Tropical Cyclone – Population Exposure Analysis


Severe Tropical Cyclone Harold was a very powerful tropical cyclone which caused widespread destruction in the Solomon Islands, Vanuatu, Fiji, and Tonga in April 2020. It was first noted as a developing tropical low within a trough of low-pressure on April 1, when it was located to the east of Papua New Guinea. Over the next day, the system moved south-eastwards over the Solomon Sea, before it was classified as a tropical cyclone and named Harold by the Australian Bureau of Meteorology.

TC Harold Made landfall in Kandavau, Fiji on the 08/04 00:00 UTC, however the predicted cyclone path was available by the 3rd of April. Utilising geospatial overlay techniques, it is possible estimate the potential exposure for different villages and communities before the landfall and take necessary measured for prepared and evacuation.

Learning Objectives

By going through this exercise, we are going to perform Population Exposure Analysis using some geoprocessing functionalities of ARCGIS,

  • Overlay wind speed zones with admin boundary
  • Prepare zonal statistics of wind speed population exposure by admin

Data Inputs


File NameData TypeSource
fji_ppp_2020_UNadj.tifRaster (tif)HDX
fji_polbnda_adm1_district.shpPolygon Shapefile (Vector)HDX
fji_polbnda_adm2_province.shpPolygon Shapefile (Vector)HDX
fji_polbnda_adm3_tikina.shpPolygon Shapefile (Vector)HDX
HAROLD20_Cyclone_Path_JRC.shpPolygon Shapefile (Vector)GDACS

Load Data and Prepared the Workspace in ArcGIS Desktop

  • Start ArcGIS Desktop by double-clicking on the icon on your desktop. If an icon is not present, you can use the Start Menu: Programs > ArcGIS Desktop
  • Load the“HAROLD20_Cyclone_Path_JRC.shp” from the following location X:\UNOSAT\Training_Material\M4\Practical\Data_Input\Vector
  • Also, load the “fji_polbnda_adm3_tikina.shp” from the same location
Fix coordinate system of the data frame if needed

As soon as the datasets are loaded, you will notice the geodata is visualised in extreme left and right corners, and that zooming into feature extent or whole extent does not help investigate Fiji closely. This unique challenge is due to Fiji’s geographic location on both sides of the ante meridian line (180 Degree longitude line). To resolve this issue, we will need to change the coordinate system of the Data frame into a local coordinate system known as Fiji Map Grid 1986. Notice that you might not have this problem if your coordinate system is already set up as Fiji Map Grid 1986.

  • From table of contents, right click on the data frame name “Layers
  • Select Properties, General tab
  • In the name section, typeTC Harold Data”
  • In the Units Section, select Decimal Degrees for Display Units > Click apply
  • Select the coordinate system tab > Expand the Projected Coordinate Systems > National Grid > Oceans > Pacific Ocean > Fiji 1986 Fiji Map Grid
  • Click apply and AddtoFavorites (so it will be easy to find next time)
  • Click the transformation box > Select the first method that is listed > Click ok
Transforming Coordinate System
  • After the transformation is set, Click apply > Click ok

After this you shall be able see the whole of Fiji in the map display without being divided by the 180 line.

Result after correction
  • Save the workspace as “M4_Population_Exposure_TC.mxd” in the location X:\UNOSAT_ADV_FJI\Training_Material\M4\Practical\Workspace


What is the purpose of setting the transformation in the data frame properties?

Overlay the Windspeed Zones with Administrative Boundaries

  • Apply appropriate colours/ symbology to both the layers
  • Inspect the different admins which might be affected by 120km/hr. wind speed
Wind Speed Zones


From a visual inspection, can you please list five Tikina that might be severely affected by the cyclone? What are the criteria that you have applied to come up with the list?


“Intersect” is a common tool used in GIS for overlay analysis. This tool extracts the overlapping portion of an input and an intersect feature. The following graphic demonstrates the operation of the intersect tool.

At this step of the exercise we are going to intersect “Tikina Boundary” and “Windspeed zones” to identify which part of Tikina will be exposed to what level of wind hazard.

Intersect tool
  • From the layer window, open its attribute table of “fji_polbnda_adm3_tikina.shp” by right-clicking Open Attribute Table.

You will notice that this layer contains records with Tikina, Province and Division level.

Administrative levels in the attribute table of “fji_polbnda_adm3_tikina”

To calculate the estimated affected population by administrative unit, we need to intersect both the “HAROLD20_Cyclone_Path_JRC.shp” polygon layer and the “fji_polbnda_adm3_tikina.shp” administrative unit layer.

  • From geoprocessing menu, click Intersect
  • On the Intersect Dialog Box, select the two layers you want to intersect (“HAROLD20_Cyclone_Path_JRC.shp”and “fji_polbnda_adm3_tikina.shp”)
  • Name the output layer file “adm3_wind_zones.shp” > choose the following output location X:\UNOSAT_ADV_FJI\Training_Material\M4\Practical\Data_Output\Vector
  • Click ok to run the tool
Intersect Tool – Using the tool and the result

The output file “adm3_wind_zones” will be automatically added to the layers window.


