Scottish Fire and Rescue Service (SFRS) - wildfire: incident reporting system - data analyses

5 Investigation of patterns in the IRS wildfires dataset

5.1 Overview

We used the information recorded in the filtered IRS wildfire dataset to identify patterns of wildfire occurrence related to a) time and seasonality; b) vegetation and fuel types; c) size of burnt areas; d) motives of ignition along with causes of ignition, ignition source and items causing ignition and spreading fire (where available) and e) accessibility and proximity to urban areas. For the latter we used the Scottish Government Urban Rural Classification (downloaded in ESRI shapefile format from https://spatialdata.gov.scot) which provides a consistent way of defining urban and rural areas across Scotland and is based upon two main criteria: a) population as defined by National Records of Scotland (NRS), and b) accessibility based on drive time analysis to differentiate between accessible and remote areas in Scotland.

For this analysis, we used the full dataset of identified wildfires (n=9,725) and a subset of wildfire incidents with damage area greater than 1,000 m² (n=1,325); we used this threshold in consultation with the SFRS to select fire incidents most likely to be bigger wildfires with significant impacts on seminatural habitats and filter out those wildfires that are more likely to be small incidents driven mainly by anti-social behaviour or relatively trivial cases of negligence with fire. Results of this analysis are presented in a consistent way at a) National level and b) Local Authority (LA) level to enable project results to be communicated to local staff in the SFRS and NatureScot. To facilitate an easier comparison of wildfire occurrence patterns in different LAs, we used cluster analysis to group LAs based on similar wildfire characteristics and presented respective patterns for the LA groupings identified by the analysis. More details on this are given in Section 5.4.

5.2 Patterns at National Level

5.2.1 Temporal – Seasonality

5.2.1.1 Annual

The number of yearly wildfire incidents (with complete records) in the full dataset ranged from 584 incidents in 2014 to 1,184 incidents in 2013 (Figure 5.1a). Overall, there are no indications of the presence of an increasing or decreasing trend over time in the number of wildfires occurring in Scotland based on the IRS records.

Regarding the subset of bigger wildfires, yearly fire incidents ranged from 50 incidents in 2020 to 134 incidents in 2010 (Figure 5.1b), with the exception of 2013 (Figure 5.1b) when 268 wildfires were recorded, most in the Highland and Na h-Eileanan Siar areas (125 and 83 fire incidents, respectively). This spike of wildfire activity was followed by three years of lower wildfire frequency up to 2016, while wildfire numbers increased again between 2017 to 2019 before dipping again in 2020 to just 50 big wildfires.

5.2.1.2 Monthly

Looking at seasonality at a monthly basis for all years, it is evident that most wildfires occur during the spring season: almost 29% of all wildfires occur in April, followed by 20% of wildfires in May and 15% in March (Figure 5.2a). A further 28% of wildfires occur in the summer season and in September, with small numbers occurring in late autumn and in the winter season. Occurrence of spring wildfires is much more dominant in the subset of bigger wildfires with around 81% of wildfires occurring in April (40.4%), March (22.3%) and May (20.3%) (Figure 5.2a). Moreover, a smaller proportion of bigger wildfires was recorded in summer months, especially in July and August when only 4.8% and 0.6%, respectively, of all bigger wildfires occurred compared to 9.7% in July and 4.2% in August of wildfires from the full dataset.

The same seasonal patterns emerge for the wildfires from the full dataset when looking at each year individually (Table 5.1). Most wildfires occur in the spring season, either in April or May. The only exception is year 2018 where most fires occurred in July and the second most fires in May; this could be explained by the prolonged period of above average temperature and below average precipitation that Scotland experienced from late spring to mid-summer that year, which could have caused an increased number of ignitions during this period.

Similarly for the subset of bigger wildfires, most wildfires occurred in April followed by wildfires occurring in May (Table 5.2). Looking at bigger wildfires occurring in the summer months, these accounted for just 2% of bigger fire incidents recorded in 2017 to 27% of fire incidents recorded in 2014.

Table 5.1 Yearly counts and monthly proportions of IRS wildfires for years 2009 to 2020.

Year	Sum	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2009	544	0	0	0	39	16	16	11	6	4	4	3	1
2010	1017	1	2	12	33	21	17	6	3	3	1	2	0
2011	733	1	1	18	35	21	6	8	2	1	2	3	1
2012	706	1	2	42	11	20	11	3	4	3	1	2	1
2013	1184	1	3	21	35	8	8	14	4	4	1	1	1
2014	584	3	1	12	28	10	10	14	5	9	5	3	1
2015	701	1	4	11	39	11	10	4	5	7	5	2	0
2016	778	0	1	17	21	28	15	3	4	2	4	2	1
2017	975	2	3	14	25	38	6	3	3	2	2	2	1
2018	1009	1	2	10	9	19	11	29	6	4	3	2	2
2019	671	4	6	7	47	14	5	6	3	3	3	1	1
2020	823	2	1	7	29	22	13	8	6	6	2	2	1

Table 5.2 Yearly counts and monthly proportions of IRS wildfires with damage area greater than 1,000 m ² for years 2009 to 2020.

Year	Sum	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2009	88	0	0	0	58	16	19	5	2	0	0	0	0
2010	134	0	4	13	51	21	7	1	1	1	1	0	0
2011	117	0	0	15	50	28	3	4	0	0	0	0	0
2012	112	0	1	52	7	21	14	3	1	1	0	0	0
2013	268	0	2	36	43	4	7	6	1	1	0	0	0
2014	73	0	0	16	42	11	8	18	1	0	0	3	0
2015	60	0	0	15	60	12	8	0	0	0	5	0	0
2016	61	0	0	16	39	30	11	0	2	0	2	0	0
2017	122	0	10	22	12	53	2	0	0	0	0	0	0
2018	128	0	2	30	21	22	6	16	0	2	1	1	0
2019	112	4	6	4	73	10	1	1	0	0	0	0	0
2020	50	0	0	8	38	42	6	0	0	6	0	0	0

5.2.2 Fuel types

5.2.2.1 Annual

Tables 5.3 and 5.4 give fire incident counts at Broad Habitat (BH) level classified from Level 3 Property Types for the full wildfire dataset and the subset of bigger fires, respectively, for years 2009 to 2020, while Tables 5.5 and 5.6 provide the same information for BHs classified from LCS88 classes. In addition, Figures 5.3 and 5.4 give the stacked proportions of individual BHs classified from Level 3 Property Types and LCS88 classes, for each year and wildfire dataset.

Investigation of Table 5.3 and Figure 5.3a. shows that there is limited inter-annual variability in the fuel time composition based on BHs classified from Level 3 Property Types. Wildfires burning grasslands are the most frequent for most years, followed by wildfires burning heathlands/shrubland. On the other hand, bigger wildfires burning heathland and grassland, based on Level 3 Property types, are the most frequent for all years and the proportion of fires on shrubland in relation to all recorded wildfires is greater than fires on grassland for all years (Table 5.4 and Figure 5.3b).

Table 5.3 Counts of IRS wildfires at Broad Habitat ( BH) level classified from Level 3 Property Types for years 2009 to 2020.

