A.M.F.Lagmay (a), G. Bagtasa (b), C.P. David (a), I. Crisologo (a), B.A. Racoma (a)
(a) National Institute of Geological Sciences, University of the Philippines C.P. Garcia corner Velasquez street, U.P. Diliman, Quezon City. 1101 Philippines
(b) Institute of Environmental Sciences and Meteorology, University of the Philippines, C.P. Garcia cor. Velasquez Street, U.P. Diliman, Q.C.
Heavy southwest monsoon rains from 18-21 August 2013 flooded Metro Manila and its suburbs, paralyzing the nation’s capital for an entire week. Called the 2013 Habagat event, it was a repeat of the 2012 Habagat or extreme southwest monsoon weather from 6-8 August, which delivered record rains in the mega city since the 1950s. In the 2012 and 2013 Habagat events, cyclones passing northeast of the Philippine archipelago enhanced the southwest monsoon rains and are primarily blamed for the heavy rains. Analysis of Doppler data, rainfall measurements, and WRF simulations show that two large stratovolcanoes, Natib and Mariveles, across from Manila Bay and approximately 70 km west of Metro Manila, played a substantial role in delivering extreme rains and consequent floods to Metro Manila. The study highlights how volcanoes, with its shape and height create an Orographic eﬀect and dispersive trail of rain clouds, which constitutes a significant flood hazard to large communities like Metro Manila with a dense population of more than 11 million people.
The 2013 Habagat event made a repeat of the previous year’s heavy rains in Metro Manila (Figure 1), comparable to the 2012 Habagat floods that swamped the country’s capital during last year’s summer monsoon season. Lacking names because extreme weather events are only given to tropical cyclones in the Philippines, the two consecutive years of extreme flooding have been referred to as the “2012 and 2013 Habagat Floods.”
Habagat is the local name for the summer southwest monsoon, a natural weather pattern that occurs during the months from June to September in the Asian subcontinent. The Habagat is characterized by warm and humid climate and occurs when warm moist air flows over the country from the southwest. It is a weather phenomenon also responsible for bringing significant rainfall that may last for a week to the Philippine archipelago’s western section (PAGASA, 2013).
A total of 1,067.4 mm of rainfall was recorded in Sangley Point, Cavite, during this year’s Habagat event. This was 125% or 2.25 times higher compared to average monthly rainfall of 475.4 mm in August (Figure 2), and 36% more rainfall compared to the 2012 Habagat event for this station. For reference, Hurricane Katrina in Louisiana, USA, dumped the highest precipitation total of 417 mm (Roth, 2008; Knabb et al., 2005) in Perrine, about halfway between Miami and Homestead, USA. Of the total in Perrine, 384 mm fell in 24 hours.
Cavite is in the suburbs, south of the metropolis (Figure 1B) which received more rain than in Metro Manila on the first day of the Habagat 2013. Rainfall was only 700.7 mm, 23% in the Port area of the city of Manila, less than the 2012 Habagat event, but still 62% more than the average rainfall recorded by this recording station for the month of August (Figure 2B). In Science Garden, where the head oﬃce of the Philippine weather bureau in Metro Manila is located, the rainfall total is 545.5 mm, 85% less than the 1007.4 mm of rainfall recorded for the 2012 Habagat event and 8% more than the average monthly rainfall of 504.2 mm in this area for August (Figure 2C).
According to the National Disaster Risk Reduction Council (NDRRMC, 2013), the Habagat event resulted in 27 fatalities, mostly from drowning in 12 provinces of northern Philippines. One died in the nation’s capital whereas property damage as of NDRRMC’s latest situation report was estimated at PhP688.1 million, including PhP 137.9 million in infrastructure and PhP 550.2 million in agriculture.
This study investigates the rainfall distribution in Greater Metro Manila and how two volcanoes, Natib and Mariveles, magnified and focused the rains of the southwest monsoon during the 2013 and 2012 Habagat events. Natib volcano is 1,253 meters high with a base diameter of 26 kilometers, while Mariveles volcano is 1,388 meters high with a base diameter of 22 kilometers. These two large and potentially active stratovolcanoes (Phivolcs, 2013) are located approximately 70 km west of Metro Manila and act as a barrier for the flow of warm moist air from the West Philippine Sea (also known as South China Sea) during the Habagat season (Figure 1). The air mass rises rapidly causing condensation to form rain clouds that bring extreme precipitation and floods to Metro Manila.
