Australian Bushfires

Introduction

Bushfires in Australia are a regular occurrence, but the 2019-20 season was an unparalleled to previous years in the amount of impact it had on Australia as a whole. These fires started as early as June and sprung out of control in September 2019. The 2019-20 Bushfires have a major impact on every facet of Australia with estimated economic loss predicted to be over $A 4.5 billion and over 1 billion animals died due to these fires and over 18 million hectares of area burned and more than 5,800 buildings burned.

But in this report, we will focus on the direct impact some major bushfires over the last 20 years have had on the Australian land. This report will analyse a dataset and look through multiple sources to understand fully the impact and draw a conclusion.

The Dataset we will be using is from the Australian Institute of Criminology (2020).  The dataset the effect each bushfire over the years had in terms of Area Burned and Buildings Burned. Below in Table 1, you can see the Summary Statistics of the dataset and though no conclusion can be drawn from it is a good place to start. One of the statistics that stands out to me is the Mode of the Buildings Burned is 0 means the most of bushfires didn’t have as much as an effect on buildings as Area Burned which makes sense considering most of the fires were in forests.

Table 1 – Summary Statistics of the Dataset

How many Buildings were Burned per Bushfire?

We concluded above that maybe the bushfires didn’t have the same effect it had on Land from a purely statistical standpoint. This is further emphasised in Figure 1 where we can observe that most of the buildings burned were on the lower side of the spectrum on the graph as 29 of the 30 bushfires analysed burned less than 420 buildings except for the Black Saturday Bushfires 2009 which burned 2029 buildings. According to this data, we can conclude that the Buildings are not impacted by the bushfires as much as the Land cover, but this raises an interesting question of how much correlation do they have for each bushfire?, which is analyse further in the report.

Figure 1 – Histogram of Buildings burned per Bushfire

How Much Area Burned per Bushfire?

Figure 2 – Boxplot of Area Burned (in hectares) per Bushfire

Figure 2 shows us the Area Burned by each Bushfire and by this Boxplot we can now visualise more the numbers we saw in the Summary Statistics of Area Burned. We have an average of 35,000 hectares of area burned per bushfire with most of the Bushfire again in the lower end but push forward by some unusually large outliers like Black Saturday Bushfires 2009 at 450,000 hectares and Black Christmas Bushfires at 300,000 hectares etc. One of the more interesting things to analyse is the range of Area burned by these fires starting at 1,200 hectares Tathra Bushfires to the Black Saturday Bushfires, this could be further analysed to find out the reasons of such range whether it be the weather, type of land or the way we dealt with these fires but that is a topic for a different report.

Which State Had the Most Bushfires over the last 20 years?

Figure 3 – Number of Bushfires in Each State

From Figure 3 we can conclude the majority of bushfires in the last 20 years were in Western Australia which is more than 1/3 of the bushfires compared in this dataset, followed by NSW approximately 1/3 of the bushfires and the rest of the bushfires by the remaining States with Tasmania with the least with just 3%. Now, this could be analysed further by finding more data and by comparing the total Land area in each of these states or even the total Forest Area in each of these states to give this data more meaning and understanding how much that will change this figure. Figure 3 also points out one of the major limitations of this dataset which is the minimal number of Bushfires that are being compared which would be further discussed in conclusion.

Do Area & Buildings Burned to have any Correlation?

Figure 4 – Scatterplot of Area vs Buildings Burned

Figure 4 answers one of the questions raised earlier in the report which did Area & Building Burned correlated and by the Scatterplot above we can say they are somewhat positively correlated meaning the more the Area is burned the more the buildings would be burned which would seem logical but its better if numbers could back it up. Their correlation coefficient as 0.73 which is high all things considered. Again, the major outliers are the Black Saturday and Black Christmas Bushfires.

Which State Was Most Impacted?

Figure 5 – Line graph of Total Area & Buildings Burned per State

Figure 5 is b culmination of the previous graphs and according to it Victoria was the most impacted state in terms of Area & Buildings burned. Again, to point out the major reason for that is the Black Saturday Bushfires which contributes almost 60% of the land burned and approximately 80% of the buildings burned. One other thing I would like to point out is the Total Area in this graph is in thousands of hectares compared to hectares in rest of the graphs to make the graph easier to read so the Figure 5 shouldn’t be taken at face value but as another indication that the Area Burned and Buildings Burned as correlated to each other which can be significantly seen in this graph.

How devastating were the 2019-20 Bushfires compared to previous years?

Bushfires are quite frequent in Australia, so Figure 6 gives us an understanding of how devastating were the 2019-20 Bushfires to Australian Land especially compared in Figure 6.  2019 has more than 3x the numbers of fires as compared to any other year in the last 20 years which signifies that the 2019-20 Bushfires were not a normal occurrence and where a consequence of multiple things which resulted in 2019 becoming the hottest year on record since 1880. (UN Environment 2020).

Figure 6 – Yearly Fires in NSW by the Land cover (UN Environment 2020)

Conclusions

In conclusion, in the last 20 years as compared in this dataset and report Black Saturday 2009 Bushfires comes as the outlier in all the graphs and an anomaly in data. Through analyses of this data, we can also conclude that Area Burned and Buildings burned are positively correlated to each other. Figure 6 also shows how devastating the 2019-20 Bushfires were and if compared as a whole to this dataset would obliterate these numbers from a statistical point of view. In summary, this report raises more questions than it gives answers to which I couldn’t get to due to the length of the report.

But before fully giving an opinion we should understand the limitations of this dataset and by extension, this report which is this dataset only compares 30 Bushfires which is relatively a small number of data to make any significant conclusions and also only considers the direct impact the Bushfires had on the Land while ignoring the importance of different land types, how populated they were by humans and animals, fires impact on atmosphere or public health, bushfires economic impact etc.