Year	Shrub land	Grass lands	Broad leaves	Conifers	Arable	Fresh water	Linear features	Built- up	Refuse	Other	None
2009	138	214	71	14	5	2	22	1	55	22	0
2010	275	377	135	50	11	1	52	11	75	30	0
2011	202	231	121	36	8	1	27	3	70	34	0
2012	222	239	76	25	11	0	33	5	57	38	0
2013	383	498	88	30	7	2	49	11	79	37	0
2014	164	194	64	35	6	1	22	2	61	35	0
2015	168	296	68	31	8	2	32	7	53	36	0
2016	200	340	76	30	14	0	34	11	51	22	0
2017	254	354	134	66	4	3	52	8	65	35	0
2018	284	351	119	74	8	0	40	6	75	52	0
2019	243	195	76	26	10	0	26	1	60	34	0
2020	157	273	138	76	11	2	36	4	86	39	1
Sum	2,690	3,562	1,166	493	103	14	425	70	787	414	1

Table 5.4 Counts of IRS wildfires with damage area greater than 1,000 m ² at Broad Habitat ( BH) level classified from Level 3 Property Types for years 2009 to 2020.

Year	Shrubland	Grasslands	Broadleaves	Conifers	Arable	Linear features	Refuse	Other
2009	41	35	6	5	0	0	0	1
2010	86	32	3	10	0	2	0	1
2011	76	22	7	7	1	0	2	2
2012	75	22	4	8	0	2	0	1
2013	169	71	10	10	2	3	2	1
2014	50	17	1	5	0	0	0	0
2015	35	19	3	3	0	0	0	0
2016	37	19	1	3	0	1	0	0
2017	78	29	7	7	0	1	0	0
2018	87	26	2	11	0	0	1	1
2019	93	13	0	6	0	0	0	0
2020	26	12	4	8	0	0	0	0
Sum	853	317	48	83	3	9	5	7

The picture is similar for fuel type composition when LCS88 classes are used to classify BHs (Table 5.5 and Figure 5.4a, full dataset), with grassland fires being the most frequent. But there is slightly more variability in fuel type composition regarding wildfires burning peatlands and bogs and heathland/shrubland in some years compared to BHs classified from Level 3 Property Types. Nevertheless, there is no evident trend in fuel type composition over the 2009 to 2020 period for the full wildfire dataset.

Regarding the subset of bigger wildfires, fires on shrubland are the most frequent for all years based on LCS88 classes (Table 5.6 and Figure 5.4b), but to a lesser degree than when BHs classified from Level 3 Property Types were used. When combined, wildfires on bogs and peatland and on shrubland outnumbered all grassland fires (on both seminatural and improved grass) from 2010 to 2014 and from 2017 to 2020, the opposite was observed in 2015, 2016 and 2020, while the same number of fires on bogs/peatlands and shrubland and on grasslands was recorded in 2009.

Table 5.5 Counts of IRS wildfires Broad Habitats ( BH) classified from LCS88 classes for years 2009 to 2020.

Year	Bogs & peatlands	Shrub land	Monta ne	Cliffs	Grass lands	Impr grass	Broad leaves	Mixed wood	Conifers	Arable	Fresh water	Built- up	No Data
2009	35	46	0	0	103	120	33	42	92	32	1	31	9
2010	74	92	1	0	198	239	38	84	149	52	2	76	12
2011	43	80	0	0	179	161	30	67	97	31	5	28	12
2012	67	89	0	2	143	173	42	66	75	14	1	28	6
2013	95	159	0	0	240	289	41	101	152	28	5	63	11
2014	40	76	0	2	101	137	18	42	107	23	1	31	6
2015	24	47	0	0	181	157	22	66	120	35	2	41	6
2016	41	52	0	0	190	193	26	79	104	23	2	55	13
2017	50	87	1	0	228	163	50	127	166	39	1	49	14
2018	47	113	0	1	187	217	48	92	211	39	2	42	10
2019	54	85	1	0	120	156	34	79	83	22	0	30	7
2020	34	47	0	1	136	183	35	105	193	30	2	39	18
Sum	604	973	3	6	2,006	2188	417	950	1,549	368	24	513	124

Table 5.6 Counts of IRS wildfires with damage area greater than 1,000 m ² at Broad Habitats ( BH) level classified from LCS88 classes for years 2009 to 2020.

Year	Bogs & peatlands	Shrub land	Monta ne	Cliffs	Grass lands	Impr grass	Broad leaves	Mixed wood	Conifers	Arable	Fresh water	Built- up	No Data
2009	12	23	0	0	12	23	10	2	4	0	0	0	2
2010	22	40	0	0	29	18	8	2	10	0	0	2	3
2011	14	36	0	0	28	20	7	3	7	0	0	2	0
2012	22	40	0	1	11	20	8	2	7	1	0	0	0
2013	46	88	0	0	47	53	6	8	14	0	2	2	2
2014	15	26	0	0	14	8	2	1	6	0	0	0	1
2015	10	13	0	0	18	13	0	1	5	0	0	0	0
2016	12	10	0	0	9	21	3	0	3	0	0	3	0
2017	20	32	0	0	25	15	8	7	14	1	0	0	0
2018	16	40	0	0	16	27	3	3	18	1	0	3	1
2019	20	39	1	0	15	21	5	3	7	0	0	1	0
2020	4	10	0	0	9	8	2	2	14	0	0	0	1
Sum	213	397	1	1	233	247	62	34	109	3	2	13	10

5.2.2.2 Monthly

Regarding fuel type composition at a monthly basis for the whole 2009-2020 period and based on BHs classified from IRS Level 3 Property Types (Table 5.7 and Figure 5.5a), most wildfires burning heathlands/shrubland and grasslands (both managed and seminatural) occurred in April (36% and 33% respectively), while most wildfires on broadleaves and conifers occurred in May (21% and 24%, respectively). Overall, 70-80% of all wildfires on shrubland and grasslands occurred in the spring season, while around 80% of wildfires on woodlands (broadleaves and conifers) occurred from late spring to the end of the summer season. On the other hand, wildfires from loose refuse and on wasteland show smaller seasonal variability with monthly proportions ranging from 7.5% to 15% for almost all months in the spring, summer, and autumn seasons.

Table 5.7 Monthly proportions (in %) of IRS wildfires per Broad Habitat ( BH) classified from Level 3 Property Types for the 2009-2020 period.

BH	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Shrubland	0.8	2.4	19.7	36.3	19.7	8.4	6.8	1.4	1.9	1.4	1.0	0.2
Grasslands	1.0	1.6	17.0	33.4	19.3	9.3	8.5	3.7	2.9	1.6	1.1	0.5
Broadleaves	1.1	1.8	6.9	18.8	21.3	16.6	14.6	7.1	6.3	3.0	1.8	0.8
Conifers	0.8	1.6	4.3	18.3	23.9	13.8	19.5	6.7	6.5	2.8	0.8	1.0
Arable	2.9	1.9	10.7	20.4	16.5	7.8	10.7	5.8	9.7	7.8	4.9	1.0
Freshwater	0.0	7.1	7.1	14.3	21.4	14.3	7.1	0.0	14.3	7.1	7.1	0.0
Linear features	1.2	4.0	11.8	24.2	20.5	12.9	8.2	5.2	4.0	4.0	1.9	2.1
Built-up	2.9	2.9	10.0	14.3	20.0	11.4	10.0	10.0	10.0	5.7	2.9	0.0
Refuse	4.4	3.4	9.5	15.1	15.2	10.2	10.3	7.9	7.5	5.3	7.4	3.7
Other	3.6	4.6	7.5	15.0	14.7	12.3	13.5	5.3	4.3	7.0	8.0	4.1
None	0.0	0.0	0.0	0.0	0.0	0.0	0.0	100.0	0.0	0.0	0.0	0.0

Table 5.8 Monthly proportions (in %) of IRS wildfires with damage area greater than 1,000 m ² per Broad Habitat ( BH) classified from Level 3 Property Types for the 2009-2020 period.

