Enhancing Risk Analysis Capacities for Flood, Tropical ... · and the provinces of Laguna, Rizal, Cavite, and Bulacan, is particularly vulnerable to the devastating effects of natural

Enhancing Risk Analysis Capacities for Flood, Tropical Cyclone Severe Wind and Earthquake for Greater Metro Manila Area

Component 1 – High Resolution Digital Elevation Data and Imagery

NATIONAL MAPPING AND RESOURCE INFORMATION AUTHORITY

GEOSCIENCE AUSTRALIA

Ofelia T. Castro1, Leo B. Grafil1, Luke Peel

2

1. National Mapping and Resource Information Authority 2. Geoscience Australia

© Republic of the Philippines and the Commonwealth of Australia (Geoscience Australia) 2014

Enhancing Risk Analysis Capacities for Earthquake, Tropical Cyclone Severe Wind and Flood for the iii Greater Metro Manila Area – High Resolution Digital Elevation Data and Imagery

Contents

1 Introduction ............................................................................................................................................ 1

1.1 Area covered by LiDAR and aerial photography ............................................................................. 1

1.2 Data requirements ........................................................................................................................... 1

2 Data ....................................................................................................................................................... 2

2.1 Data collection ................................................................................................................................. 2

3 Methods ................................................................................................................................................. 4

3.1 Sensors ............................................................................................................................................ 4

3.2 GPS station and Ground Control Point (GCP) ................................................................................ 4

3.3 Data processing ............................................................................................................................... 5

3.4 QA / QC............................................................................................................................................ 5

3.4.1 Vertical accuracy assessment .................................................................................................... 6

3.4.2 Horizontal accuracy assessment ............................................................................................... 6

3.4.3 Classification assessment .......................................................................................................... 7

3.4.4 LiDAR-derived DEM assessment ............................................................................................... 7

3.4.5 Ortho-rectified imagery assessment .......................................................................................... 8

4 Conclusion ............................................................................................................................................. 9

5 Acknowledgements .............................................................................................................................10

6 References ..........................................................................................................................................11

7 Attachments .........................................................................................................................................12

iv Enhancing Risk Analysis Capacities for Earthquake, Tropical Cyclone Severe Wind and Flood for the Greater Metro Manila Area – High Resolution Digital Elevation Data and Imagery

Enhancing Risk Analysis Capacities for Earthquake, Tropical Cyclone Severe Wind and Flood for the v Greater Metro Manila Area – High Resolution Digital Elevation Data and Imagery

Executive Summary

Due to its geographical location, the Philippines is highly prone to natural disasters resulting from

earthquakes, volcanic eruptions, tsunamis, tropical cyclones, flooding, extreme winds and landslides.

The effect of climate change, urbanization and rapid population growth increase the frequency and

intensity of disasters in the Philippines.

The Greater Metro Manila Area (GMMA), comprising 16 cities and one municipality of Metro Manila,

and the provinces of Laguna, Rizal, Cavite, and Bulacan, is particularly vulnerable to the devastating

effects of natural disasters. It has a population of 21 million residing on land that is cut by active

earthquake faults and subject to intense riverine flooding. GMMA is also frequently affected by

typhoons, which can result in severe wind damage, storm surge, intense flooding and landslides. The

risk from these natural hazards is further exacerbated by informal settlers living along flood drainage

and riverbanks, areas frequently affected by flooding.

The project “Enhancing Risk Analysis Capacities for Flood, Typhoon, Severe Wind and Earthquake for

Greater Metro Manila Area” or Risk Analysis Project (RAP) aims to analyse the risk from flood, severe

wind and earthquake. An essential input to this activity is high resolution Digital Elevation Model

(DEM), a highly detailed representation of the earth’s surface. Without a high-resolution DEM it is

extremely difficult, if not impossible, to develop a flood risk model in densely urbanized areas like

Metro Manila which has a very flat topography.

The most appropriate methodology to derive high-resolution DEM is through topographic LiDAR.

