Report & Socioeconomic Loss Analysis: James Daniell (Earthquake Report/KIT) & Armand Vervaeck (EQ Report); Shelter: Susan Brink & Friedemann Wenzel (KIT); Social Sensors: Joachim Fohringer, Silke Eggert (GFZ), Andre Dittrich, Christian Lucas (KIT); Disaster Response: Trevor Girard (KIT); Landslides: Bijan Khazai (KIT); Weather: Bernhard Mühr (KIT); Social Vulnerability: Chris Power (KIT), Werner Trieselmann (GFZ); Bohol Pictures and on-site data: Pieter Nierop, Julie Jaramillo & Maria Docoy-Boucher (Bohol, Philippines); General Help & Dissemination: Carlos Robles, Jens Skapski (Earthquake Report), Lee-Jérôme Schumann (GFZ)
Bericht als pdf (engl.)
- Bericht 6 – 02.11.2013, 02:00pm GMT
- Bericht 5 – 21.10.2013, 8.00am GMT
- Bericht 4 – 19.10.2013, 6.00am GMT
- Bericht 3 – 18.10.2013, 10.00am GMT
Preferred Hazard Information:
|EQ Latitude||EQ Longitude||Magnitude||Depth [km]||Fault Mech.||Source||Spectra|
|9.866||124.011||7.1 - 7.2 Mw||20||Thrust||USGS||none avail|
|Country||ISO||Province||Most Impact||Building PF||HDI (2012)||Urbanity||Population|
|Philippines||PH||Bohol||West Coast||Average||0.729||25%||1.3 million|
Preferred Hazard Information:
|Fig. 1: MMI Intensity (USGS Shakemap)||Fig. 2: Aftershocks up to 11am GMT 17.10.2013 & Deaths and Missing up to 17.10.2013|
Hazard Description (Intensities and Ground Motion)
Intensities reached VII on the PEIS scale – very well built structures received slight damage. Older buildings suffered great damage. There was also limited liquefaction. The damage seen corresponds to VIII and perhaps very isolated VIII-IX locations on the MMI scale. Over 3000 aftershocks have occurred, with magnitude 5 earthquakes continuing to pepper the region around Clarin, Loon and Tagbilaran on Bohol. The fault sense can start to be seen well from the PHIVOLCS data, with the fault break running at about WSW-ENE. At least 94 of these have been strong enough to be felt.
Key Hazard Metrics
(VIII‐IX) Epicenter, Loon, Clarin, (VII‐VIII) Tagbilaran City, West Bohol, (VI‐VII) Cebu City, East Coast Cebu, East Bohol
Vulnerability and Exposure Metrics
|Fig. 3: Population per Barangay (2013) aggegated via Tatem et al. (2012)
Municipalities via PhilGIS - Analysis via CATDAT
|Fig. 4: Slope (degree) – CATDAT through ASTER GDEM and the Road system on Bohol (blackl lines)|
The island of Bohol has a capital stock around $5‐6 billion USD with approximately 1.3 million inhabitants. It is mountainous in nature and has the chance for many landslide. Cebu is a key tourist area in the Philippines with 2 million arrivals per year as of 2013. Still, the average income and GDP per capita is about the same as that of the whole of the Philippines. Bohol has a lower GDP per capita in comparison. The main industries are dominated by agriculture which could be affected.
What have been the 2 largest comparable damaging events in the past? None in this region.
|Date-Name||Impact Size||Damage %||Social % or Insured %||Economic Loss|
|1990 Bohol||Mw 6-6.8, VII PEIS||7000 homeless||6 deaths, 200 injured||154m PHP ($ 7m US)|
|1996 Bohol||Mw 5.6, VI PEIS||Poorly built structures||No deaths||Minor|
Preferred Building Damage Information:
Description: Many government, churches and private (over 45641 so far) The counting of buildings destroyed has not been undertaken with only a few houses included in the current count of 10020 destroyed and 35621 damaged. Based on families displaced, this value could be up to at least 15000 destroyed. Loon has been particularly hard hit as well as Clarin, Antequera, Carmen, Tagbilaran and others. See the pictures for locations of current counting.