Can you identify how many Tikina is inside 120 km/hr wind speed zones?

Zonal Statistics Tool

In this step we are going to utilize the zonal statistics tool to calculate the exposed population inside each council and province. The tool is available on the Spatial Analysis toolbox. It works by taking a raster input value and a feature or raster as zone data and aggregates the values for any zones.

There are many sources for population data available. However, due to its accuracy, accessibility, and availability we have chosen a Worldpop dataset for this exercise.

How Zonal Statistics tool works


Estimated total number of people per grid-cell. The dataset is available to download in Geotiff format at a resolution of 3 arc (approximately 100m at the equator). The projection is Geographic Coordinate System, WGS84. The units are number of people per pixel with country totals adjusted to match the corresponding official United Nations population estimates that have been prepared by the Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat (2019 Revision of World Population Prospects). The mapping approach is Random Forest-based dasymetric redistribution.

  • Load “fji_ppp_2020_UNadj.tif” raster from the following location X:\UNOSAT_ADV_FJI\Training_Material\M4\Practical\Data_Input\Raster

A dialog box will pop up asking you to calculate pyramid for faster display of raster.

  • Select no for this specific case. Display will be slow due to the data crossing antemeridian line
A dialog box asking to calculate pyramid

To obtain the estimation of population living within different wind speed zones areas by administrative units we are going to use Zonal Statistics Tools available in Arc Toolbox:

  • Search in Arc Toolbox: Spatial Analyst Tools > Zonal > Zonal Statistics as table
  • Select the following parameters:
    – Input raster or feature zone data => “adm3_wind_zones”
    – Zone field => “FID” (unique identifier for each “adm3_wind_zones” features)
    – Input value raster => “fji_ppp_2020_UNadj.tif”
    – Statistic type => “SUM”
  • Name the output table as “pop_winds_adm3.dbf
  • Save iton X:\UNOSAT_ADV_FJI\Training_Material\M4\Practical\Data_Output\Vector
Inserting data in the Zonal Statistics Tool

If you inspect the new “pop_winds_adm3.dbf”, you will see it has only FID_, COUNT, AREA, SUM but no name of admin. This can be solved by joining “pop_winds_adm3.dbf” with “adm3_wind_zones” using a common key FID_ and FID.

Common key between two layers
  • From TOC, right click on the layer “adm3_wind_zones” > Join and Relate > Join and set the following options:
Joining data
  • Click No in the “create index” dialogue box.
  • Once joined, open the attribute table to “adm3_wind_zones” again and you shall see the info “pop_winds_adm3.dbf” is already linked.


What is the total estimate of affected population by provinces (Give only three most affected provinces)? Also provide the total affected population.

The joined you have performed earlier is temporary. To save the join into a new file:

  • Right-click on “adm3_wind_zones” > Data > Export Data
  • Save the new file as “adm3_wind_zones_pop.shp” in the following location X:\UNOSAT\Training_Material\ModA2\s1-s2\Practical\Data_Output\Vector

Summarize Admin-wise data using Excel

To summarize the output of “adm3_wind_zones” we need to copy attribute data to excel and use a pivot table to organize the results.

  • Right-click on “adm3_wind_zones” > Open attribute table
  • Select all data cell using CTRL+A > Copy the data from the from attribute table
Select and copy all content of the attribute table
  • Open Microsoft Excel > Paste the copied data.
  • As soon as the data is pasted, format the whole dataset as table. This step will ensure selection of data from table easier
In Excel, format as table
  • From the table Design Ribbon in excel, rename the table name to “exp” > Click “Summarize with Pivot Table”
In Excel, summarize with PivotTable
  • Select the option to create the pivot table in new worksheet
Creating PivotTable
  • Drag and Drop
    – ADM3_NAME to “row label”
    – Windspeed to “Column”
    – Value to “Sum of SUM”

As a result, you will see a table as follows:

Pivot Table – Drag and Drop

The table shows the total population by Tikina likely to be exposed to each wind speed throughout the cyclone path.


Change the text format to number and remove decimals for better data representation. And save it as “Population Exposure” in the following folder X:\UNOSAT_ADV_FJI\Training_Material\M4\Practical\Data_Output\


If you are planning evacuation of safety precautions which Tikinas will be your priority?

Challenge: Calculate Potential Exposure of Human Settlements and Critical Infrastructures

Congratulations on completing a population exposure analysis using forecasted cyclone path. It is important to remember the forecasted cyclone paths are updated frequently. And it is always a good idea revise these exposure statistics as soon as new forecasts is available and shows a lot of change.

Using the same principle of superposition or overlaying you shall now to need create an exposure analysis report which may include following features:

  • Number of villages
  • Schools
  • Hospital
  • Any other important facilities

The data can be found in the following folder: X:\UNOSAT_ADV_FJI\Training_Material\M4\Practical\Data_Input\Vector

While performing the analysis keep an emergency management end user in mind.