Geographic Information Systems

Geography and Planning

University of Western Australia

Week 9 - Lab Eta : Fire hazard analysis using QGIS

Throughout all labs there will be reflective questions to answer. Write down your answers in a notebook, or digitally if
you prefer. Make sure your responses are comprehensive as these notes will help your learning progress and provide a
5 valuable information resource for your unit assessments. If you do not know the answers, talk to your peers and ask
your tutors – they are there to help guide your learning during the labs. If you are unsure of answers outside of lab time,
post your queries to the Blackboard forum.

1. Introduction
10
In this lab you will further advance your environmental GIS analysis by processing some of the data you have already become
familiar with, using more terrain operators. By doing this you will be able to build up a spatial picture of how the landscape may be
prone to lower or higher levels of wildfire hazard.

15 To complete this investigation, you will use some new topographic processing techniques, as well as spatial operators you are
already familiar with including proximity and reclassification. Again, you will use the raster calculator to overlay your layers to build
an environmental model which identifies the spatial distribution of wildfire hazard for the southwest region.

2. Learning outcomes
20
• Processing of elevation data to obtain topographic characteristics
• Creation of buffer zones as distance measures
• Improved use of map algebra algorithms in raster calculator to combine multiple criteria
• Linking key theory of raster data models and topographic analysis within an applied GIS context
25

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 1
 Geographic Information Systems
Geography and Planning
University of Western Australia

3. Some background
Wildfire is a natural hazard which, if particular populations (e.g. human; ecological) are exposed to wildfire, results in an increased
level of risk, especially if population vulnerability is also high. Some natural systems rely on fire regimes for maintaining
30 ecosystems.

However, with the world’s increased population and pressure on land use, wildfire has become an increasing threat to maintaining
safe environments for human and ecological populations. This is because the level of exposure to fire as a hazard has increased,
and hazards themselves have increased in intensity, exacerbated by factors such as a changing climate.
35
We can spatially assess the risk to populations from fire hazards to enable more effective emergency response planning.

Over the next two labs you will build up a picture of environmental risk to wildfire (note: Risk = f(Hazard, Vulnerability) for the
southwest region.
40
• Hazard corresponds to the physical characteristics of a burning fire.
• Vulnerability corresponds to the level of exposure (by a human population or species) to the hazard, as well as the
adaptive capacity available to reduce the potential impacts of the hazard on the population (in this lab, we will consider the
emergency response mechanisms for mitigating the fire hazard as our adaptive capacity).
45
In this week’s lab you will look at assessing the fire hazard. You will assess the hazard by evaluating the environmental factors that
contribute to forest fires. For this lab, these factors will include:

• land cover,
50 • topographic characteristics; and
• meteorological variables.

To look at the potential overall fire hazard, you will need to standardise each factor (just like you did for the cost surface model you
built in the last lab). You will use a similar method to Gai et al. 2011 to reclassify the factors which contribute to fire hazard from 1 to
55 10, where low to high values represent very low to extremely high fire hazard levels, respectively. Then we will use weights to
weight the factors after reclassification.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 2
 Geographic Information Systems
Geography and Planning
University of Western Australia

Finally, you will start to think about preparing some of the data using spatial processing for the next lab which will be needed for
assessing mitigation measures and emergency response (adaptive capacity) for fire risk management and disaster risk
60 reduction.

• Download the data for the lab and unzip the folder.
• Open your QGIS project from last week and create a new group called Lab Eta in the layers panel. Set up your project
home to Lab Eta.
65 • Add all the new data layers to map window.

