-
Home
-
About CHARIM
-
Countries
-
Methodology book
-
Use case book
-
Data management book
-
Forum
-
Geonode
You are here
6.2 National scale exposure analysis for river flooding in low lying areas
Keywords:
Flood hazard, risk analysis, loss estimation
| Before you start: | Use case Location: | Uses GIS data: | Authors: |
|
You will need an appreciation of what flood hazard information is and how it is generated. Please see8.6 national flood hazard methodology for more information. Some basic GIS spatial analysis skills will be useful if you wish to follow the analysis method for yourself or apply it to your own data. |
The Use Case location is at the Belize national scale. Note, the analysis applies equally at smaller scales (e.g. District) and for other areas, provided the data at that level of detail is available to carry out the analysis. |
Yes, indicative flood hazard layers for Belize, road vector layers and population density raster data, see table in Data Requirements section below. All are projected in WGS84, UTM Zone 16N. |
Mark Trigg, Mark Brussel |
Introduction:
At a national level, a government must make many decisions that require a balanced understanding of multiple issues, and therefore needs specific and reliable information on those issues. One issue which is integral to many planning decisions is that of flood risk. For example, a government may need to know what part of the population is exposed to potential flood hazard and where that risk is most serious. This can guide investment in disaster preparedness, in terms of shelter location and their capacity, as well as evacuation plans and event response procedures. Infrastructure at risk of flooding can also be identified and evaluating potential losses that may occur, can help decision making, in terms of protecting valuable or strategic assets, or upgrading structures. Indeed, “the value of vulnerability assessment and reduction” is recognized as one of the guiding principles of the National Hazard Mitigation Policy of Belize.
Even where good flood hazard information does not exist in a formal sense, it is likely that responsible persons within a government will use experience and historical knowledge to guide their plans to some extent. However, this knowledge is often lost when the experienced person retires and hence planning decisions may not include all potential risk, in an objective way. It is also difficult to integrate this informal knowledge into an increasingly data rich world, making flood risk analysis difficult and potentially inconsistent.
Many governments now use national flood hazard information which has been formally defined for different return periods (probability of occurrence), sometimes simplified into low, medium and high hazard categories (see definitions in the Methodology Book, for instance, sections 3 flood hazard assessment and section 5 Risk Assessment. Traditionally, this information is provided in map form, but is at its most useful in digital GIS format. A computerized digital format allows cross-analysis of the flood hazard information against the many other digital datasets that are now available. This analysis can be carried out rapidly and more accurately, and is consistent and repeatable.
Below is a schematic which illustrates the definition of flood risk (see section 5.1 of the methodology book for more detail). This Use Case focuses on using the flood hazard information defined in Use Case 8.6 national flood hazard method to identify exposure at a national scale. The vulnerability element is not explicitly included in the analysis due to a lack of data, although it is discussed so that appropriate data can be collected for future work.
While this Use Case focuses on providing a pre-disaster assessment of flood risk, it should be noted that methods are similar to and closely tied to post-disaster assessments. In Belize, these assessments are coordinated by NEMO (National Emergency Management Organization) and are called DANA (Damage Assessment and Needs Analysis) reports. Many government agencies contribute to these reports and they provide a valuable analysis of losses arising from a natural disaster. An example for Hurricane Ernesto in 2012 can be found on the Belize NEMO website.
The main difference is that a pre-disaster assessment uses predicted hazard outputs from modeling exercises and a post-disaster assessment uses information for an actual event, collected in the field by survey teams or by satellite/airplane.
Objectives:
The overall objective of this Use Case is to demonstrate, using two example analysis categories, how to carry out a flood risk analysis at a national scale. It uses national indicative flood hazard maps and national level data to identify what and who may be at risk during floods, i.e. exposure. Population and roads are used as example categories in this Use Case analysis, but the method can be applied to other categories as well. The exposure information is then aggregated into enumeration areas to demonstrate how this can provide further value from the information in a planning context. Finally, it outlines how further analysis of the vulnerability of the flood exposure can be used to provide an estimate of the actual losses that might occur.