BH	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Shrubland	0.7	2.6	25.1	41.6	18.9	6.0	3.3	0.5	0.6	0.5	0.4	0.0
Grasslands	0.0	2.5	21.5	41.3	20.8	7.9	4.1	0.9	0.3	0.6	0.0	0.0
Broadleaves	0.0	0.0	14.6	29.2	25.0	16.7	12.5	2.1	0.0	0.0	0.0	0.0
Conifers	0.0	2.4	4.8	30.1	28.9	12.0	15.7	0.0	6.0	0.0	0.0	0.0
Arable	0.0	0.0	0.0	33.3	33.3	0.0	33.3	0.0	0.0	0.0	0.0	0.0
Linear features	0.0	0.0	22.2	11.1	33.3	22.2	11.1	0.0	0.0	0.0	0.0	0.0
Refuse	0.0	0.0	0.0	60.0	20.0	0.0	20.0	0.0	0.0	0.0	0.0	0.0
Other	0.0	0.0	0.0	71.4	14.3	14.3	0.0	0.0	0.0	0.0	0.0	0.0

Regarding fuel type composition at a monthly basis for the subset of bigger wildfires and based on BHs classified from IRS Level 3 Property Types (Table 5.8 and Figure 5.5b), again most wildfires burning heathlands/shrubland and grasslands occurred in April (42% and 41% respectively), but this was also the case for woodland/forest fires as most wildfires on broadleaves and conifers also occurred in April (29% and 30%, respectively). Overall, around 85% of all wildfires on shrubland and grasslands occurred in the spring season, while around 65% to 70% of wildfires on woodlands (broadleaves and conifers) occurred in the spring season and around 30% during summer months. There were only 24 bigger incidents with BHs other than shrubland, grasslands and woodlands, with most also occurring in the spring season.

Similar to BHs based on Level 3 Property Type, most wildfires burning on bogs and peatlands, heathlands/shrubland and seminatural grasslands for the 2009-2020 period, based on BHs classified from LCS88 classes, occurred in April and around 70%-75% of all wildfires for those fuel types occurred in the spring season (Table 5.9 and Figure 5.6a). The picture is also similar for wildfires on improved grasslands. For wildfires on woodlands (broadleaves, conifers and mixed), most wildfires (67%-75%) occurred from April to end of July. Looking at the subset of bigger wildfire incidents (Table 5.10 and Figure 5.6b), most wildfires burning on bogs and peatlands, heathlands/shrubland, seminatural and improved grasslands, broadleaves and mixed woodland occurred in April and around 76%-87% of all wildfires for those fuel types occurred in the spring season. Most conifer wildfires were recorded in May and a greater number of bigger wildfires on conifers occurred during the summer season (30% of all conifer fires) compared to wildfires on mixed woodland (24%) and on broadleaves (16%), indicating that there might be a greater risk of conifer forest fires in the summer compared to other woodland types.

Table 5.9 Monthly proportions (in %) of IRS wildfires per Broad Habitat ( BH) classified from LCS88 classes for the 2009-2020 period.

BH	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Bogs & peatlands	1.0	1.8	20.5	36.8	19.4	10.6	6.0	1.0	1.0	0.7	1.0	0.3
Shrubland	1.0	2.9	22.1	35.1	15.9	8.6	7.9	2.1	1.2	1.3	1.5	0.2
Montane	0.0	0.0	33.3	33.3	33.3	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Cliffs	0.0	16.7	50.0	0.0	0.0	0.0	33.3	0.0	0.0	0.0	0.0	0.0
Grasslands	1.5	2.1	16.7	31.7	20.8	8.6	6.8	4.0	3.2	2.0	1.8	0.8
Improved grass	1.6	3.1	16.2	29.4	19.1	9.4	7.5	3.9	3.1	2.7	3.0	1.2
Broadleaves	1.0	1.4	9.8	27.8	25.2	12.2	9.4	4.3	3.1	3.6	0.7	1.4
Mixed woodland	1.7	1.9	10.4	22.0	20.7	12.3	12.4	5.4	5.7	2.9	2.7	1.8
Conifers	1.2	1.4	5.0	19.9	19.1	15.2	17.9	6.9	7.5	3.3	1.7	0.8
Arable	2.4	2.7	14.1	25.5	18.2	8.7	11.7	2.7	6.3	3.8	2.4	1.4
Freshwater	0.0	0.0	8.3	33.3	12.5	12.5	8.3	12.5	0.0	8.3	4.2	0.0
Built-up	0.8	1.9	17.3	35.1	16.4	9.6	6.6	4.3	2.7	2.3	1.8	1.2
NoData	0.8	2.4	17.7	24.2	22.6	8.9	9.7	4.8	1.6	5.6	1.6	0.0

Table 5.10 Monthly proportions (in %) of IRS wildfires with damage area greater than 1,000 m ² per Broad Habitat ( BH) classified from LCS88 classes for the 2009-2020 period.

BH	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Bogs & peatlands	0.5	2.8	27.7	43.2	15.5	6.6	2.3	0.9	0.5	0.0	0.0	0.0
Shrubland	0.0	2.3	25.4	41.8	16.6	7.8	4.0	0.8	0.5	0.3	0.5	0.0
Montane	0.0	0.0	0.0	0.0	100.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Cliffs	0.0	0.0	100.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Grasslands	1.7	1.7	20.2	40.8	24.0	3.9	5.2	0.9	0.9	0.9	0.0	0.0
Improved grass	0.4	4.0	25.9	42.9	18.6	4.5	2.8	0.0	0.0	0.8	0.0	0.0
Broadleaves	0.0	1.6	9.7	43.5	29.0	12.9	3.2	0.0	0.0	0.0	0.0	0.0
Mixed woodland	0.0	0.0	8.8	35.3	32.4	8.8	11.8	2.9	0.0	0.0	0.0	0.0
Conifers	0.0	1.8	7.3	24.8	30.3	16.5	13.8	0.0	5.5	0.0	0.0	0.0
Arable	0.0	0.0	33.3	0.0	33.3	0.0	33.3	0.0	0.0	0.0	0.0	0.0
Freshwater	0.0	0.0	50.0	50.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Built-up	0.0	0.0	23.1	38.5	7.7	15.4	7.7	0.0	0.0	7.7	0.0	0.0
No Data	0.0	0.0	10.0	40.0	30.0	10.0	0.0	0.0	0.0	0.0	10.0	0.0

5.2.2.3 Daily

Tables 5.11 and 5.12 give the daily total counts for the 2009-2020 period of wildfires from the full dataset per Broad Habitats (BH) classified from Level 3 Property Types and LCS88 classes, respectively, while Tables 5.12 and 5.13 give the same information as above for bigger wildfires with damage area greater than 1,000 m² per. Data for the full wildfire dataset show that more wildfires occurred during weekends (1,641 on Saturdays and 1,517 on Sundays) and on Mondays (1,472), compared to the rest of the weekdays which had a similar number of recorded wildfires (1,256 – 1,311). However, there does not seem to be a strong pattern regarding fuel type composition in relation to the day of wildfire occurrence. Using both BH classifications for wildfires from the full dataset, there is an indication that wildfires on woodlands are generally more frequent during weekends, and this also applies for bogs and peatlands, shrublands and grasslands, but to a smaller extent. But overall, 60% to 70% of wildfires for the main fuel types (bogs, shrublands, grasslands and woodlands) occurred during weekdays, so there is no strong indication of an increased ignition risk during weekends due to, for example, recreational effects.

