/GPM/main/index.html; Kirschbaum
et al., 2017). On the other hand, the NO.20180823 event, a
tropical depression (TD “Luis”) located in the southwest of
Taiwan, China led to the deepening of the Asian monsoon
trough, which induced the strong southwesterly monsoon
prevailing during the event. At the same time, convective
thunderstorms formed in Davao city amidst the weak synoptic
wind environment in the area. The estimated rain in the Davao
region from the GPM rainfall data is about 20e30 mm.
The results of the river discharge simulation of the
MRB during the two events are shown in Table 3. The
Fig. 7. Liuxihe model structures of MRB with SRTM3 data sources. observed basin-wide peak rainfall and average rainfall for the
NO.20170503 event is 12.52 mm/h and 1.73 mm/h, respec-
Table 1 tively, and for the NO.20180823 flood event is 7.51 mm/h and
The initial values of land use/cover-related parameters. 1.574 mm/h, respectively. For the NO.20170503 flood event,
Land use/cover Evaporation Roughness the resulting peak river discharge is 21.246 m3/s and a runoff
coefficient coefficient coefficient of 0.433. On the other hand, for the NO.20180823
Cultivated and managed areas 0.7 0.15 event, peak river discharge has a slightly higher value of
Mosaic: Cropland/Tree Cover/Other 0.7 0.35 25.061 m3/s and a runoff coefficient of 0.54. In both
natural vegetation GPM-derived and interpolated observed rainfall amounts, the
Artificial surfaces and associated areas 0.7 0.1
NO.20180823 event had an overall smaller rainfall depth
compared to the NO.20170503 flood event. However, the
second event had higher resulting river discharge. This
be referred to as NO.20170503 and NO.20180823, respec- apparent inconsistency is explained by the spatial differences
tively. The NO.20170503 event occurred in May during the between the observed station rainfall data. Fig. 7 shows the
onset of the southwest monsoon season. A series of convective hydrograph of the two events simulated. For the NO.20170503
thunderstorms forming along the western regions of the event in Fig. 7a, it can be observed that rainfall was located
country extending from the northernmost island of Luzon to more upstream as upstream stations recorded a significant
the southernmost island of Mindanao. Using the JRA55 portion of rainfall in the basin. In the NO.20180823 event
(Japanese 55-year atmospheric reanalysis) reanalysis wind shown in Fig. 7b, most of the observed rainfall was recorded
data at 850 mb (Kobayashi et al., 2015), easterly trade winds in the PG station, indicating more localized precipitation
were prevailing over most of the Philippines with a low-level along the main river of the MRB. Despite the overall lower
cyclonic flow to the northwest. It is estimated that 50e60 mm rainfall amount, the difference in the spatial distribution of

Table 2
The initial values of soil-related parameters.
Type Thickness of Saturated water Field Saturated hydraulic Soil characteristic Wilting
soil layer (mm) content capacity conductivity of soil (mm$h1) coefficient point
CN4478 1000 0.458 0.278 15.66 2.5 0.137
CN4546 1000 0.522 0.442 1.22 2.5 0.328
CN4548 1000 0.469 0.298 13.96 2.5 0.137