2. Geographical setting and watersheds of Metro Manila and suburbs
Metro Manila is located on an Isthmus between two bodies of water, the Manila Bay, which opens to the South China Sea and Laguna de Bay (Figure 1A), a 10 x 20 km wide freshwater lake partly formed by two calderagenic eruptions. The metropolis lies on one of the widest floodplains in the Philippines underlain by mostly volcanic and coastal sediments (Lagmay et al., 2010).
The entire region of Metro Manila is composed of one major river basin called the Marikina River Basin (535 km2) and eight smaller, highly urbanized river sub-basins (683 km2) that drain directly towards Manila Bay and Laguna de Bay (Figure 1B). The Marikina and Pasig Rivers serves as the main outlet for tributaries of the Marikina River Basin.
South of Metro Manila is Cavite Province, within the watershed on the northern slopes of Ancestral Taal Volcano. North of Metro Manila is Bulacan Province which in part is within the La Mesa watershed and in part the Pampanga watershed, a large 10,707.95 km2 river basin composed of 7 major tributaries that drain towards north Manila Bay (Figure 1A).
Doppler data for the area of Metro Manila and vicinities, rain gauge data, and satellite imagery were investigated to determine the contribution of the Natib and Mariveles volcanoes (Figure 1B) in enhancing the heavy southwest monsoon rains from 18-21 August 2013. Simulations using the WRF were used to determine the extent of the Orographic Eﬀect in generating heavy precipitation that inundated Metro Manila, adjacent suburbs and rural villages during the Habagat event of 2013.
The Doppler data were combined to reflect the daily total rainfall and compared with the isohyetal maps generated from rain gauge data. The Doppler data was also examined relative to the Automated Rain Gauge (ARG) measurements from 27 stations to assess relative bias in the radar measurements. Although the Doppler data indicated rain in the clouds (precipitable rain) while ARG data were measurements of rain in the ground, statistical analysis was conducted in order to demonstrate that the Doppler radar from Subic Station was providing reasonable data for analytical work. Relative Bias is the systematic overestimation or underestimation of the radar measurement from the ARG measurements. For the bias analysis, the formula is shown below:
RELATIVEBIAS = ARGMEASUREMENT−RADARMEASUREMENT (1)
If relative bias is negative, then it means that the radar is overestimating the amount of rainfall relative to the ARG. If the relative bias is positive,then it means that the radar is underestimating the amount of rainfall relative to the ARG. If the bias is zero, then other statistics may be necessary to get a clearer picture of the diﬀerences.
Satellite imagery was used to determine the location and trajectory of the tropical cyclone northeast of the Philippines. Called tropical storm Maring (international codename Trami), this cyclone boosted the southwest monsoon winds and rains. A similar scenario,with a tropical storm located northeast of the Philippines was also present during the 2012 Habagat and comparisons between the two Habagat events are provided in the discussion.
The Weather Research and Forecasting (WRF) model was used to simulate the eﬀect of Natib and Mariveles volcanoes on moist air mass from the West Philippines Sea, drawn to the east-northeast by a tropical cyclone northeast of the Philippines from 18-20 August 2013. The WRF model is a nonhydrostatic mesoscale numerical weather prediction model used for both operational forecasting and atmospheric research. It is based on an Eulerian solver with a terrain following pressure coordinate. To simulate the eﬀects of both Mt. Natib and Mt. Mariveles on precipitation, two (2) simulation cases were conducted. One was a control run where topography was left unchanged while the second was a simulation run where the two volcanoes were removed. The simulation used a cloud-resolving 1 km grid scale, with terrain heights from the 30-arc second USGS global topography data set (Figure 3).
4.1 Doppler data
During the Habagat event of 2013, Doppler radar was able capture precipitable rain clouds up to 250 km radius from its location, including Metro Manila and its vicinity. The images used came from the Doppler radar at Subic station, maintained and operated by the Philippine Atmospheric, Geophysical and Astronomical Science Administration (PAGASA). Blind spots in the southern sector and northern sector are due to blocked radar signals by mountain terrain (see figure 4). It is assumed that reliability of radar data decays with distance from the instrument station and the analysis is best within 100 km radius of the Subic Doppler radar.
Isohyetal maps from ARG data reveal heavy rainfall in Metro Manila and vicinities from 18-19 August 2013 (Figure 4A and B), waning in intensity as continuous rains entered the 3rd and 4th days on 21 and 22 August 2013 (Figure 4C and D). On the 18th of August, rainfall was heaviest in Cavite (Figure 4A) while on the 19th of August, was heaviest in Metro Manila (Figure 4B).