Table 5.11 Daily counts of IRS wildfires per Broad Habitat ( BH) classified from Level 3 Property Types for the 2009-2020 period.

BH	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Shrubland	428	375	366	359	349	430	383
Grasslands	552	477	470	449	468	585	561
Broadleaves	165	134	157	153	176	189	192
Conifers	68	62	50	48	80	90	95
Arable	19	9	16	11	14	12	22
Freshwater	1	1	2	2	2	5	1
Linear features	69	48	43	53	53	84	75
Built-up	15	7	6	12	4	13	13
Refuse	108	99	103	106	107	151	113
Other	47	44	56	66	58	81	62
None	0	0	0	0	0	1	0
Sum	1,472	1,256	1,269	1,259	1,311	1,641	1,517

Table 5.12 Daily counts of IRS wildfires per Broad Habitat ( BH) classified from LCS88 classes for the 2009-2020 period.

BH	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Bogs & peatlands	115	87	76	64	66	104	92
Shrubland	165	141	130	129	144	149	115
Montane	0	1	1	0	0	0	1
Cliffs	1	0	0	0	4	1	0
Grasslands	297	288	252	259	250	329	331
Improved grass	361	271	283	312	289	388	284
Broadleaves	53	45	68	51	56	67	77
Mixed woodland	143	112	128	131	116	175	145
Conifers	206	179	189	184	240	267	284
Arable	41	40	59	42	66	65	55
Freshwater	0	7	4	4	4	1	4
Built-up	76	66	63	66	61	81	100
NoData	14	19	16	17	15	14	29
Count	1,472	1,256	1,269	1,259	1,311	1,641	1,517

Results from the subset of bigger wildfires show a different picture than those from the full dataset, as more bigger wildfires occurred on Mondays (248) compared to weekends (205 on Saturdays and 153 on Sundays), while fire incident numbers ranged from 168 to 203 for the rest of the weekdays (Tables 5.13 and 5.14). Looking at results using BHs based on Level 3 Property types (Table 5.13), there does not seem to be a strong pattern regarding fuel type composition in relation to the day of wildfire occurrence. Using BHs classified using LCS88 classes, there is an indication that wildfires on bogs and peatlands and on conifer forests are slightly more frequent during weekends than compared to wildfires on heath and on grass, but overall and for both BH classifications 63% to 77% of bigger wildfires for the main fuel types (bogs, shrublands, grasslands and woodlands) occurred during weekdays.

Table 5.13 Daily counts of IRS wildfires with damage area greater than 1,000 m ² per Broad Habitat ( BH) classified from Level 3 Property Types for the 2009-2020 period.

BH	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Shrubland	172	135	127	111	99	120	89
Grasslands	49	53	34	38	43	58	42
Broadleaves	12	3	3	6	10	8	6
Conifers	12	8	9	12	11	17	14
Arable	1	1	0	0	0	0	1
Linear features	1	2	1	3	1	1	0
Refuse	0	1	0	1	2	1	0
Other	1	0	0	3	2	0	1
Sum	248	203	174	174	168	205	153

Table 5.14 Daily counts of IRS wildfires with damage area greater than 1,000 m ² per Broad Habitat ( BH) classified from LCS88 classes for the 2009-2020 period.

BH	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Bogs & peatlands	33	37	26	20	23	42	32
Shrubland	81	57	56	55	58	53	37
Montane	0	0	1	0	0	0	0
Cliffs	0	0	0	0	0	1	0
Grasslands	46	37	24	37	21	43	25
Improved grass	50	45	37	30	26	39	20
Broadleaves	8	8	14	7	9	6	10
Mixed woodland	11	0	4	8	5	1	5
Conifers	18	13	11	14	19	17	17
Arable	0	1	1	0	1	0	0
Freshwater	0	2	0	0	0	0	0
Built-up	1	1	0	3	2	3	3
No Data	0	2	0	0	4	0	4
Sum	248	203	174	174	168	205	153

5.2.3 Ignition motives

Based on IRS recorded information on ignition motives, most wildfires in the full dataset were caused by a deliberate ignition (n=5,560) compared to an accidental ignition (n=2,506) and unknown motive (n=1,659). On the other hand, most of the bigger wildfires were caused by accidental ignitions (n=615), followed by an unknown ignition motive (n=441) and deliberate ignitions (n=269).

Regarding temporal patterns, the proportions of wildfires for each ignition motive type recorded in the IRS (accidental, deliberate, and not known) appear to be quite similar for each day for both the full wildfire dataset (Table 5.15) and the subset of bigger wildfires (Table 5.16). Accidental wildfires from the full dataset accounted for 23% to 29% of daily fire incidents, while deliberate wildfires accounted for 55% to 60% of daily wildfires.

On the other hand, for bigger wildfires, accidental fire incidents were the most frequent for all days and accounted for 42% to 51% of daily fire incidents, while deliberate wildfires accounted for 17% to 24% of daily wildfires. Therefore, there does not seem to be a clear indication of an emerging daily pattern regarding the different motive types for wildfire ignition for both the full wildfire dataset and the subset of bigger wildfires.

Table 5.15 Daily counts and proportions (in %, brackets) of IRS wildfires per recorded motive.

Motive	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Accidental	387 (26%)	353 (28%)	329 (26%)	312 (25%)	374 (29%)	396 (24%)	355 (23%)
Deliberate	823 (56%)	685 (55%)	741 (58%)	731 (58%)	723 (55%)	950 (58%)	907 (60%)
Not known	262 (18%)	218 (17%)	199 (16%)	216 (17%)	214 (16%)	295 (18%)	255 (17%)
Sum	1,472	1,256	1,269	1,259	1,311	1,641	1,517

Table 5.16 Daily counts and proportions (in %, brackets) of IRS wildfires with damage area greater than 1,000 m ²per recorded motive.

Motive	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Accidental	113 (46%)	99 (49%)	85 (49%)	73 (42%)	86 (51%)	89 (43%)	70 (46%)
Deliberate	54 (22%)	35 (17%)	31 (18%)	41 (24%)	31 (18%)	46 (22%)	31 (20%)
Not known	81 (33%)	69 (34%)	58 (33%)	60 (34%)	51 (30%)	70 (34%)	52 (34%)
Sum	248	203	174	174	168	205	153

However, there are indications of patterns related to fuel type composition and ignition motive types. In particular, a greater proportion of wildfires from the full dataset on bogs and peatlands and shrublands was caused by accidental ignitions compared to deliberate ignitions (Tables 5.7a and 5.8a). On the other hand, deliberate ignitions were more frequent for seminatural and improved grasslands than accidental ones. At the same time, most of the wildfires with an unknown motive of ignition occurred on shrublands based on BHs classified using the Level 3 Property Types (Figure 5.7a). However, looking at BHs classified using the LCS88 classes (Figure 5.7b) shows that roughly the same numbers of fires on bogs and peatlands and shrublands and fires on grasslands (seminatural and improved) had an unknown ignition motive. In addition, most wildfires on loose refuse and wasteland had a deliberate ignition motive.