Before you commence your analysis have a look at the Gai et al. (2011) paper (link above) and make some notes on what you
think you might be looking for in your evaluation of fire hazard. Try to get a clear idea of how your spatial analysis will produce an
output. If you are unsure of anything, discuss with your peers and the tutors.
70

4. Fire hazard factors

75 4.1 - Elevation
Elevation generally relates to temperature and allows us to calculate secondary terrain characteristics such as slope and aspect.
Areas of higher elevation have less oxygen, which is needed for fire to burn, and are also cooler as altitude increases. In some
geographies elevation may also have an influence over precipitation regimes, where hill or mountain ranges may generate
orographic uplift. Elevation plays a highly complex role in all of these potential fire hazard factors, but for our modelling purposes,
80 we will assume that as elevation increases our fire hazard potential decreases.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 3
 Geographic Information Systems
Geography and Planning
University of Western Australia

• Open the reclassification tool, which you should know how to use well by now.
• Reclassify the raster SRTM_DEM (you have these data from last week’s lab) to:
o <200 = 10,
85 o 200-300 = 8,
o 300-400 = 6,
o 400-500 = 4,
o >500 = 2.
o Remember to select Use no data when no range matches value for all reclasses that you do.
90

4.2 - Slope
Slope provides an important factor for controlling the speed of fire as it spreads across the terrain. Generally, steep slopes increase
the speed of fire and therefore should be associated with a larger fire hazard potential; conversely, shallower slopes are associated
95 with slower fire spreading speeds. According to Geoscience Australia every 10-degree (°) increase in slope will double the fire
speed i.e. the speed increases four-fold for a 20 degree slope compared to flat terrain. We can use the elevation information from
the DEM to calculate slope. The GIS will look at the change in elevation from one pixel to the next to determine the corresponding
change in slope.

100 • Open the “Slope” toolbox and use the SRTM_DEM as the input raster, save the output as “Slope” and run the tool.
• Now reclassify your output using:
o 0-0.5 = 1,
o 0.5-1 = 3,
o 1-3 = 5,
105 o 3-5 = 7,
o >5 = 10.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 4
 Geographic Information Systems
Geography and Planning
University of Western Australia

4.3 - Aspect
110 Aspect (direction in which the terrain slope is facing) affects fire behaviour through variations in the amount of solar radiation and
wind that different aspects receive. Generally, in the southern hemisphere, northerly aspects are most favourable for fire to start
and spread. These slope directions receive more sunshine and therefore have lower humidity and potentially higher fuel
temperatures.

115 • Open the “Aspect” toolbox and use SRTM_DEM as the input raster,
• save the output as “Aspect” and run the tool.
• Now reclassify your output using the diagram below to infer what values you will need to set as your reclassification
ranges.

120 • Reclassify as follows:

o Flat = 8,
o N = 10,
o NE = 8,
o E = 6,
125 o SE = 4,
o S = 2,
o SW = 4,
o W = 6,
o NW = 8.
130

• Remember that north facing slopes receive the highest sunshine in the southern hemisphere.
135

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 5
 Geographic Information Systems
Geography and Planning
University of Western Australia

4.4 - Land cover

140 Vegetation cover types such as shrubs and grasslands contribute more to a significant fuel load through ease of fire ignition and potential to
burn rapidly. Fuel load describes the amount of fallen bark, leaf litter and small branches accumulating in the landscape, as well as the potential
for fixed vegetation to burn. The greater the fuel load, the hotter and more intense the fire. Reclassify the Land_Cover.tif raster layer using the
Reclassify by Layer tool.

145 Find the table called “Land_cover.tif.vat” in the LandCover.gpkg and edit the LandCover_Reclass field to allocate values to each of the land
cover values. In undertaking your reclassification process you should allocate an integer value to each of the listed land cover types: A value of
0 indicates zero fuel load/potential for ignition and a value of 10 indicates a very high fuel load/potential for ignition. Use the table below to help
you work through this process. Once you have allocated your values run the Reclassify by Layer tool to reclassify Land_Cover.tif.