LiDAR is an aerial survey technique that produces a very high-resolution DEM with vertical accuracy

of <15 cm. It has added benefit of accurately defining the height of features such as buildings and

forest canopies. The aerial photographs will compliment LiDAR in the classification of points, 3D

visualization and 3D modelling.

Acquisition of airborne LiDAR and imagery will provide other benefits to the Government of the

Philippines including detection and mapping of active fault lines, improved ability to develop accurate

exposure information for GMMA, and an ability to accurately determine the impact of different sea

level rise scenarios in the Manila area.

The High Resolution Elevation Data and Imagery (LiDAR and Aerial Photography) Component is one

of the six Components of the “Enhancing Risk Analysis Capacities for Flood, Typhoon, Severe Wind

and Earthquake for Greater Metro Manila Area” or Risk Analysis Project (RAP). The primary purpose

of this component was to develop a high resolution Digital Elevation Model (DEM) derived from Light

Detection and Ranging (LiDAR) technology and orthoimagery from digital aerial photography. The

DEM will be used as the base dataset for natural hazard risk analysis and climate change impacts.

The deliverables included an orthoimage from digital aerial photographs, unclassified and classified

point cloud, LiDAR derived products such as intensity, DEM comprising of Digital Terrain Model

(DTM), Digital Surface Model (DSM), Canopy Elevation Model (CEM) and Foliage Cover Model

(FCM). The extent of these outputs is the cities and municipalities of Metro Manila, along with parts of

Bulacan, Rizal, Laguna and Cavite provinces.

vi Enhancing Risk Analysis Capacities for Earthquake, Tropical Cyclone Severe Wind and Flood for the Greater Metro Manila Area – High Resolution Digital Elevation Data and Imagery

This report documents the acquisition of LiDAR and aerial photography by Fugro Spatial Solutions

(FSS) from Australia from March to April 2011 and the processing of derived products. The results of

the comprehensive QA/QC process conducted by Geoscience Australia (GA) and National Mapping

and Resource Information Authority (NAMRIA) on the deliverables are also presented in this report.

This project was funded by the Australian Agency for International Development (AusAID)

implemented by Office of Civil Defense (OCD) through the National Disaster Risk Reduction and

Management Council (NDRRMC) in partnership with Geoscience Australia (GA) and the Collective

Strengthening of Community Awareness of Natural Disaster (CSCAND) agencies composed of the

Mines and Geosciences Bureau (MGB), the Philippine Institute of Volcanology and Seismology

(PHIVOLCS), the Philippine Atmospheric, Geophysical and Astronomical Services Administration

(PAGASA), and the National Mapping and Resource Information Authority (NAMRIA).

Enhancing Risk Analysis Capacities for Earthquake, Tropical Cyclone Severe Wind and Flood for the 1 Greater Metro Manila Area – High Resolution Digital Elevation Data and Imagery

1 Introduction

Light Detection and Ranging (LiDAR) technology is now widely accepted for the collection of terrain

data to generate high resolution DEMs. It offers fast acquisition and processing of data with minimum

human dependence since most of the processing is done automatically and is capable of day and

night data collection.

As agreed on by the partner agencies, the LiDAR mission was carried out to develop a seamless high

resolution elevation dataset for Greater Metro Manila Area (GMMA). Since there was no LiDAR

service provider in the Philippines at the time of the project, a Request for Quotation based on tender

specifications was issued by the Optical, Geospatial, Radar, and Elevation (OGRE) Supplies and

Services Panel of Geoscience Australia (GA) to all Optical Services Providers in Australia for the

acquisition of LiDAR and photography for the GMMA coverage area. There were three companies that

submitted proposals of which Fugro Spatial Solution (FSS) was selected based on the OGRE

assessment.

This report is based on the final report submitted by FSS covering the LiDAR and aerial photography

mission details, data processing, deliverables and QA/QC results of the generated data by GA and

NAMRIA.