Similarly, looking at BHs of bigger wildfires based on Level 3 Property types, a greater proportion of shrubland fires had an accidental ignition motive than a deliberate one, while most of the wildfires with an unknown motive of ignition occurred on shrublands (Figure 5.7b). A clearer pattern emerges when looking at BHs based on LCS88 classes (Figure 5.8b); it is evident that more bigger wildfires on bogs and peatlands and on heathland/shrubland were caused by accidental than deliberate ignitions, while wildfires on grassland accounted for most fire incidents with deliberate ignitions.

Overall, these results indicate that wildfires on more remote or less accessible seminatural habitats, such as peatlands and heathlands, are more likely to be caused by accidental ignitions, compared to wildfires occurring in more accessible and/or managed land where they are more likely to be caused by deliberate ignitions.

In addition, we investigated patterns of causes of ignition, ignition sources, first items ignited and items causing spread in relation to ignition motive (Tables 5.17-5.24). This information was available for 919 wildfire incidents from the full dataset and 232 fire incidents from the subset of bigger wildfires. Regarding the cause of ignition, bringing together a heat source and combustibles was almost exclusively the cause of deliberate ignitions (>90%), for all fire incidents irrespective of damage area size (Tables 5.17 and 5.18). On the other hand, most accidental wildfires were caused by bonfires or some other form of intentional burn that got out of control; 24% and 27% of wildfires from the full dataset and 26% and 33% of the bigger wildfires, respectively. Regarding wildfires of unknown motive, most had an ignition cause of "Other", followed by overheating of unknown cause and some form of intentional burn that got out of control (Tables 5.17
and 5.18).

There was limited information to extract regarding ignition sources, since an unknown ignition source was recorded for most fire incidents from the full dataset and the subset of bigger wildfires, regardless of ignition motive (Tables 5.19 and 5.20). Other recorded ignition sources included match and candles, naked flame and smoking-related ignitions, that each accounted from 12%-24% of wildfires from the full dataset and 9% to 19% of the bigger wildfires. Interestingly, 3%-4% of wildfires (from the subset and full dataset, respectively) were attributed to natural occurrence; however no further information was provided for those fire incidents.

Vegetation was recorded as the item or material first ignited for 709 of the 919 wildfires from the full dataset and 204 of the 232 bigger wildfires (Tables 5.21 and 5.22). Looking at the main types of vegetation recorded (given in field "ItemFirstIgnitedSubtype") for the wildfires from the full dataset, there were 383 records for grassland, heath and scrub, 199 records for trees, 49 records for leaves and 34 records for straw/stubble, while grassland, heath and scrub dominated the subset of bigger wildfires (178 records), followed by 20 records for trees. More wildfires from the full dataset with grass, heath or scrub as the recorded vegetation type of ignition were caused by an accident (153 records) than by deliberate action (125 records) or unknown motive (105 records), whilst most wildfires with trees and straw/stubble as recorded ignition materials were caused by deliberate action (118 and 21 records, respectively) than by accident (35 and 9 records, respectively). On the other hand for bigger wildfires, accidental ignitions greatly outnumbered deliberate ignitions for fire incidents with grass, heath or scrub as the recorded vegetation type of ignition (94 records for accidental ignitions compared to 23 records for deliberate ignitions).

Vegetation was also recorded as the most frequent item or material causing fire spread for 731 of the 919 wildfires from the full dataset and 222 of the 232 bigger wildfires (Tables 5.23 and 5.24). Again, investigating the main types of vegetation recorded (given in field "ItemCausingSpreadSubtype") for the wildfires from the full dataset showed that there were 404 records for grassland, heath and scrub, 209 records for trees and 33 records for straw/stubble, while grassland, heath and scrub dominated the subset of bigger wildfires (192 records), followed by 25 records for trees. Looking different vegetation types causing spread in relation to ignition motives revealed the same patters as in the case of items causing ignition. More wildfires from the full dataset with grass, heath or scrub as the recorded vegetation type causing spread had an accidental ignition motive (162 records) compared to a deliberate motive (127 records) or unknown motive (115 records), whilst most wildfires with trees and straw/stubble as recorded materials responsible for fire spread were caused by deliberate action (127 and 23 records, respectively) than by accident (37 and 7 records, respectively). As expected for bigger wildfires, accidental ignitions greatly outnumbered deliberate ignitions for fire incidents with grass, heath or scrub as the recorded vegetation type of ignition (101 records for accidental ignitions compared to 64 records for deliberate ignitions).

Although results given in Tables 5.17 to 5.24 are based on the analysis of a limited number of wildfire incidents, they give a clear indication that types of seminatural vegetation (grass, heath, scrub) are more likely to be responsible for ignition and fire spread in fires caused by accident, especially in the bigger wildfires where almost all records had grass, heath, and scrub as the vegetation types responsible for ignition and fire spread. On the other hand, trees and to a lesser degree straw/stubble are more likely to be the most frequent vegetation type responsible for ignition and fire spread in, mainly smaller and less trivial, fires caused by deliberate ignitions.

Table 5.17 Total counts and proportions of cause of ignition per motive of IRS wildfires.

Motive	Sum	Accumulation flammable material	Bonfire out of control	Careless handling- careless disposal	Combustible articles close to heat source	Coo king	Faulty fuel supplies- Gas/ Electricity/ Petrol	Heat source + combustibles brought together	Natural occurrence	Negligent use of equipment	Other	Other intentional burn, out of control	Over heating, unknown cause	Playing with fire (or heat source)
Accidental	248	2	24	13	3	1	0	0	6	3	14	27	5	2
Deliberate	454	0	0	0	0	0	0	96	0	0	4	0	0	0
Not known	217	3	7	2	6	1	0	0	11	0	35	15	17	3

Table 5.18 Total counts and proportions of cause of ignition per motive of IRS wildfires with damage area greater than 1,000 m ².

Motive	Sum	Accumulation flammable material	Bonfire out of control	Careless handling- careless disposal	Combustible articles close to heat source	Coo king	Faulty fuel supplies- Gas/ Electricity/ Petrol	Heat source + combustibles brought together	Natural occurrence	Negligent use of equipment	Other	Other intentional burn, out of control	Over heating unknown cause	Playing with fire (or heat source)
Accidental	117	1	26	11	3	1	1	0	4	1	15	33	4	0
Deliberate	37	0	0	0	0	0	0	92	0	0	8	0	0	0
Not known	78	0	4	1	5	3	0	0	9	0	41	15	21	1

Table 5.19 Total counts and proportions of ignition sources per motive of IRS wildfires.

Motive	Sum	Bombs and explosives	Cooking appliance	Electricity supply	Fuel/ chemical related	Industrial Equipment	Matches and candles	Naked flame	Natural Occurrence	Not known	Other	Smoking Related	Spread from secondary fire	Vehicles only	Wet hay
Accidental	250	2	5	1	0	0	14	12	4	29	8	12	10	2	1
Deliberate	456	0	0	0	2	0	26	18	0	40	0	14	0	0	0
Not known	221	0	1	0	0	0	8	2	5	77	4	3	0	0	0

Table 5.20 Total counts and proportions of ignition sources per motive of IRS wildfires with damage area greater than 1,000 m ².

Motive	Sum	Bombs and explosives	Cooking appliance	Electricity supply	Industrial equipment	Matches and candles	Naked flame	Natural occurrence	Not known	Other	Smoking Related	Spread from secondary fire	Vehicles only
Accidental	118	4	2	1	2	19	13	3	32	10	9	5	0
Deliberate	38	0	0	0	0	29	26	0	42	0	3	0	0
Not known	78	0	4	0	0	5	1	3	79	3	4	0	1

Table 5.21 Total counts and proportions of first item ignited per motive of IRS wildfires.