fid VALUE ISO_CLASS CLASSLABEL LandCover_Reclass

1 1 Primarily Non-Vegetated_Bare Areas
Bare Areas
2 2 Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation Woody Shrubs Sparse Chenopods
Chenopod Shrubs - Sparse
3 3
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation Herbaceous Graminoids Open Hummock Grasses Hummock Grasses - Open
4 4
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation Herbaceous Graminoids Sparse Hummock Grasses Hummock Grasses - Sparse
5 5
Primarily Non-Vegetated_Waterbodies_Water Inland Waterbodies
6 6
Primarily Non-Vegetated_Waterbodies_Water Brine Salt Lakes
7 7
Primarily Vegetated_Cultivated & Managed Lands_Herbaceous_Graminoids_Irrigated_Pasture Irrigated Pasture
8 8
Primarily Vegetated_Cultivated & Managed Lands_Herbaceous_Graminoids_Rainfed Rainfed Cropping
9 9
Primarily Vegetated_Cultivated & Managed Lands_Herbaceous_Graminoids_Rainfed_Pasture Rainfed Pasture
10 10
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Shrubs_Closed Shrubs - Closed
11 11
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Shrubs_Open Shrubs - Open
12 12
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Shrubs_Scattered Shrubs - Scattered
13 13
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Shrubs_Sparse Shrubs - Sparse
14 14
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Trees_Closed Trees - Closed
15 15
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Trees_Open Trees - Open
16 16
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Trees_Scattered Trees - Scattered
17 17
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Woody_Trees_Sparse Trees - Sparse
18 18
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation Herbaceous Graminoids Open Tussock Grasses Tussock Grasses - Open
19 19
Primarily Vegetated_Natural & Semi-Natural Terrestrial Vegetation_Herbaceous_Graminoids_Sparse_Tussock Grasses Tussock Grasses - Sparse
150

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 6
 Geographic Information Systems
Geography and Planning
University of Western Australia

4.5 - Climate
In addition to topographic and land-use/cover factors, forest fires are also strongly linked to weather and climate. Air temperature
plays an important role in fire behaviour. The air temperature influences the temperature of the potential fuel load and therefore the
quantity of heat energy required to raise vegetation to its ignition point. Wind also causes greater rates of fire spread affecting both
155 the intensity and extent of an active fire.

Q3: What is the difference between climate and weather?

160 For our modelling we will consider the longer-term climate factors which could influence fire risk.
For the annual maximum temperature layer (°C) reclassify the values to:
• <20 = 3,
• 20-22 = 5,
• 22-24 = 7,
165 • >=24 = 9.
The higher the average temperature, the higher the potential fire risk.

For the annual average wind speed layer (m s-1) reclassify the values to
• <5 = 3,
170 • 5-6 = 5,
• 6-7 = 7,
• >=7 = 10.
The higher the average wind speed, the higher the potential fire risk.

175

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 7
 Geographic Information Systems
Geography and Planning
University of Western Australia

5. Calculation of Fire Hazard

Next, before we can combine our standardised layers we need to weight each of the input factors. In multiple criteria analysis, an
expertise is often involved to help make the decision on which environmental factors have the greatest or the lowest influence on
180 fire hazard. For the purposes of this exercise we will simply use equal weights to weight each factor.

Criteria Weights

Elevation 0.167
Slope 0.167
Aspect 0.167
Temperature 0.167
Wind speed 0.167
Land cover 0.167

Q4: Do you agree with this weighting approach?

185

Q5: What is your opinion on which environmental factors you think have the greatest influence on fire hazard?

190

Q6: How might you suggest improving the method for weight allocations for each factor?

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 8
 Geographic Information Systems
Geography and Planning
University of Western Australia

5.1 - Mapping fire hazard

195
Now we are going to combine our environmental fire hazard factors using raster calculator and weight each layer according to the
table above. This is similar to the process you already used in the last lab to combine your environmental layers for generating the
habitat suitability map. We will use an equation (R = ∑Wi Xi) to determine the fire risk, where R is an index of wildfire risk, W is a
weight for the environmental factor i determined by our proportional weights, and X is the value of reclassified values for factor i.
200 Just like you did for the cost surface previously. Open the raster calculator tool and using the weights from the table write up your
expression to combine the data.

Q7: What does your mapped output indicate?

205

Q8: Where is the highest fire hazard spatially?

210
Q9: How does this distribution relate to the distribution of your input environmental factors and the weights you have assigned?

215

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 9
 Geographic Information Systems
Geography and Planning
University of Western Australia

5.2 - Map layout for Fire Hazard and Contributing Factors

Mapping - Fire Hazard Inputs

220 Look to symbolise the inputs into your hazard layer using a suitable colour ramp and apply the range of values in your raster using
the following rationale.