Secondary Effect Information:
|Type||Impact||Damage %||Social %||Economic %|
|Landslides||Many roads blocked, infrastructure damage||Minor||At least 20 deaths||1-5%|
Preferred Social Impact Information:
|Deaths||230 incl. 8 missing||May rise||Hypocenter played a major role in fatality estimation: 20 to 400 = various models||Daniell, CATDAT, Earthquake Report.|
|**NB: The lowest death toll is currently 186 as 11 are missing. The BQ mall may unfortunately have more victims according to eyewitnesses|
|Injuries||967||Approx. 1000||877 Bohol, 96 Cebu||NDRRMC|
|Sheltered Homeless||87000||Up to 105000||87146 currently in shelters||NDRRMC Data 02.11.2013|
|All Homeless||368691||380906 peak||368691 currently displaced – see below||NDRRMC|
|Affected||3.2m||3.5m peak||Cebu, West Bohol, Negros||NDRRMC|
Preferred Current Economic Impact Information:
|Reconstruction Cost||$162m||$100m-200m||Reconstruction costs have been estimated at 7 billion PHP||NDRRMC/Govt|
|Total Losses||$89.4m||$55m-121m||Total estimate (using rapid loss model combined with damage for range)||CATDAT/James Daniell|
|Insured Losses||<$2m||$1m-5m||Minor insurance takeout but Cebu some||CATDAT|
|Aid Impact||$8m||EU: $3.3mn, AU: $3mn - $46.8mn desired.||NDRRMC|
|Direct Economic Damage (Total) - Summary||Weather|
|There have been estimates of some components of the infrastructure damage being 2.2 billion PHP (around 50 million USD).
The rapid loss estimation of CATDAT/James Daniell, gives a total damage value coming out to between 55-100 million USD (up to 4.5 billion PHP) with a median 89.4 million USD (3.9 billion PHP).
The reconstruction cost has been estimated by the government to be approx 7 billion PHP (162m USD) This is a significant percentage of the gross capital stock of the location, with a MDR approaching 2%.
Social Sensors & Disaster Response
Analysis has been undertaken as to the social media response during the earthquake, and also the associated aid and disaster response after the disaster. See the main chapters for more info.
Insured Loss Estimates
Some public infrastructure damage occurred, and in addition there was minor damage to tourist facilities in various locations. It is still expected that the damage will be insignificant for the insurance industry. In addition no global impacts on supply chains.
Abridged Summary Description from full CATDAT description sources
|A catastrophic earthquake hit the densely populated area of Cebu, and the less densely populated island of Bohol with catastrophic consequences.|
|CATDAT Economic Index Rank||8 – Very Damaging||CATDAT Social Index Rank||8 – Destructive|
Over 3117 aftershocks have occurred so far, with the main ones shown here. Two aftershocks have caused minor additional damage (Earthquake Report). 94 main felt aftershocks have occurred. However, there is still potential for larger aftershocks up to M=6 which could cause additional damage. The fault trace is from the description from PHIVOLCS. There are ranges of the fault trace map from N40E to N60E. Mapping of the fault continues.
|The peak ground acceleration was likely around 0.7-0.8g with the highest recording being 0.37g. Data was collected 02.11.2013 from USGS.|
Geological Effects and Earthquake-Induced Landslides
The fault scarp has been found from Brgy. Napo in Inabanga to Brgy. Anonang in Buenavista (according to PHIVOLCS). The range of the height of the fault scarp has been measured from 10cm to 5m.
Fig. 5: PHIVOLCS-DOST website
In addition, there has been coastal uplift observed in Maribojoc. It is expected that the fault extends from Inabanga/Buenavista, down to Loon and Maribojoc in an approximately NE-SW direction.
Fig. 6: Annotated by PHIVOLCS-DOST website
In total, at least 20 people died due to landslide effects. This is much less than the 107 dead from landslides out of 113 dead and missing from the February 2012 Negros earthquake.
The magnitude 7.2 earthquake on 15 October 2013 main shock, triggered shallow landslides that can be observed on the steep natural slopes of the famous Chocolate Hills in Bohol. The shallow disaggregated landslides are typically not associated with particular geologic units and/or type of slopes. They are usually as deep as the root zone of the vegetative cover, anywhere from several decimeters to a meter deep, and consist of dry, highly disaggregated and fractured material that cascaded down-slope to flatter areas at or near the base of slopes.
Shallow disaggregated landslides account form most the failure types after earthquakes. However, some of the landslides shown on the Chocolate Hills (when looked at more closely are more deep- seated rock and earth slumps that involve relatively large volumes of material (see Figures at the end of the report from Julie Jaramillo). Earthquake triggered landslides contributed to the following noted disruptions as shown and more than 40 barangays have reported landslides :
- The highway in Cortes particularly in Lilo-and was rendered impassable due to a landslide. A part of Cortes’ highway was also damaged.
- In Balilihan, the Bohol Mayor, Dominisio Chatto has confirmed that 5 people died from a landslide due to the earthquake.
- In Loon, Bohol, numerous landslides have affected areas.