Motive	Sum	None	Not known	Other	Paper/Cardboard	Rubbish/Waste/Recycling	Vegetation	Wood
Accidental	250	2	7	1	0	2	86	2
Deliberate	456	3	14	0	5	4	70	4
Not known	221	1	16	0	0	0	79	4

Table 5.22 Total counts and proportions of first item ignited per motive of IRS wildfires with damage area greater than 1,000 m ².

Motive	Sum	Not known	Other	Paper/Cardboard	Rubbish/Waste/Recycling	Vegetation	Wood
Accidental	118	8	1	0	3	87	1
Deliberate	38	11	0	3	0	86	0
Not known	78	13	0	0	1	86	0

Table 5.23 Total counts and proportions of items causing fire spread per motive of IRS wildfires.

Motive	Sum	Foam, rubber, plastic	None	Not known	Other	Paper/ Cardboard	Rubbish/ Waste/Recycling	Vegetation	Wood
Accidental	250	0	6	2	2	0	0	89	1
Deliberate	456	1	12	6	0	1	2	72	6
Not known	221	0	9	6	0	0	0	83	2

Table 5.24 Daily counts and proportions of items causing spread per motive of IRS wildfires with damage area greater than 1,000 m ².

Motive	Sum	None	Not known	Vegetation
Accidental	118	2	3	95
Deliberate	38	3	5	92
Not known	78	1	3	96

5.2.4 Burnt area size

The purpose of this analysis was to investigate patterns of burnt area size related to fuel type composition, i.e., investigate whether certain fuel types cause more burnt areas and greater wildfires than others. For this reason, we aggregated the thirteen (13) classes of the Outdoor Damage Area field recorded in the IRS (see Section 2) into four (4) classes:

a) None: Wildfires with no burn area estimates

b) DA0_100: Wildfires with burnt areas of 0-100 m²;

c) DA0100_1000: Wildfires with burnt areas of 100-1,000 m²; and

d) DA>1000: Wildfires with burnt areas >1,000 m² (this is the same as the subset of bigger wildfires described in previous Sections).

Overall, most wildfires (67%) were small wildfires with estimated burnt areas less than 100 m², while wildfires of moderate size (100 – 1,000 m²) and bigger wildfires (>1,000 m²) accounted for 15% and 14% of all wildfires, respectively. Regarding fuel types, wildfires in almost all Broad Habitats (BH) are small with estimated burnt areas less than 100 m², with the clear exception of bogs and peatlands and shrubland. Based on Broad Habitats BHs classified from IRS Level 3 Property Types, small wildfires with less than 100 m² burnt areas accounted for around 60% to 90% of all wildfires for each BH, with small wildfires being the most frequent for shrubland (43%) but to a smaller extent compared to the other BHs, while bigger wildfires accounted for 32% of shrubland wildfires (Table 5.25 and Figure 5.9a).

Table 5.25 Counts of IRS wildfire Broad Habitats ( BHs) classified from Level 3 Property Types per recorded Damage Area ( DA) group: None; 0-100 m ²; 100-1,000 m ² and >1,000 m ².

BH	None	DA0_100	DA100_1000	DA>1000
Shrubland	57	1158	622	853
Grasslands	145	2522	578	317
Broadleaves	49	971	98	48
Conifers	11	329	70	83
Arable	7	86	7	3
Freshwater	1	13	0	0
Linear features	25	355	36	9
Built-up	10	57	3	0
Refuse	83	681	18	5
Other	67	320	20	7
None	0	1	0	0

Regarding wildfire BHs classified from LCS88 classes, bigger wildfires (burnt area >1,000 m²) accounted for 35% of wildfires on bogs and peatlands and 41% of shrubland wildfires, compared to small wildfires (burnt area <100 m²) that accounted for 41% and 34% of all wildfires, respectively (Table 5.26 and Figure 5.9b). Bigger wildfires accounted for only 11%-12% of wildfires on seminatural and improved grasslands, while most broadleaved, mixed and conifer woodland fires were small wildfires (68%, 82% and 77%, respectively).

Table 5.26 Counts of IRS wildfire Broad Habitats ( BHs) classified from LCS88 classes per recorded Damage Area ( DA): 0-100 m ²; 100-1,000 m ² and >1,000 m ².

BH	None	DA0_100	DA100_1000	DA>1000
Bogs & peatlands	16	248	127	213
Shrubland	22	329	225	397
Montane	0	1	1	1
Cliffs	1	1	3	1
Grasslands	80	1321	372	233
Improved grass	140	1504	297	247
Broadleaves	25	282	48	62
Mixed woodland	51	780	85	34
Conifers	64	1188	188	109
Arable	16	311	38	3
Freshwater	2	14	6	2
Built-up	32	427	41	13
No Data	6	87	21	10

Overall, these results indicate that wildfires on peatlands and heathlands tend to be larger in burnt area size and cause greater damage (based on recorded damage area estimates) than wildfires in other seminatural vegetation such as grasslands and woodlands, which are more likely to be small fires causing limited damage.

5.2.5 Urban-Rural classification

Accessibility and population density and proximity to populated areas are established factors influencing wildfire ignitions and wildfire risk. For this reason, we used the 8-fold Scottish Government Urban Rural (UR) Classification (Table 5.27) to investigate, in an integrated way, patterns of wildfire occurrence related to accessibility and population numbers.

Table 5.27 Description of the 8-fold Urban Rural ( UR) classification system.

UR Folds	Description	UR Folds	Description
Fold01	Large Urban Areas	Fold05	Very Remote Small Towns
Fold02	Other Urban Areas	Fold06	Accessible Rural
Fold03	Accessible Small Towns	Fold07	Remote Rural
Fold04	Remote Small Towns	Fold08	Very Remote Rural

Figure 5.10a shows that most IRS wildfires occurred within accessible rural areas (57%), followed by wildfires in very remote rural areas (22%) and in remote rural areas (14%). Only 5% of wildfires occurred within large urban and other urban areas (477 wildfire incidents in total). In contrast, most of the 1,325 bigger wildfires occurred in very remote rural areas (58%) and then in accessible rural areas (24%) and remote rural areas (15%), while only six (6) bigger wildfires were recorded in urban areas (Folds 1 and 2) and just nine (9) wildfires in small towns (one (1) in accessible small towns and eight (8) in very remote small towns) (Figure 5.10b).

Regarding fuel type composition, most grassland and woodland wildfires occurred in accessible rural areas for Broad Habitats (BH) classified from both the IRS Level 3 Property Types (Figure 5.11a) and LCS88 classes (Figure 5.12a). However, most wildfires on bogs and peatlands and heathlands/shrubland occurred in very remote rural areas, which shows a clear indication that these wildfires that tend to be larger in burnt area size and of accidental motive are more likely to occur in areas with very low population density and with very limited accessibility.

This is further supported when looking at the subset of bigger wildfires, where a greater proportion of wildfires on bogs and peatlands and on heathlands occurred in very remote areas compared to when all wildfires where investigated (Figures 5.11b and 5.12b). In addition, a greater proportion of bigger grassland wildfires occurred in very remote areas than accessible rural areas when BHs based on Level 3 Property types were used (Figure 11b); this is also true for bigger wildfires on managed or improved grasslands using BHs based on LCS88 classes (Figure 12b).