1.1 Area covered by LiDAR and aerial photography

The total project area is approximately 1,232 km2 covering GMMA including the flood investigation

area, fault, Pasig-Marikina river basin, and the shoreline of Manila Bay extending from Bulacan to

Cavite.

1.2 Data requirements

The project area requires high resolution DEM through airborne LiDAR and digital aerial photography

for GMMA. Specifications for data acquisition and products are defined in the Tender Specifications

developed by OGRE with inputs from CSCAND agencies. The next section discusses the important

requirements and specifications.

2 Enhancing Risk Analysis Capacities for Earthquake, Tropical Cyclone Severe Wind and Flood for the Greater Metro Manila Area – High Resolution Digital Elevation Data and Imagery

2 Data

2.1 Data collection

Aerial photography and LiDAR were collected separately. FSS in consultation with GA, decided to

capture aerial photography first since LiDAR could be captured at night and under poorer weather

conditions. Both LiDAR and aerial photography are weather dependent. Data acquisition is limited

during the dry season between January and May. Cloud free imagery is preferred, and atmospheric

and sun effects including sun reflection from water surfaces should be minimised or avoided.

The acquisition for aerial imagery used a metric digital camera to generate seamless ortho-rectified

imagery. The preferred Ground Sample Distance (GSD) is<=20cm with 3 or 4 band multispectral RGB

and infrared channels. The horizontal accuracy required is <= +/- 45cm at the 68% confidence interval.

The aerial photography survey started on 13th March 2011 and continued for 16 days using a Leica

ADS40 digital camera flying at a nominal height of 2,750 meters above mean sea level.

The preferred GSD of <=20cm was not achievable due to limitations in flying time and height (not

allowed to fly below 9,000ft) imposed by the Civil Aviation Authority of the Philippines (CAAP). As a

result, the GSD was increased to 25cm. The northern part following the West Valley Fault (WVF) was

covered by clouds throughout the aerial photography mission. A decision was made not to wait for a

clear sky in that particular area so as not to cause substantial delay in the project

Similarly CAAP restrictions applied to the LiDAR acquisition. The day and night LiDAR mission lasted

for 15 days and was completed on 12th April 2011 covering the entire project area. A Leica ALS50

system was used in the LiDAR survey flying at a nominal height of 2,750 meters above mean sea

level with a swath width of 602 meters and 12.5 degrees field of view. LiDAR and aerial photography

project coverage is shown in Figure 1 covering an area of 1,311km2and 1,291 km

2 respectively. The

total area is more than the required area due to water boundaries along the coast. This resulted in a

variation to the contract to cover the bigger area and additional work.

A minimum Nominal Post Spacing (NPS) of two outbound LiDAR pulses per square meter with 10%

flight line overlap was required. All unclassified and classified point clouds, meaning all returns (all

collected points) were fully calibrated and adjusted to the Philippine Reference System of 1992

(PRS92) datum and required to be LASer File Format (LAS) v1.2 or v1.3 compliant. Deliverables were

provided in orthometric and ellipsoidal vertical datum and in Philippine Transverse Mercator Zone 3

projection system (PTM - Zone III). Fundamental spatial accuracy of the survey must conform to the

Intergovernmental Committee on Surveying and Mapping (ICSM) Category 1 standard with

Fundamental Vertical Accuracy (FVA) of <= +/- 30cm and Fundamental Horizontal Accuracy (FHA) of

<= +/- 80cm both at 95% confidence interval. All classified point cloud data must adhere to the

American Society for Photogrammetry and Remote Sensing (ASPRS) classification scheme based on

ICSM’s level 2 classification.

The extents of the LiDAR and aerial photography are detailed in Figure 2.1.


Figure 2.1. LiDAR data collection extent (shown in blue) and aerial photography coverage (shown in red).


3 Methods

3.1 Sensors

Table 3.1 shows some important system settings used in airborne LiDAR data acquisition within this

project. System settings for the capture of airborne imagery are detailed in Table 3.2.