Fig. 7: Landslides in each barabgay according to NDRRMC data
|Chocolate Hills Landslides (courtesy: @docjolt)||Julie Jaramillo on site at Choc Hills|
To aid in the rescue and relief as well as resettlement operations, the National Anti‐Poverty Commission (NAPC) on Wednesday released the maps of Cebu, Negros Oriental, and Siquijor indicating location of poor households in barangays that are exposed to high risks of landslides.
The landslide distances are within the bounds seen in historical earthquakes as tested by Bijan Khazai for ChiChi, Northridge, Seattle (Nisqually) and El Salvador earthquakes, and James Daniell for Cebu‐Bohol (after Keefer, 1999; Khazai, 2004).
|Keefer (1984a) presents magnitude‐distance relationships using two distance definitions (epicentral and fault projection) for three different landslide categories: coherent, disrupted slides and falls, and lateral spreads and flows. The figure above shows the earthquake magnitude and the maximum distance from the epicenter limit curves obtained by Keefer (1984a) for both coherent and disrupted slides. Superimposed on these curves is a suite of more recent events that plots well within this envelope, indicating that for the most part the types of landslides that occurred were quite typical of what can be expected in major earthquakes. The range of landslides is shown in this diagram, but only the furthest away counts as the maximum distance (in this case around 65km).|
|Fig. 8: Range of Landslides|
|Poverty incidence (2009) Observed home values vs. spatially weighted average of neighbors shows that Bohol
Source: Philippine Human Development Report 2012/2013; Geography and human development in the Philippines: page 5
This region holds 10.75 % of the Total Elderly Population, as a percent of the Population of the Region VII the elderly make up 4.75 % of the total population which is higher than the national average (3.77 %).
Most of the dead and missing are in Loon (68), Antequera, Maribojoc, Tubigon, Clarin, Sagbayan and are primarily located on the western side of Bohol island. There are currently 222 dead and 8 missing reported.
At least 20 deaths are via landslides, and around 20 deaths are also via heart attacks and panic-related injuries. The proportion of the other 190 are expected to be shaking related.
Since the last update, consolidation has occurred with the numbers – with only 96 people injured in Cebu compared with 877 in Bohol. 1 person has been injured in Siquijor, 1 in Negros Oriental and 1 in Ilolio.
Shelter and damage to homes
The number of homeless from this earthquake is somewhere between 87,000 and 370,000. Due to aftershocks, damaged houses, lack of basic services, 87,000 people are in shelters, but many remain also in tents next to their homes or along roads (around 280,000 are outside of shelters but still in need of basic support.
Emergency Situation according to NDRRMC site report 34 (02.11.2013) for displaced families and persons who are in or outside evacuation centers.
|Provinces||No. of evacuation centers||Displaced families/persons||Families/persons inside centers||Families/persons outside centers|
The numbers are still being counted and we must wait for more information as to the total of long- term homeless vs. the current short-term homeless numbers and in addition, the destroyed and damaged buildings vs. the homeless numbers.
|Fig. 9: Displaced as of 21.20.2013 in the Bohol Earthquake, NDRRMC Data & CATDAT Analysis||Fig. 10: Affected as of 21.20.2013 in the Bohol Earthquake, NDRRMC Data & CATDAT Analysis|
|Fig. 11: Destroyed Buildings counted as of 21.10.2013 in the Bohol Earthquake NDRRMC Data & CATDAT Analysis DATA IS CURRENTLY STILL BEING COLLECTED BY NDRRMC||Fig. 12: Damaged Buildings counted as of 21.10.2013 in the Bohol Earthquake NDRRMC Data & CATDAT Analysis DATA IS CURRENTLY STILL BEING COLLECTED BY NDRRMC|
Over 40% of people logging into the Earthquake Report website in the first 10 mins were from Cebu City, 12% of people logging into the website in the first 6 hours were from Cebu City, and around 30% from Philippines. The following diagram shows visitors in the first 6 hours from each city. The darker circle in Cebu City, and the other blue circle is Manila. Individual peaks were seen with each major aftershock and the initial alert after 1 minute was from IP address increases.
The social sensoring project of CEDIM at GFZ (Joachim Fohringer and Silke Eggert) has been active with characterising Twitter responses. Here is the full analysis that they have undertaken:
Method & Data
To get rapid information on what eye-witnesses report from the epicentral area, we analyze twitter messages related to the Oct. 15 Mw7.1 Philippines (Bohol) earthquake. We extract Twitter messages, so called tweets containing one or more specified keywords from the semantic field of “earthquake” and use them for further analysis shown below. For the time frame of Oct. 15 to Oct 18 we get a data base of in total 50.000 tweets whereof 2900 tweets are localized and 470 have a photo attached.