Rainfall distribution is consistent with the distribution of rain clouds seen in the the Doppler images. Plumelike concentrations of precipitable rain (whitish area in the Doppler images) are observed to emanate from the regions near the summit of the Natib and Mariveles volcanoes. Furthermore, less rain clouds are observed in the West Philippines Sea (also known as the South China Sea) compared to those seen on or near land in the Doppler image.
4.2 Relative Bias of Doppler vis-a-vis ARG data
The average of relative biases from the 27 stations per day from 19-22 August 2013 shows radar measurements overestimating rainfall relative to the ARG measurements. Bias measurements were generally negative in the four days of Habagat with radar measurements in many stations consistently overestimating ARG rainfall measurements. Furthermore, radar measurements tend to overestimate the ARG measurements as rainfall measurements get larger.
Trend lines in Figure 5A show the trend of bias in the data while Figure 5B shows the relationship between rainfall value and the degree of bias. The degree of bias is greatest at the end of the third day of the Habagat 2013. Also noticeable is that as rainfall values measured by the ARGs became larger, the Doppler radar readings tend to overestimate rainfall readings. It was noticed that on 14 out of 27 stations the Doppler radar readings were consistently overestimating the rainfall measures relative to the ARG results in all four days (Table 1). The 14 stations are highlighted in yellow in Table 1. These ARG stations are from Pampanga, Cavite, Tarlac, and Zambales. The five stations with the highest average bias per time have their statistic marked with red ink. The San Felipe Town Hall Station had the largest negative biases in August 20 and 21 rainfall measures. Two stations have large underestimations of rainfall readings; these came from Bataan Orani State University and Quezon City Science High School Station, with positive biases. Both stations had negative relative biases in the August 21 readings. These stations are highlighted in blue.
4.3 WRF Simulations
The model runs of the WRF for the Habagat 2013 event show heavy precipitation on the windward side of Mount Natib and Mariveles (Figure 6A) a feature not seen in the simulation scenario when the two volcanoes are stripped oﬀ (Figure 6B). In addition to heavy precipitation due to the orographic eﬀect in the windward side of Natib and Mariveles volcanoes, there is also more rainfall observed over Manila Bay and extended rain distribution in the southern portion of Metro Manila.
The radar images in figure 4, however, was not able to detect the precipitation on the windward side of the mountain ranges to the east and north-east of Metro Manila due to the 10◦ scan angle of inclination of the doppler radar to minimize ground clutter contribution from the same mountain ranges.
When modeled in WRF for water mixing ratio, both Mt. Natib and Mt. Mariveles produced a trail of rain clouds which brought higher precipitation over Manila Bay and the western portions of Metro Manila (Figure 7A). In these simulations, cloud trails emanate from Natib and Mariveles volcanoes extending towards Metro Manila. East of Metro Manila where there are mountains, cloud trails are enhanced (right middle portion of Figure 7A). In the model runs where Natib and Mariveles volcanoes are stripped oﬀ, there are no cloud trails over Manila Bay (Figure 7B)). Without Natib and Mariveles, the cloud trails only begin to form upon entering the mountainous region east of Metro Manila (right middle portion of Figure 7B). Mount Pinatubo, which was not removed in the topography for the WRF simulation runs (case 1 and case 2), produced west- to east-directed cloud trails.
Apart from the updraft in the windward side of Mariveles volcano, the simulated vertical wind speed cross section along Mt. Mariveles shows a wave breaking region with as much as 2.4MS−1 upward speed (see circled region in Figure 8). This updraft region resembles a lee wave, referred to as a “dispersive trail” of nonhydrostatic waves by Smith (1979). According to Cotton et al. (2011) the size of the volcanoes determine the maximum amplitude of the nonhydrostatic waves and can be observed at low levels on the leeward slopes of the mountain. This dispersive trail can be maintained for appreciable distance downwind. In the case of Mariveles and Natib volcanoes, the dispersive trail was maintained for more than 70 kilometers reaching the heavily populated Capital of the Philippines. Clouds formed from this orographic updraft are clearly seen in the radar images in Figure 4).
Recently acquired Doppler Radar stationed at Subic, Zambales has allowed the study of mesoscale weather events, such as the intense rains that appear to emanate from the Natib and Mariveles volcanoes. Previously, such rain cloud patterns that form during the southwest monsoon season could not be viewed and studied in detail. All that was known was tropical cyclones, especially those passing through the northeast side of the Philippine archipelago, enhanced the southwest monsoon and bring forth heavy rains to the country’s western section.