Value range Hazard Level

0–2 Low
>2 – 4 Medium
>4 – 6 High
>6 – 8 Very High
>8 - 10 Extreme

This is an equal interval translation between the range of values in our standardised hazard layers and the use of a linguistic
225 scale (or use of terminology) to describe the level of hazard associated with each input.

Mapping – Fire Hazard Output

Ok – time to put your thinking cap on! We have just undertaken an overlay process to create a weighted output from six
230 standardised inputs with values ranging from 0 to 10. It is easy to think that the output of this process will also result in an output
with a range of values from 0 – 10. BUT whilst that is possible, it may not be the case. Everyone is likely to have a slightly different
output because the range of input values from the standardisation process may not be exactly the same.

When we conducted this exercise with broadly similar inputs as you have done, our output layer had a range of values running from
235 approximately 3.833 to approximately 8.667.

So – what does this mean and why is it happening? What is happening here is that due to the weighting process we end up with a
compression of the output values to create a narrower range of values – this is quite normal and does not mean that anything has
gone wrong, it just means that at the lower and upper extremes of our value range there is a lower probability that we could end up
240 with a value of 0, 1 or 10 (for example). In order to generate an output value of 10 every corresponding cell in all six of the layers
would need to have a value of 10 going into the model and the same for a value of 0 at the other end of the scale. In the middle of
Lab Eta – Fire hazard analysis assessment using QGIS
Version 1.4 – September 2024
Page 10
 Geographic Information Systems
Geography and Planning
University of Western Australia

the numeric scale there is are just more ways that the weighted values could sum up to those particular values. The result of this is
that we tend to get a compression of our range of values towards the mid-range of values.

245 If we were to use the same linguistic scale that we used for our inputs on our output layer we will tend to get an output that seems
to display a lot of the colours present in the mid-range values (in this case we can see a lot of yellow and orange colours present in
the map).

Figure 5.1 – Illustration of Hazard Output using the same linguistic scale as the inputs.

250

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 11
 Geographic Information Systems
Geography and Planning
University of Western Australia

However, if we examine our output and adjust our terminology to reflect the narrower range of values in our output, we can
generate a different interpretation of hazard. In Figure 5.2 below we can now see a greater amount of red, green or blue cells, so it
255 is potentially easier to distinguish between low or extreme levels of hazard.

Figure 5.2 – Illustration of Hazard Output using the newly derived range of values.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 12
 Geographic Information Systems
Geography and Planning
University of Western Australia

Once you have decided on your value break points make a note of the break values and decide what terminology you are going to
260 use for your output.

Figure 5.3 – Using QGIS to help with your translation between values and terminology.

265

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 13
 Geographic Information Systems
Geography and Planning
University of Western Australia

Mapping Exercise - Produce a single map layout that includes your overall fire hazard as the main map frame PLUS six smaller
individual map frames representing the reclassified contributing factors (elevation, slope, aspect, land cover, max temperature and
windspeed), look to include relevant topographic information and labelling to give your map frames spatial context.

270 You can add any key map elements that you like. Make sure your map layout is informative for you to provide to a fire risk manager
to use to help with spatial decision-making. We’ve not given you a template this week as you should practice producing your own
layouts. If you’ve forgotten how to do anything, then you can refer back to the previous lab instructions to refresh your memory.
We’ll deal with mapping the adaptive capacity next week.

275

6. Calculation of Adaptive capacity (emergency response)

There are various mechanisms that we can put in place to help reduce vulnerability to fire hazards and ensure an enhanced
adaptive capacity to mitigate for fire impacts. For example, any nearby Department of Fire and Emergency Service (DFES)
280 stations, water hydrants, and road access to the fire incident can enhance our adaptive capacity to help with fire risk management
should a fire occur.