When do people twitter?
In general, people start to twitter seconds after or even during the shaking they feel. Studies showed that the first tweets related to an earthquake are sent ~20 second after the p-wave arrival. Figure 1 shows the number of tweets per 10 minutes for the first three days after the mainshock. The number of messages decays with the time but major aftershocks still evoke a peak of tweets even in later time. The graph also reflects very well day-time and night-time (11pm – 7am are highlighted in grey in the time axis).
Where do people twitter?To see the spatial distribution of the messages, we created a heat map, shown in figure 2. Apart of the world wide reaction on the earthquake, the main concentration of tweets is located on the Philippines Island, of course. The hotspot Manila, the country’s capital, reflects the national reaction on the event. A very interesting hotspot is the city of Cebu, the closest and therefore most affected city to the epicenter.
Fig. 13: Spatial distribution of tweets on national level. Two hotspots of Twitter activity Manila and Cebu City are marked clearly.
What do people twitter?
To gain insight into what people affected by the earthquake report from the epicentral area, how eye- witnesses appraise the situation and what they report about damage and losses, we categorize the tweets related to their content. We set up 7 categories: earthquake alerts, felt reports, damage, evacuation, call for pray/ help for the victims, private news and media news. Figure 3 shows that half of the messages come from the two last groups of private and media news. Since these messages are not relevant for hazard assessment we focus our studies on the other 5 categories. Messages most interesting for rapid hazard assessment make about 1⁄4 of all messages: tweets on felt aftershocks can be used to map intensities and dimensions of the epicentral area; tweets on evacuations and damages can be used for rapid assessment.
Fig. 14: Categories of tweets and their percental distribution: Above left: all tweets, below right: tweets directly related to the earthquake.
Case study Cebu
While messages calling for spiritual support or containing personal information are distributed all over the country, tweets related to damage, evacuation and felt shaking are located in the epicentral area. They give a first view of the earthquake’s effects and dimensions. Cebu (870.000 inhabitants) is the closest and probably most affected city to the epicenter. Figure 4 shows a close up view of the municipal area with all tweets categorized by their content. Messages having a photo attached are highlighted. Most tweets report about felt aftershocks but especially in the city center messages report about damage and evacuations.
Fig. 15: Distribution of tweets in the municipal area of Cebu City. Text messages (circle) and messages with photo included (camera) are colored by category. Two examples are shown in detail.
As mentioned above, we focus our studies on two aspects: damage and intensities. We classify all tweets related to damage and evacuation according to the type of damage. Furthermore, we “translate” the content into intensities and color it according to the European Macroseismic Scale EMS-98 (see figure 5).
Fig. 16: Tweets related to damage in the municipal area of Cebu. The symbols represent the type of damage, colors mark the intensity according to the EMS-98 scale.
Most people report about destroyed buildings, mainly about their home’s interior affected by the shaking. Mainly fallen objects or small cracks in the walls are observed, not more than slight damages (V/ VI). The damaging reports (VII) come from two heavily damaged churches with collapsed steeples and some major cracks. Some people report about destroyed bridges and impassable roads whereas very few tweets are related to debris, landslides or power outage. Minor but nonetheless useful information for rapid response is also the group of tweets reporting about no damage. The map gives a first overview about the situation after the earthquake in the city of Cebu. Intensities are always subjective and influenced by the author’s impression. But numerous tweets on one building or attached pictures can serve as a proof for correctness.
To see where and with what intensity people still feel the shaking of the aftershocks, we translate all tweets related to the category “felt” into EMS-98 intensities: felt nothing (I), shaking (V), strong shaking (VI), and very strong shaking (VII). An interpolated map of the intensities within the municipal area of Cebu is shown in figure 6. Especially for the city center, we get a detailed, though still subjective image of the shaking. Peripheral areas should be treated with caution.
Fig. 17: Intensity map for Cebu City created from the tweets related to felt shaking. The intensity is colored according to the EMS-98 scale, tweets included in the calculation are marked with black circles
Twitter messages related to the Mw7.1 Bohol earthquake give fast and eye-witness based information on the impact within the epicentral area. Nationwide, people mainly distribute news related to the earthquake and call for prayers, tweets from the epicentral area concern mainly aftershocks felt. For the city of Cebu we could produce both an intensity map and a damage map converting the content into the EMS-98 scale. Both maps give detailed information on local scale. Reliability of the information is an issue but attached photos and a high numbers of tweets concerning the same issue can be used as a correction factor or proof. All things considered, the information extracted from Twitter is an important and useful piece to the situational awareness of the earthquake’s impact.