93
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

rainfall led to higher river discharge in the case of the References

NO.20180823 event.
Though there is a lack of automated river height or Abbott, M.B., Bathurst, J.C., Cunge, J.A., O'Connell, P.E., Rasmussen, J.,
discharge measurements in the MRB, the simulation result is 1986a. An introduction to the European hydrologic system-system
hydrologue europeen, ‘SHE’, a: history and philosophy of a physically-
consistent with the flood incident report conducted by the based, distributed modelling system. J. Hydrol. 87, 45e59.
government during the evacuation of the affected population. Abbott, M.B., Bathurst, J.C., Cunge, J.A., O'Connell, P.E., Rasmussen, J.,
In NO.20170503, government monitoring indicated an 1986b. An introduction to the European hydrologic system-system
impending flood event at 2107H local time. By 2155H, hydrologue europeen, ‘SHE’, b: structure of a physically based, distrib-
evacuation of affected residents commenced. Peak rainfall uted modeling system. J. Hydrol. 87, 61e77.
Arya, L.M., Paris, J.F., 1981. A physicoempirical model to predict the soil
occurred between 2200H and 2300H, followed by river water moisture characteristic from particle-size distribution and bulk density
level of 0.46 m at 2330H and a peak of 0.59 m water level at data. Soil Sci. Soc. Am. J. 45, 1023e1030.
0030H past midnight the following day. The report of the Bagtasa, G., 2017. Contribution of tropical cyclones to rainfall in the
authorities is consistent with the model simulation of the Philippines. Journal of Climate 30 (10), 3621e3633.
NO.20170503 event where peak river discharge onset time Bagtasa, G., 2020. Influence of maddenejulian oscillation on the intraseasonal
variability of summer and winter monsoon rainfall in the Philippines.
from peak rainfall is approximately 80 min. In NO.20180823, Journal of Climate 33 (22), 9581e9594.
the peak water level reached 1.03 m and had an onset time Burnash, R.J.C., 1995. In: Singh, V.P. (Ed.), The NWS River Forecast System-
from peak rainfall to peak discharge of about 50 min, also Catchment Modeling, Computer Models of Watershed Hydrology. Water
consistent with the ground reports. Resource Publications, Littleton, CO, USA.
Cabrera, J.S., Lee, H.S., 2018. Impacts of climate change on flood-prone areas
in Davao Oriental, Philippines. Water 10 (7), 893.
5. Conclusion Cabrera, J.S., Lee, H.S., 2020. Flood risk assessment for Davao Oriental in the
Philippines using geographic information system-based multi-criteria
The flood forecasting model of the Matina river basin analysis and the maximum entropy model. Journal of Flood Risk Man-
(MRB) is established by using the Liuxihe model. The MRB agement 13 (2), e12607.
is a small and ungauged river basin located in the southern Census of Population, 2015. Region XI (Davao Region). Total Population by
Province, City, Municipality and Barangay. PSA. (Accessed 18 February
Philippines. It is an area that is typical in the Philippines - 2021).
prone to flooding, vulnerable to urban sprawl, and increased Chang, L.C., Chang, F.J., Yang, S.N., Tsai, F.H., Chang, T.H., Herricks, E.E.,
risk to climate change. We present the determination of 2020. Self-organizing maps of typhoon tracks allow for flood forecasts up
model parameters using the PSO algorithm to optimize the to two days in advance. Nat. Commun. 11, 1.
hydrological model. Results show that the chosen parameters Chen, Y., 2009. Liuxihe Model. Science Press, Beijing, China, 198p.
Chen, Y.B., Ren, Q.W., Huang, F.H., Xu, H.J., Cluckie, L., 2010. Liuxihe
are robust enough with acceptable model simulation accu- model and its modeling to river basin flood. J. Hydrol. Eng. 16, 33e50.
racy. The simulation results conform to the limited obser- Chen, Y., Li, J., Xu, H., 2016. Improving flood forecasting capability of
vations during the selected flood events, where the runoff physically based distributed hydrological model by parameter optimiza-
coefficients of the two simulated flood events vary between tion. Hydrol. Earth Syst. Sci. 20, 375e392.
0.3 and 0.6. Overall, despite the limitation in the data used, Chen, Y., Li, J., Wang, H., Qin, J., Dong, L., 2017a. Large-watershed flood
forecasting with high-resolution distributed hydrological model. Hydrol.
the model showed reasonable results which makes the model Earth Syst. Sci. 21, 735e749.
sensible, thus can provide early warning references for flood Chen, Y., An, X., Dou, P., Zhang, T., 2017b. Estimating land use/cover of
management. The study also suggests that even without foshan city in southern China with landsat remote sensing imagery for
automated gauges, simple citizen observations can validate flood modeling. In: Proceedings of the International Multidisciplinary
and help in further developing the model parameters used in Scientific GeoConference Surveying Geology and Mining Ecology Man-
agement, vol. 17, pp. 79e86. SGEM, Albena, Bulgaria.
the MRB. This study presents a new numerical tool that can CLUP, 2013. Davao City, Comprehensive Land Use Plan for 2013e2022,
be a basis in developing a flood forecasting system among (Philippines). https://cpdo.davaocity.gov.ph/wp-content/uploads/2016/10/
the rivers of Davao City and similar locations in the Vol1CLUP2013-202220151215084007.pdf. (Accessed 18 February 2021).
Philippines. Crawford, N.H., Linsley, R.K., 1966. Digital Simulation in Hydrology, Stan-
ford Watershed Model IV. Technical Report No. 39. Department of Civil
Engineering, Stanford University, Stanford, CA, USA.
Declaration of competing interest Dumayas, A.D.R., 2015. City development in emerging economies. The case
of Davao city in the Philippines. Firms’ Location Selections and Regional
The authors declare no conflict of interest. The funders had Policy in the Global Economy 267e280. https://doi.org/10.1007/978-4-
no role in the design of the study; in the collection, analysis, or 431-55366-3_15.
interpretation of data; in the writing of the manuscript; or in Eberhart, R.C., Shi, Y.H., 2001. Particle swarm optimization: developments,
applications and resources. Proceedings of the 2001 Congress on Evolu-
the decision to publish the results. tionary Computation. IEEE, Seoul, Korea, pp. 81e86.
Hearne, D., Gamboa, R., Monsanto, V.M., 2019. From ideas to action: a re-
view of implementing HELP principles in river basins with limited re-
Acknowledgments
sources and capacity. Water SA 34 (4), 504. https://doi.org/10.4314/
wsa.v34i4.183663.
This study was supported by the Natural Science Founda- Interface Development Interventions, 2016. (IDIS) Inc. Clean Water. Healthy
tion of China (No. 51961125206). Watersheds.