For us to spatially map these factors we will undertake proximity analysis using Euclidean distance. Remember, we covered the
steps required to undertake this in the last lab.
285
You have vector layers provided for DFES stations (points), water hydrants (points) and state roads (lines). You will need to
add a new field to each attribute table, populate it with “1”, rasterise your vector layers, and then run proximity analysis. Remember
to set the resolution of the outputs to match the resolution we are working with (here it is 1km) and to set the processing extent to
the southwest region. Undertake this now and if you get stuck with what to do, just switch back to last week’s lab instructions to
290 refresh your memory. Suggested output names: DFES_Proximity, Hydrant_Proximity, Roads_Proximity.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 14
 Geographic Information Systems
Geography and Planning
University of Western Australia

Q10: Check your three output maps. By visually looking at these three layers where do you think we have the lowest adaptive
capacity for reducing our vulnerability to fire risk across this region?
295
Once you have developed your proximity raster layers, undertake the reclassification process for each of the three layers
(DFES_Proximity, Hydrant_Proximity, Roads_Proximity) using the reclassify by layer tool. First, you will need to build yourself a
reclassification table for each of the three proximity layers i.e. Reclassify the distance to DFES stations, existing water hydrant and roads
and do this in metres.
300
IMPORTANT: To build a reclass table you can use Excel to create the tables and export the table into a geopackage format, OR you can
follow the steps below.

Layer > Create Layer > New Geopackage Layer…

305 Create a new geopackage using the Database box
Give the table a name
Geometry type > No Geometry
New Field > MIN_VALUE > integer > Add to Fields list
New Field > MAX_VALUE > integer > Add to Fields list
310 New Field > RECLASS_VALUE > integer > Add to Fields list
OK

Next: Open your new attribute table and open an edit session. To add rows to your table click on the Add feature icon. You can
manually edit the values as necessary in your edits session. Once complete run the reclassify by layer tool making sure that you set the
315 range boundaries appropriately.

The principle that you are applying here is that we want to give lower scores to the locations which are closer to each of our adaptive
capacity factors, as they will contribute to less overall risk; when you map the adaptive capacity, higher scores will mean the area is
contributing to overall higher risk.
320
When doing this you should open the histogram for each of the proximity layers to investigate the range of values to help you pick out
appropriate values. Remember, your layers are in metres, so we need to use metres here, not kilometres.

Hint: As a starting point your lowest risk values should be <= 1,000m and these should be allocated a value of 1. Your highest risk
325 values should be allocated a value of 10. Everything in between should be allocated an appropriate value – this is up to you.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 15
 Geographic Information Systems
Geography and Planning
University of Western Australia

6.1 - Mapping adaptive capacity (emergency response)

Criteria Weights

Adaptive DFES 0.333

capacity Water hydrant 0.333
State roads 0.333
330
Use the weights in the above table to build an expression using the raster calculator to combine your three adaptive capacity layers
together. We will use a similar approach to that taken for mapping Hazard. In this case (AC = ∑Wi Xi) to determine adaptive
capacity, where AC is an index of adaptive capacity, W is a weight for the environmental factor i determined by our proportional
weights, and X is the value of reclassified values for factor i.
335
Q11: What does your output dataset reflect?

340 Q12: What does the spatial distribution look like in comparison to your fire hazard map?

Q13: Is this easy to compare?

345

Q14: What do you think of the weightings? How might you vary these?

350

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 16
 Geographic Information Systems
Geography and Planning
University of Western Australia

Symbolise the input to your adaptive capacity layer using the range of values in your raster using the following rationale.

Value range Adaptive

Capacity
0–2 Very good
>2 – 4 Good
>4 – 6 Moderate
>6 – 8 Poor
>8 - 10 Very poor
355
6.2 - Map layout for Adaptive Capacity and Contributing Factors

Produce a single map layout that represents adaptive capacity. If you like you can include the three elements representing the
reclassified contributing factors (road proximity, water hydrant proximity and DFES sites). As with hazard above, you may need to
360 have a think about the translation between output values and terminology when building your combined adaptive capacity map.

You can add any key map elements that you like.

Coming up next Week...

365 Next week is your final lab week. So, we’ll finish off our fire risk mapping by looking at the exposure of human populations to fire
risk and bringing back in our ecological mapping for assessing the vulnerability of our native western quoll species to wildfire. You’ll
be using some additional data for mapping population distribution for generating an exposure map.

Lab Eta – Fire hazard analysis assessment using QGIS

Version 1.4 – September 2024
Page 17