TENAS – Twitter Event Notification and Analysis Service - KIT (Andre Dittrich and Christian Lucas)
(spatial resolution: 0.25° x 0.25°; temporal resolution: approx. 1 minute)
The grid-based Twitter Event Notification and Analysis Service of the CEDIM FDA Taskforce detected the earthquake and delivered an automatic e-mail alert at 15th October, 12:15:07 UTC, i.e. a delay of 2 minutes and approx. 30 seconds. The automatic alerts from official notification services took for example about 30 minutes (GEOFON) and 39 minutes (USGS).
The first alert was generated in the grid cell with the center at 123.875° (longitude) and 10.375° (latitude), i.e. Cebu, Philippines, based on geo-referenced messages created between 12:14 and 12:15 UTC.
|Cell center||123.875° (longitude) 10.375° (latitude)|
|Centroid of messages||123.9094° (longitude) 10.334° (latitude)|
|Keywords||earthquake 5 times*|
|Number of messages||8|
|Cell mean number of messages||2.1429|
|Cell standard deviation of number of messages||1.801|
|*English and tagalog occurences combined
**Probability that the statistical test accepts the alternative hypothesis (“Event occurred”) if it is true
Following alerts for the same event occurred for several hours. Scattered messages reporting damage mainly concerning infrastructure elements were collected for further analyses. Most of the aftershocks exceeding an approx. magnitude of 4.7 could be separately detected and classified as such. In total, the grid-based approach collected 1191 geo-referenced messages related to the earthquake and its subsequent aftershocks during a time range of about one week. Further relevant Tweets without direct geo-reference (i.e. geographical coordinates) were identified via keyword searches.
The disaster related keywords that were mainly used to communicate this specific event on Twitter were:
earthquake, quake, shaking, shock, aftershock, epicenter, warning, depth, tremor, collapse, alarm, tsunami, magnitude, crack, destroyed, damage
The figure shows the automatically generated map provided through a link in the e-mail alarm.
Over 530 felt reports were collected on Earthquake-Report.com, with additional intensity estimates from many people which are currently being geocoded. It can be seen that a lot of the responses came from Cebu City, but scattered responses came also from Bohol.
Information Gap Analysis
Each chart is the result of an analysis of the information produced within the first three days following the disaster. Information for the 15 Oct 2013 earthquake in Bohol was retrieved as it was released, and was mostly obtained from Earthquake Report, NDRRMC Situation Reports obtained from ReliefWeb, and various other documents released on ReliefWeb. The information for the 06 Feb 2012 earthquake in Negros Oriental was obtained from reports archived on ReliefWeb, with the majority being NDRRMC Situation Reports. All information obtained from the referenced sources was categorized under the headings listed on the left side of the graph. Each piece of critical information was separated into three types, being basic data (who, what, where, when), analysis (how, so what), and root causes (how come, why so). Refer to Review of Publicly Available Information document in appendix which summarizes information provided.
It is not yet clear which requirement for shelters will emerge from this situation. There are no obvious natural aggravating factors to seek shelter. The weather conditions are fair (day temperatures of about 30 centigrade, night temperatures of 24 degrees; little rain announced); the aftershock activity is (currently) moderate. As most affected areas are rural people likely have social mechanisms to avoid shelters. The total number of shelter seeking persons should be below 100000.
The attached GFS rain forecast for the next 120 hours is shown below. No organized tropical systems will affect the Cebu/Bohol area within next 5‐7 days. Most rain expected over the Mindanao‐Sea and the southern half of Bohol, especially in mountainous areas. All other areas have to take into account some thunderstorms, only isolated, however may be heavy for a short while.
October is still monsoon (and typhoon) season. A tropical cyclone must taken into account within next weeks. 3 and 6 hourly rain forecast (updated every 6 hours) can be found at Wettergefahren-Frühwarnung.
The vast rain area to the east is related to the track of "Francisco" another typhoon heading for Japan, but not affecting Philippines.
It can be seen that around 100‐200mm of rain may fall in the next 5 days in Bohol, increasing the chances of major landslides, if large aftershocks occur. 200mm‐300mm is the average for October on Bohol using data from 1961‐ 1990.
|Fig. 18: Long term average precipitation (1961-1990) over the Philippines Data source: Climate Research Unit (CRU)||Fig. 19: Forecast of 192h-accumulated rain amount until 26 October 2013. Base: GFS-Model, Initialisation: 18 October 2013, 00 UTC|
by Julie Jaramillo (All rights reserved)