Table 3.1. Airborne LiDAR system specifications.

LiDAR Sensor LEICA ALS50

Nominal Flying height 2,750 meters(9,000 feet)

Scanner Field of View 12.5 degrees

Nominal Swath width 602 meters

Nominal Airspeed 150 knots

Average point density 2 points per square meter

Footprint diameter 0.62 meter

Navigation Mode GPS based

Positioning Mode DGPS/IPAS

Table 3.2. Airborne digital camera system specifications.

Digital Sensor LEICA ADS40

Nominal Flying height 2,750 meters(9,000 feet)

Field of View 12.5 degrees

Imagery Side Overlap 30%

Imagery Forward Overlap Continuous

Nominal Airspeed 150 knots

Image resolution 25 cm GSD

Navigation Mode GPS based

Positioning Mode DGPS/IPAS

3.2 GPS station and Ground Control Point (GCP)

The ground survey component of the project was carried out by a local contractor FF Cruz & Co., hired

by FSS. The ground control consisted of 35 primary control points and 50 ground validation points

strategically located throughout the project area. The primary control points will serve as reference for

the transformation of the imagery and LiDAR point cloud. FF Cruz was not able to mobilise promptly to

place targets prior to the aerial photography mission, so the ground control was then post-identified

using the acquired imagery.


The Differential Global Positioning System (DGPS) method of control survey referencing nearby

ground control stations was adopted. A base station was established at the main airport in Manila.

This was used to position the LiDAR and imagery sensors in the aircraft. The base station was

surveyed with respect to surrounding major control points and a large GPS dataset was collected to

enable an accurate calculated point position in the International Terrestrial Reference Frame 2005

(ITRF05) datum. The survey returns were turned over to FSS on April 19, 2011.

3.3 Data processing

Full details of LiDAR and aerial photography processing are provided in Attachment A. This includes

the initial post processing of the DGPS observations to obtain the in-flight positions of the LiDAR

sensor. The LiDAR point cloud was subjected to an automatic filtering to classify each point in

accordance to the project specification. Based on the resulting classified point cloud, DEM (DTM and

DSM) were generated and formatted into 1 km by 1 km tiles. The LiDAR DEM was best fitted to the

primary Ground Control Points (GCP). A comparison was made with the adjusted LiDAR heights that

yielded a good correlation between the LiDAR data and the GCP. All products have undergone a

comprehensive accuracy validation. Hydro flattening, a data manipulation process that creates planar

surfaces for waterways and water bodies, has been undertaken for natural and mad-made water

bodies and water courses but is limited to non-tidal water bodies with surface area > 625m2 and

nominal width > 30 m.

Main LiDAR derived products are the 1 meter DSM generated from the “first return”, which includes

ground and non-ground points such as vegetation and buildings, DTM from LiDAR mass point data

classified as “Ground” only, so that it defines the “bare earth” surface, CEM that represents the height

of the tree canopy and FCM that represents density of vegetation greater than 2 meters high, and the

intensity image.

The Leica ADS40 is a pushbroom digital camera using the three-line-scanner principle, whereby linear

arrays on the focal plane capture imagery looking forwards, downwards and backwards from the

aircraft. The ADS40 simultaneously captures data from three panchromatic and four multispectral

bands. Together with the collected GPS and IMU data during flights, the raw images were processed

and combined to generate image strips. Afterwards, aerial triangulation to determine the exact location

and orientation of the image at the time of exposure was carried out using the primary ground controls.

The LiDAR derived DTM was used to rectify the image strips to derive a seamless ortho-rectified

imagery. Image corrections and enhancements have been applied to minimize the effect of cloud

shadows, contrast changes, colour balance and hot spots issues. The orthorectified imagery was

delivered as 1 km x 1 km tiles with 4 bands (RGBI) in TIF format.