In the last two years of extreme Habagat rains and floods in Metro Manila and its suburbs, volcanoes have played an important role. Acting as barriers for the flow of warm moist air from the West Philippine Sea, Natib and Mariveles volcanoes cause rapid uplift of moist air mass and consequent condensation that bring forth focused and magnified precipitation to Metro Manila and its suburbs. This process must have taken place for many thousand years in the region but was only observed recently because of the availability of Doppler data.
Comparison between the Doppler data for accumulated rain clouds and precipitation measured in the ground from ARG stations show reasonable correlation. Doppler overestimates the measured value for the same location where there is an ARG, but is an error that is expected since the rain cloud may not have fallen yet on the ground. As ground rainfall values become larger, the overestimate of the Doppler radar value also becomes larger. However, there is still positive correlation, enough to demonstrate qualitatively that Doppler radar represents the distribution of rain clouds, in particular their major source regions, which come from Natib and Mariveles volcanoes.
Simulations using the WRF corroborate the observations in Doppler radar, showing that the two volcanoes are major sources of rain clouds during the Habagat 2013 event (case 1). Without the volcanoes (case 2), there are no major rain clouds formed. The same phenomenon was observed in Doppler radar imageries for the Habagat 2012 event.
Extreme rainfall with consequent floods is not listed as a hazard associated with volcanoes. Although the Orographic Eﬀect is known in volcanoes to generate more rainfall in the windward side and deplete moisture in the leeward side, enhanced rains and cloud trails with intense rainfall that travel for more than 70 km beyond the leeward slopes of the volcano is not documented. The case of the lethal 2012 and 2013 Habagat floods where Natib and Mariveles volcanoes enhanced precipitation is remarkable, delivering more 1000 mm of southwest monsoon rain over a highly populated metropolis. This phenomenon should be enough to be considered in the list of volcano-associated hazards.
In 1950, the core municipality of Manila had a population of under 1 million people and represented approximately 60 percent of the urban population of the Nation’s Capital. Manila has expanded into a sprawling metropolis area called Greater Metro Manila that extends to the provinces of Rizal, Cavite, Laguna, Bulacan and Batangas with a total population estimated at 26.5 million people in 2010 and expected to rapidly expand to about 45-50 million people by 2050 (Cox, 2011)).
The presence of the volcanoes west of Metro Manila enhances the rains during the southwest monsoon season and will continue to threaten the rapidly growing populace of the metropolis. In part, the reason for the heavy rains is due to the presence of Natib and Mariveles volcanoes. Many volcanoes situated near urban centers, which have similar weather as the Philippines may also experience the same orographic and dispersive tail of nonhydrostatic waves phenomena (Smith, 1979), heretofore not associated with massive volcanic cones. Because of its severity and flood impact especially to large populations, it is worthy of further investigation in the Philippines and elsewhere to better understand the phenomenon for possible hazard mitigating solutions, if any.
Cotton, W., Bryan, G., and van den Heever, S.: Storm and Cloud Dynamics: The Dynamics of Clouds and Precipitating Mesoscale Systems, Academic Press, 2011.
Cox, W.: The evolving urban form: Manila, New Geography http://www.newgeography.com/content/002198-the-evolving-urban-form-manila, accessed 9 September 2013, 2011.
Knabb, R., Rhome, J., and Brown, D.: Tropical Cyclone Report, Hurricane Katrina, National Hurricane Center Reports, 2005.
Lagmay, A., Rodolfo, R., and Bato, G.: The perfect storm: Floods devastate Manila, Earth, 55, 50–55, 2010.
NDRRMC: Eﬀects of Southwest Monsoon (Habagat) enhanced by tropical storm “Maring”, National Disaster Risk Reduction and Management Council Sitrep No. 20., 2013.
PAGASA: Definition of terms: Philippine Atmospheric Geophysical and Astronomical Services Administration, http://kidlat.pagasa.dost.gov.ph/cab/define.htm, accessed 8 September 2013, 2013.
Phivolcs: Philippine Institute of Volcanology and Seismology list of active, potentially active and inactive volcanoes, http://www.phivolcs.dost.gov.ph, accessed 8 September 2013, 2013.
Roth, D.: Hurricane Katrina – August 24-September 1, 2005, Tech. rep., Hydrometeorological Prediction Center, 2008.
Smith, R.: The influence of mountains on the atmosphere, Advances in Geophysics, 1979.