Looking at bigger woodland fires based on LCS88 classes (Figure 12b), it is evident that most wildfires on broadleaf woodlands occurred in remote rural areas (61%) with only 10% of wildfires on broadleaves occurring in accessible rural areas. On the other hand, most wildfires on mixed woodland occurred in very remote areas (53%) and in accessible rural areas (32%). Finally, most bigger conifer forest fires occurred in accessible rural areas, with an almost even number of conifer forest fires occurring in remote and very remote rural areas (21% and 19%, respectively). These results indicate that bigger wildfires on broadleaves and mixed woodland are more likely to occur in remote and very remote rural areas compared to wildfires of smaller damage area size on the same vegetation/fuel types that are more likely to occur in more accessible rural areas.

5.3 Patterns per Local Authorities

5.3.1 Overview

Wildfire data from both the full wildfire dataset and the subset of bigger wildfires are presented for each Local Authority (LA) using tables of the total count of wildfires per LA along with the proportions of those wildfire counts falling into the different time periods (years, months, days), fuel types, ignition motive types and burnt area sizes. Results at LA level are presented using tables because the large number of LAs made generated figures difficult to read and interpret (LA n=32 for the full wildfire dataset and n=31 for the bigger wildfires; no bigger wildfire incident recorded in East Renfrewshire).

5.3.2 Temporal - Seasonality

5.3.2.1 Annual

Most wildfires from the full dataset occurred in the Highland area (1,852), followed by the North Lanarkshire (690) and West Lothian (684) areas, while Clackmannanshire, Angus and East Renfrewshire areas had the least recorded wildfires (79, 54, 35 incidents, respectively) (Table 5.28). For most LAs, the greater annual number of wildfires within their areas occurred during the 2016 to 2020 period, but there was no clear indication of any trend of inter-annual wildfire occurrence.

Table 5.28 Total counts and yearly proportions (in %) of IRS wildfires per Local Authority ( LA) for the 2009-2020 period.

LA	Sum	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Aberdeen City	125	6	6	10	4	6	2	7	2	10	21	11	14
Aberdeenshire	281	5	6	9	5	10	6	9	5	9	11	11	16
Angus	54	7	7	11	2	7	7	6	6	13	7	9	17
Argyll & Bute	355	5	19	9	9	14	3	5	8	7	7	10	5
Clackmannanshire	79	9	8	3	8	11	9	17	3	24	6	0	4
Dumfries & Galloway	177	6	19	14	6	9	3	5	6	9	10	3	12
Dundee City	150	1	5	9	4	4	5	13	11	16	17	6	9
East Ayrshire	609	3	15	5	6	17	4	8	10	14	8	4	5
East Dunbartonshire	81	4	9	9	6	7	9	16	5	11	9	5	11
East Lothian	154	4	9	8	3	7	2	12	6	18	11	12	8
East Renfrewshire	35	3	3	3	17	26	3	3	11	11	6	9	6
Edinburgh City	200	10	13	7	7	13	6	10	7	6	12	5	8
Falkirk	357	5	15	6	3	9	9	6	13	12	9	3	10
Fife	414	13	12	7	6	9	5	10	8	7	13	4	8
Glasgow City	198	3	7	9	4	11	10	13	15	11	5	9	4
Highland	1852	7	10	8	9	15	7	5	5	8	12	9	7
Inverclyde	245	0	2	6	9	9	5	9	25	18	7	4	7
Midlothian	202	10	8	9	6	7	5	7	6	16	5	10	11
Moray	323	6	7	8	6	7	7	6	7	4	19	10	13
Na h-Eileanan Siar	644	3	7	6	15	23	6	5	6	6	14	5	5
North Ayrshire	187	2	6	6	10	9	4	8	9	17	16	10	5
North Lanarkshire	690	4	8	9	8	14	7	10	11	9	6	7	8
Orkney Islands	101	11	11	5	5	13	6	5	7	6	16	9	7
Perth & Kinross	182	9	13	6	7	9	9	7	10	13	9	4	7
Renfrewshire	195	4	3	9	7	10	7	11	14	14	9	3	10
Scottish Borders	257	7	15	11	4	7	6	6	4	16	8	8	8
Shetland Islands	83	6	18	6	6	8	4	5	4	6	21	2	15
South Ayrshire	87	2	17	8	9	5	8	6	6	8	16	3	12
South Lanarkshire	368	7	9	8	4	9	7	10	12	10	4	8	12
Stirling	211	10	13	7	3	5	3	7	6	11	13	7	17
West Dunbartonshire	145	2	8	8	2	21	8	8	12	13	6	4	6
West Lothian	684	6	16	6	7	9	6	6	6	9	11	8	11

Most bigger wildfires occurred in the Highland and Na h-Eileanan Siar areas (656 and 251 fire incidents, respectively) (Table 5.29). There is an indication that a greater proportion of bigger wildfires occurred in 2013 when most fire incidents were recorded for the Na h-Eileanan Siar (33%) and Highland (19%) LAs, which combined accounted for 208 of the 268 wildfires recorded in 2013.

Table 5.29 Total counts and yearly proportions (in %) of IRS wildfires with damage area greater than 1,000 m ² per Local Authority ( LA) for the 2009-2020 period.

LA	Sum	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Aberdeen City	3	0	0	33	0	0	0	0	0	0	67	0	0
Aberdeenshire	27	4	4	7	15	19	11	0	4	7	11	7	11
Angus	4	0	0	0	0	0	0	0	0	25	0	50	25
Argyll & Bute	59	0	27	19	5	17	2	3	2	10	5	8	2
Clackmannanshire	7	0	14	14	14	0	0	43	0	15	0	0	0
Dumfries & Galloway	33	9	21	21	15	9	0	6	9	3	0	0	6
Dundee City	2	0	0	0	0	50	0	0	0	0	50	0	0
East Ayrshire	30	0	4	13	0	33	7	0	3	33	0	7	0
East Dunbartonshire	4	0	25	0	0	0	25	50	0	0	0	0	0
East Lothian	5	0	0	0	0	0	0	0	0	0	0	100	0
Edinburgh City	2	0	0	0	0	0	0	0	0	0	50	0	50
Falkirk	7	0	14	0	0	0	29	0	0	14	14	0	29
Fife	17	6	0	6	6	29	0	6	6	6	12	12	12
Glasgow City	2	0	0	50	0	0	0	0	0	0	0	50	0
Highland	656	9	10	8	9	19	6	3	4	9	10	10	3
Inverclyde	8	0	0	13	0	0	0	13	25	38	0	0	13
Midlothian	4	25	0	25	25	0	0	25	0	0	0	0	0
Moray	25	0	4	0	12	8	8	4	0	8	12	24	20
Na h-Eileanan Siar	251	4	7	6	11	33	5	4	4	5	13	7	1
North Ayrshire	2	0	0	0	0	0	0	0	0	0	100	0	0
North Lanarkshire	26	0	4	12	0	12	0	19	4	23	8	4	15
Orkney Islands	11	9	27	0	0	45	0	0	9	0	9	0	0
Perth & Kinross	13	8	23	8	15	15	8	0	8	15	0	0	0
Renfrewshire	3	0	0	0	0	0	0	0	67	33	0	0	0
Scottish Borders	22	9	23	23	9	5	0	0	0	5	14	14	0
Shetland Islands	11	0	9	9	0	45	0	0	9	9	18	0	0
South Ayrshire	7	0	14	14	14	0	0	0	0	14	14	0	29
South Lanarkshire	16	0	13	6	6	6	6	13	13	6	0	13	19
Stirling	15	13	13	7	0	7	13	20	0	20	0	0	7
West Dunbartonshire	14	0	0	14	0	14	14	0	29	21	0	7	0
West Lothian	39	13	8	5	10	10	10	15	10	5	5	0	8

5.3.2.2 Monthly

The greater monthly counts of wildfires for the 2009-2020 period and for all LAs were recorded in one of the spring months, with the only exception of Moray where more wildfires were recorded for July than in other months (Table 5.30). April was the month with the most recorded wildfires for most LAs, followed by May. Overall, spring wildfires outnumbered summer wildfires in all LAs, apart from in the Moray and Falkirk areas, while an almost equal number of spring and summer wildfires was recorded in the Shetlands Islands.