94
R.V. Macalalad, S. Xu, R.A. Badilla et al. Tropical Cyclone Research and Review 10 (2021) 87e95

Jenson, S.K., Domingue, J.O., 1988. Extracting topographic structure from Roxy, M.K., Dasgupta, P., McPhaden, M.J., Suematsu, T., Zhang, C., Kim, D.,
digital elevation data for geographic information system analysis. Photo- 2019. Twofold expansion of the Indo-Pacific warm pool warps the MJO
gramm. Eng. Remote Sens. 54, 1593e1600. life cycle. Nature 575 (7784), 647e651.
Julien, P.Y., Saghafian, B., Ogden, F.L., 1995. Raster-based hydrologic Strahler, A.N., 1957. Quantitative analysis of watershed geomorphology.
modeling of spatially-varied surface runoff. Water Resour. Bull. 31, Trans. Am. Geophys. Union 35, 913e920.
523e536. Sugawara, M., 1995. In: Singh, V.P. (Ed.), “Tank model.” Computer Models
Kirschbaum, D.B., Huffman, G.J., Adler, R.F., Braun, S., Garrett, K., of Watershed Hydrology. Water Resources Publications, Littleton, CO,
Jones, E., Zaitchik, B.F., 2017. NASA's remotely sensed precipitation: a USA.
reservoir for applications users. Bulletin of the American Meteorological Todini, E., 1996. The ARNO rainfall-runoff model. J. Hydrol. 175,
Society 98 (6), 1169e1184. 339e382.
Kobayashi, S., Ota, Y., Harada, Y., Ebita, A., Moriya, M., Onoda, H.,., Vieux, B.E., Vieux, J.E., 2002. VfloTM: a real-time distributed hydrologic
Takahashi, K., 2015. The JRA-55 reanalysis: general specifications and model. In: Proceedings of the 2nd Federal Interagency Hydrologic
basic characteristics. Journal of the Meteorological Society of Japan. Ser. Modeling Conference, Las Vegas, NV, USA, 28 Julye1 August.
II 93 (1), 5e48. Wang, H.W., Chen, Y.B., 2019. Identifying Key hydrological processes in
Li, J., Chen, Y.B., Wang, H.Y., Qin, J.M., Li, J., Chiao, S., 2017. Extending highly urbanized watersheds for flood forecasting with a distributed. Hy-
flood forecasting lead time in a large watershed by coupling WRF QPF drological Model. Water. 11, 1641.
with a distributed hydrological model. Hydrol. Earth Syst. Sci. 21, Wang, Z., Batelaan, O., De Smedt, F., 1997. A distributed model for water and
1279e1294. energy transfer between soil, plants and atmosphere (WetSpa). Phys.
Li, J., Yuan, D.X., Liu, J., Jiang, Y.J., Chen, Y.B., Hsu, K.L., Sorooshian, S., Chem. Earth 21, 189e193.
2019. Predicting floods in a large karst river basin by coupling Xie, L.L., Liu, P., Guo, S.L., 2018. Optimal design of seasonal flood limited
PERSIANN-CCS QPEs with a physically based distributed hydrological water levels by jointing operation of the reservoir and floodplains. Water
model. Hydrol. Earth Syst. Sci. 23, 1505e1532. Resour. Manag. 32, 179e193. https://doi.org/10.1007/s11269-017-1802-7.
Liang, X., Lettenmaier, D.P., Wood, E.F., Burges, S.J., 1994. A simple hy- Xu, H., Chen, Y., Zeng, B., He, J., Liao, Z., 2012. Application of SCE-UA
drologically based model of land surface water and energy fluxes for algorithm to parameter optimization of Liuxihe model. Singap. J. Trop.
general circulation models. J. Geophys. Res. 99, 14415e14428. Geogr. 32, 32e37.
Macalalad, R.V., Badilla, R.A., Cabrera, O.C., Bagtasa, G., 2021. Hydrolog- Zhang, D., Quan, J., Zhang, H., Wang, F., Wang, H., He, X., 2015. Flash flood
ical response of the pampanga river basin in the Philippines to intense hazard mapping: a pilot case study in Xiapu River Basin, China. Water
tropical cyclone rainfall. Journal of Hydrometeorology 22 (4), 781e794. Science and Engineering 8 (3), 195e204.
O'Callaghan, J.F., Mark, D.M., 1984. The extraction of drainage networks Zhao, R.J., 1977. Flood Forecasting Method for Humid Regions of China. East
from digital elevation data. Comput. Vis. Graph. Image Process. 27, China College of Hydraulic Engineering, Nanjing, China.
323e344. Zhou, Y.L., Guo, S.L., Chang, F.J., 2019. Explore an evolutionary recurrent
PAGASA, 2014. Climate Types. http://bagong.pagasa.dost.gov.ph/information/ ANFIS for modelling multi-step-ahead flood forecasts. J. Hydrol. 570,
climate-philippines. (Accessed 20 March 2021). 343e355.

95