3.4 QA / QC

Quality assurance and quality control procedures for the LiDAR derived products and ortho-rectified

imagery were performed primarily by GA with NAMRIA working as a training assistant. The methods of

evaluation include vertical, horizontal, and point cloud classification accuracy validation, file naming

and spatial format specification confirmation, hydro-flattening validation, and to ensure that all data

requirements are met.


3.4.1 Vertical accuracy assessment

Vertical accuracy is the principal criterion in specifying the quality of elevation data, and vertical

accuracy requirements depend upon the intended user applications. For error that is not normally

distributed, ASPRS recommends Accuracy in height be determined by 95th percentile testing. A

normal distribution can be tested by calculating the skewness of the dataset. If the skew exceeds ±0.5

this is a strong indicator of asymmetry in the data and further investigation should be completed to

determine the cause.

This assessment is derived from ground validation points and compared against the elevation of the

derived DEM.

The resulting Root Mean Squared Error (RMSE) for the collected data = 0.109 meter.

NAMRIA further collected additional survey checkpoints and used them to verify the accuracy of

LiDAR and orthophoto. A total of 79 checkpoints were evaluated and resulted in RMSE of 0.11 meter.

The result of the assessment is shown in Table 3.3.

Table 3.3. Vertical accuracy results.

GCP Checkpoints

Average dz 0.004 0.087

Minimum dz -0.277 -0.285

Maximum dz 0.186 0.770

RMSE 0.109 0.110

3.4.2 Horizontal accuracy assessment

Horizontal accuracy is another important characteristic of elevation data; however, it is largely

controlled by the vertical accuracy requirement. If a very high vertical accuracy is required then it will

be essential for the data producer to maintain a very high horizontal accuracy. This is because

horizontal errors in elevation data normally, but not always contribute significantly to the error detected

in vertical accuracy tests.

Horizontal error is more difficult than vertical error to assess in LiDAR datasets. This is because the

land surface often lacks distinct (well defined) topographic features necessary for such tests or

because the resolution of the elevation data is too coarse for precisely locating distinct surface

features. For these reasons, ASPRS does not require horizontal accuracy testing of LiDAR-derived

elevation products. Instead it requires data producers to report the expected horizontal accuracy of

elevation products as determined from system studies or other methods.

The project adopted the method of horizontal assessment by analysing the intensity LiDAR and ortho-

rectified imagery against ground validation points. Road line data from NAMRIA were tested against

50 points and were ranked within the specification of horizontal accuracy supplied by FSS.


3.4.3 Classification assessment

The classified point clouds are delivered in 1 km x 1 km tile fully compliant to LAS v1.2 format. Visual

inspection was conducted using LP360 QA/QC functionalities on various sampling sites with differing

infrastructures to gain a good cross section. Manual editing was performed to further improve the

classification of the LiDAR point cloud on water bodies, vegetation and buildings. A revised and final

dataset was submitted. Figure 3.1 illustrates the classification errors (orange colored points classified

as ground on water bodies) and the corrected point cloud on water bodies reclassified as water shown

in blue.

Figure 3.1. Detail of the coastline along Manila Bay showing water areas erroneously classified as ground (left), and the corrected point cloud (right).

3.4.4 LiDAR-derived DEM assessment

With the improvements on the classification of the LiDAR point cloud, the generated DEM has

improved the definition of the water edges along water courses and along the coastline. GA and

NAMRIA made a comparison of the first supplied DTM in September 2011 and the revised DTM of

November 2012. NAMRIA had generated 1000 random points and reported an overall height

difference of +5 m to -9 m. This may seem significant but investigation of the actual location of the

points would indicate that these height differences are due to corrections in the classified LAS dataset

and the derived DTM. GA statistically analysed these points and concluded that they are outliers or

anomalies. The mean difference for all the points was 0.026m with a standard deviation of 0.48. The

frequency distribution also indicates the majority of points have negligible or no change in height.