Flowchart:
Simple flow chart
Detailed flowchart
Problems definition and specifications:
A national government (we use Belize for this Use Case for illustration) wishes to identify what or who may be at risk for a given flood hazard. This may need to be further quantified in terms of losses that may arise if such an event occurs. The resulting information needs to be as accurate and spatially detailed as possible in order to undertake informed decisions at a national level.
For this particular Use Case we assume that the government wants to know where the largest population at flood risk is located within the country, what length of roads in the country may be affected by floods under the 1-in-100 year return period hazard (1% probability).
Data requirements:
National indicative flood hazard extent data for the return period (probability) of interest. This can be in raster or vector form as illustrated in this Use Case. For just exposure estimates, flood extent is sufficient, but for vulnerability assessment, some measure of flood magnitude (i.e. depth or velocity as a measure of intensity) is required.
Spatially detailed category information is required to estimate the category exposure to flood. This needs to be sufficiently detailed that when compared to the flood location, it is accurate enough to place the category correctly inside or outside the flood extent. It is possible to disaggregate coarse data to a finer scale using assumptions of how that category is distributed (as is the case for the population data we use here), although the results will be of more limited accuracy.
If available, vulnerability information for the category of interest is needed to assess the full flood risk for the category. However, this level of data is rarely available and even if generic information is available, it may not be relevant locally if for example, construction methods are different.
For the population example below, we use population density data from the Worldpop project (http://www.worldpop.org.uk/). This is based on national population statistics disaggregated by urban extents. We also use Statistical Institute of Belize data for population by enumeration area to adjust these data to more accurately match official national values. For the following roads example below, we use national roads data layer with details of road type and unit replacement costs per mile.
|
|
Layer name |
Type |
Description |
|
1 |
enumeration areas.shp |
Vector shapefile |
Enumeration areas of Belize |
|
2 |
BLZ_pph_v2b_2010.tif |
Geotiff raster |
2010 population density raster (100 m) from Worldpop data set. (http://www.worldpop.org.uk/) |
|
3 |
Belize_Survey_Road_Nov15_V4.0.shp |
Vector shapefile |
Road vectors surveyed as part of the BCRIP project |
|
4 |
FP_flood_1in100.tif |
Geotiff raster |
Fluvial and pluvial flood hazard (no coastal) for 1 in 100 year return period.(water depth). NFHM output |
|
5 |
FP_flood_1in100_GtZ.tif |
Geotiff raster |
Fluvial and pluvial flood hazard >0 depth for 1 in 100 year return period. Binary flood =1, dry = 0. Derived from (4) |
|
6 |
FP_flood_1in100_GtZ_Poly.shp |
Vector shapefile |
Fluvial and pluvial flood hazard >0 depth for 1 in 100 year return period. Only flood polygons. Derived from (5) |
|
7 |
population_exposed_100y.tif |
Geotiff raster |
OUTPUT: Example result of population exposed from analysis |
|
8 |
roads_100y_clip.shp |
Vector shapefile |
OUTPUT: Example result of road lengths exposed from analysis |
|
9 |
EnumArea_100y_flood_pop.shp |
Vector shapefile |
OUTPUT: Example result of population exposed aggregated by enumeration area |
Table 1: showing list and description of GIS data used
Analysis steps:
The analysis can be divided into two main analysis steps and these are described in general for all categories below, followed with details of the application of these steps to population and roads in Belize for the 1-in-100 year fluvial and pluvial hazard (but not coastal storm surge). The free open source GIS program QGIS is used as the tool for this analysis, but any GIS platform is capable of being used for this analysis.
Generic analysis applicable to all categories
1. Geospatial analysis of flood extent against a category of interest.
An analysis category can be any subject of interest for which flood risk is important, for example population, roads, or agricultural farm land. In order to identify what in the analysis category is exposed to the flood extent, the national flood hazard map for a given return period has to be spatially intersected with the spatial data for the analysis category. The analysis needs to be restricted to a given area of interest to be meaningful (e.g. National area), and the two datasets in the analysis need to be complete for this area of analysis to be accurate. You may want to subdivide this analysis by sub areas, e.g. districts or elevation categories, to understand the spatial distribution better or provide the output in a particular context for the user. The more spatially detailed the data of the category is to be analyzed, generally the better the analysis will be. However, there will always be a limit to the quality and level of detail possible in the analysis, constrained by limitations in the data and methods. This step of analysis will yield details of what in the category is at risk and where it overlaps the defined hazard area.