GA conducted a spatial analysis of the two DTM datasets by computing the height difference. The

areas with the largest value were located and checked. The results indicate that the revised DTM had

improved values. The revised Classified LAS files were checked/inspected in both plan and profile


view to ensure classification was improved in areas with buildings, vegetation, edges of the coastline,

and where hydro-flattening was applied such as in water courses especially in the tidal areas, that

were not done correctly in the original DTM. The difference surface indicates the spatial distribution of

any height alterations between the two datasets. As the statistics indicate there is little or no significant

height difference and if viewed to the full extents of the image layer there is very little distinction to

observe. The revised DTM provides improvement across the whole project area. This has resulted in a

DTM that is appropriate for detailed flood modeling.

The GMMA LiDAR data meets the ICSM category 1 standard. The overall RMSE of 0.11 m meets the

RMSE <= 30 cm at 95% confidence interval specification based on the ICSM Guidelines for Digital

Elevation Data, Version 1.0.

3.4.5 Ortho-rectified imagery assessment

The preferred GSD for aerial photography was reduced to 25 cm resolution due to the flying height

restriction, i.e. not below 9,000 ft. The accuracy of the generated ortho-rectified imagery ( 46 cm in

easting) was just over the prescribed specification of 45 cm based on FSS’s aerial triangulation

results. This was due to several factors such as the reduction of image spatial resolution, limited

photographic period and bad weather conditions.


4 Conclusion

The LiDAR and aerial photography mission in GMMA successfully acquired high resolution DEM and

imagery. A total of 1,291 km2 for digital aerial photography and 1,311 km

2 for LiDAR were captured

within a period of 16 and 15 days respectively. The data were processed and the resulting products

passed the QA/QC procedure conducted by GA and NAMRIA. Based on the Tender Specifications, all

delivered products for the GMMA-Risk Analysis Project conformed to project specifications.

Some issues encountered during the implementation of this project include the late delivery of primary

and validation GCPs that resulted in significant delays in the data processing. Another factor was the

limited flying time allowed by CAAP due to air traffic within the project area. Less than ideal weather

conditions were also a problem as not all the project areas (the northern end part following West

Valley Fault) were captured by aerial photography.


5 Acknowledgements

This project component was undertaken jointly with the NDRRMC-CSCAND agencies of the

Government of the Philippines and Geoscience Australia (GA) with the funding from the Australian

Agency for International Development (AusAID). The following organizations and individuals are

hereby acknowledged for their input into this project: Phil Tickle, Andreia Siqueira, Kriton Glenn,

Vesna Regulic, Hamish Anderson, Luke Peel, Andrew Clive (Geoscience Australia), Anne Orquiza

(AusAID), Linda SD. Papa, John Santiago F. Fabic, Ofelia T. Castro (NAMRIA), Owen Temby (Fugro

Spatial Solutions), OCD-NDRRMC, PHIVOLCS, PAGASA, and MGB.


6 References

ICSM Guidelines for Digital Elevation Data, Version 1.0, http://www.icsm.gov.au/elevation/ICSM-GuidelinesDigitalElevationDataV1.pdf [accessed 11 June 2011]

ICSM LiDAR Acquisition Specifications and Tender Template, Version 1.0, November 2010, http://www.icsm.gov.au/elevation/LiDAR_Specifications_and_Tender_Template.pdf [accessed 11 June 2011]

http://www.icsm.gov.au/elevation/ICSM-GuidelinesDigitalElevationDataV1.pdf

http://www.icsm.gov.au/elevation/ICSM-GuidelinesDigitalElevationDataV1.pdf

http://www.icsm.gov.au/elevation/LiDAR_Specifications_and_Tender_Template.pdf


7 Attachments

Owen Temby, Final Project Report for Aerial LiDAR and Imagery Surveys of Metro Area of Manila in

the Philippines, Fugro Spatial Solutions Pty. Ltd.

Enhancing Risk Analysis Capacities for Flood, Tropical ... · and the provinces of Laguna, Rizal, Cavite, and Bulacan, is particularly vulnerable to the devastating effects of natural

Documents