2. Loss analysis
If you have further information on a category, such as economic value of a particular land use, you can continue the analysis and calculate the potential loss, if that actual location was subject to flood losses. This can be as simple as multiplying the area flooded for a given category in step (1) with the, value per area, of that category.
Obviously, the reality of flood damage to a particular piece of infrastructure or property is not as simple as the complete loss or damage of the object, and actually involves an increasing loss based on how severe the flood is at that location and how vulnerable that object is to the flood. Therefore, the ultimate goal of this analysis is to identify for a given flood intensity (e.g. flood depth or velocity) what the losses may be. This requires high quality information on the flood intensity and also, locally valid, vulnerability information related to that intensity (commonly known as vulnerability curves). See definitions in chapters 3 Flood Hazards and 5 Risk Assessment of the Methodology Book, for instance, sections 3.1 and 5.3. This level of analysis requires accurate and detailed information to be valid and is not undertaken for this Use Case. However, as the information collated by a government on actual flood losses and flood hazard data improves, this type of analysis will become possible in a meaningful way in the future.
Example Application
1. Application to population exposure to flood hazard in Belize
This illustrates an analysis based on raster data.
It should be noted that the population values in step (vi) above have been adjusted in line with the official government statistics for the enumeration areas. This is necessary as the spatially distributed values in the Worldpop are limited in accuracy and tend to overestimate population density in remote rural areas, when we compare the total population values in rural enumeration areas. However the use of the Worldpop data is necessary as we need some understanding of how the population is actually spatially distributed within the enumeration area in order to intersect it with the flood extent. To adjust the exposed population calculated in step (iv), the result is scaled by the ratio of the total population provided by the government for the enumeration area, to that calculated from the Worldpop data, i.e. if the government reported total is higher than Worldpop, the exposed population is adjusted upwards by that ratio. This adjustment is carried out for each enumeration area. It would also be possible to adjust the whole Worldpop raster dataset in a similar way before using it in the analysis, but ultimately it would be better for the government to develop its own, more accurate, spatially distributed population dataset for this purpose, and we only use Worldpop here to illustrate the process.
2. Application to road length exposed to flood hazard in Belize
This illustrates an analysis based on vector data.
Note, the road analysis above does not include bridges or culverts, which would need to be included for a full picture of road infrastructure exposure.
Results:
We now have an estimate of population and road exposure in Belize to the 1-in-100 year flood hazard. This exposure assessment, in itself, has planning value in terms of identifying areas most exposed to the flood hazard and therefore where to focus attention. However, it has not been possible to carry out the risk assessment examples to a full conclusion, as vulnerability data for Belize is not currently available. For roads, for example, we know that the entire length of road affected by the flood hazard is unlikely to be destroyed completely and that the damage will actually depend on depth and velocity of flooding as well as duration in some cases (for more details, see Use Case 3.3.4 Flood mitigation for roads). Therefore, if we had locally relevant data on road vulnerability to different magnitudes (intensity) of flooding, it would be possible to extend the analysis above to take this into account.
Similarly, with more detailed population demographics, it would be possible to identify risk to population groups based on flood depth. For example, some groups such as the young and elderly would be more vulnerable to shallower flooding than healthy adults and this could be taken into account in the risk analysis.
It is possible to undertake this form of analysis with any spatially distributed data, for example agricultural land use. Although some data as to how crop types are affected by different flood depths and at what point in the growing season a flood occurs, will be important for a full flood risk analysis of agricultural risk.
It should also be remembered that the indicative flood hazard map represents the same probability hazard in all locations and will not occur all at the same time. However, this is important when comparing risk nationally in a consistent manner.
Conclusions:
Spatially detailed flood hazard information allows an assessment of the exposure of any category of interest for planning purposes, provided spatially detailed data is also available for that category. The risk analysis can be extended to account for the category vulnerability, although this requires additional locally relevant data. The accuracy of the results of these analyses depends on the category data being both up-to-date and spatially detailed enough to be meaningful at the scale of analysis. This means an investment at a national level in collecting spatial planning data will result in more accurate and relevant hazard analyses for planning.
Last update:
15-07-2016
Post new comment