Table 5.30 Total counts and monthly proportions (%) of IRS wildfires per Local Authority ( LA) for the 2009-2020 period.

LA	Sum	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sept	Oct	Nov	Dec
Aberdeen City	125	5	7	6	12	18	12	14	6	12	2	4	1
Aberdeenshire	281	2	2	13	20	17	10	12	9	9	3	2	2
Angus	54	2	6	11	7	26	9	6	9	6	2	13	4
Argyll & Bute	355	1	2	17	34	25	8	7	1	2	1	1	1
Clackmannanshire	79	0	3	8	22	29	8	13	8	11	0	0	0
Dumfries & Galloway	177	1	2	14	31	22	10	6	4	2	5	2	2
Dundee City	150	3	2	25	27	17	13	7	1	1	3	0	1
East Ayrshire	609	1	1	18	39	18	7	6	2	2	1	2	1
East Dunbartonshire	81	4	1	9	33	23	7	7	2	4	2	5	1
East Lothian	154	6	6	8	23	17	14	10	3	2	5	1	4
East Renfrewshire	35	0	3	11	26	34	6	9	0	0	6	3	3
Edinburgh City	200	0	4	8	22	16	9	22	4	7	5	4	3
Falkirk	357	1	1	6	18	19	18	16	9	6	4	2	0
Fife	414	1	2	8	23	16	17	15	6	5	5	2	0
Glasgow City	198	1	2	17	46	17	5	4	2	4	2	2	0
Highland	1852	1	2	17	33	17	10	10	4	3	2	1	0
Inverclyde	245	0	0	16	34	31	11	4	0	1	1	1	0
Midlothian	202	2	3	10	23	24	7	8	4	10	2	3	1
Moray	323	1	2	5	16	14	14	22	8	11	5	2	1
Na h-Eileanan Siar	644	1	3	30	30	13	10	6	2	1	1	2	0
North Ayrshire	187	1	4	9	25	22	11	14	6	3	2	3	1
North Lanarkshire	690	1	2	20	37	19	7	7	2	2	2	1	1
Orkney Islands	101	1	2	13	26	20	6	10	8	2	4	5	4
Perth & Kinross	182	2	3	8	27	21	13	10	5	7	2	2	2
Renfrewshire	195	1	5	13	24	27	6	8	2	5	4	3	3
Scottish Borders	257	2	2	10	21	18	15	9	9	7	2	4	2
Shetland Islands	83	0	5	20	13	7	23	7	10	4	2	4	5
South Ayrshire	87	2	2	11	13	24	16	9	5	7	5	3	2
South Lanarkshire	368	3	3	11	26	21	12	7	5	4	4	4	1
Stirling	211	1	1	8	30	22	11	13	5	2	3	2	1
West Dunbartonshire	145	1	1	15	34	19	13	6	5	3	2	1	1
West Lothian	684	2	1	11	28	26	10	10	4	3	2	2	1

Regarding the subset of bigger wildfires, greater monthly counts of wildfires for most LAs were recorded in one of the spring months (Table 5.31). For example, 43% and 39% of wildfires recorded for the Highland and Na h-Eileanan Siar areas, respectively, occurred in April, while April was the month with the most recorded wildfires for most LAs, followed by May. Overall, bigger spring wildfires outnumbered summer wildfires in almost all LAs and in those LAs with the greatest numbers of wildfire occurrence.

Table 5.31 Total counts and monthly proportions (in %) of IRS wildfires with damage area greater than 1,000 m ² per Local Authority ( LA) for the 2009-2020 period.

LA	Sum	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Aberdeen City	3	0	0	0	0	33	0	33	0	33	0	0	0
Aberdeenshire	27	0	0	26	19	15	15	19	4	4	0	0	0
Angus	4	0	0	25	25	50	0	0	0	0	0	0	0
Argyll & Bute	59	0	2	29	37	29	3	0	0	0	0	0	0
Clackmannanshire	7	0	0	29	57	14	0	0	0	0	0	0	0
Dumfries & Galloway	33	0	3	21	52	18	6	0	0	0	0	0	0
Dundee City	2	0	0	0	50	0	0	50	0	0	0	0	0
East Ayrshire	30	0	0	13	50	30	7	0	0	0	0	0	0
East Dunbartonshire	4	0	0	0	50	50	0	0	0	0	0	0	0
East Lothian	5	80	20	0	0	0	0	0	0	0	0	0	0
Edinburgh City	2	0	0	0	50	0	0	50	0	0	0	0	0
Falkirk	7	0	0	0	29	43	0	29	0	0	0	0	0
Fife	17	0	6	12	24	18	18	18	0	0	6	0	0
Glasgow City	2	0	0	0	50	0	0	50	0	0	0	0	0
Highland	656	0	2	20	43	20	8	4	1	1	0	0	0
Inverclyde	8	0	0	0	38	63	0	0	0	0	0	0	0
Midlothian	4	0	0	25	25	25	25	0	0	0	0	0	0
Moray	25	0	0	8	20	28	20	12	0	12	0	0	0
Na h-Eileanan Siar	251	0	5	35	39	11	6	3	1	0	0	0	0
North Ayrshire	2	0	0	0	0	0	0	100	0	0	0	0	0
North Lanarkshire	26	0	0	4	38	54	4	0	0	0	0	0	0
Orkney Islands	11	0	0	9	55	18	0	9	0	0	9	0	0
Perth & Kinross	13	0	0	15	38	31	0	8	0	8	0	0	0
Renfrewshire	3	0	0	0	33	67	0	0	0	0	0	0	0
Scottish Borders	22	0	5	36	32	18	0	5	5	0	0	0	0
Shetland Islands	11	0	0	64	27	9	0	0	0	0	0	0	0
South Ayrshire	7	0	0	43	14	43	0	0	0	0	0	0	0
South Lanarkshire	16	0	0	13	44	25	19	0	0	0	0	0	0
Stirling	15	0	0	0	60	33	7	0	0	0	0	0	0
West Dunbartonshire	14	0	0	14	57	14	7	7	0	0	0	0	0
West Lothian	39	0	0	18	33	31	10	8	0	0	0	0	0

5.3.2.3 Daily

The greater daily counts of wildfires for the 2009-2020 period and for all LAs were recorded during weekends (either on a Saturday or a Sunday) (Table 5.32). However, in most cases the differences between daily wildfire counts were small. Overall, total wildfires occurring during weekdays outnumbered wildfires occurring at the weekend in all LAs. The greater numbers of weekday wildfires were recorded in the Dundee City, Aberdeen City and Na h-Eileanan Siar areas, where weekday wildfires accounted for 80%, 73% and 72% of the total number of wildfires. On the other hand, weekday wildfires accounted for 59% and 58% in the Shetlands and Orkney Islands, respectively, but accounted for 60% and more of total wildfires in all other LAs.

Table 5.32 Total counts and daily proportions of IRS wildfires per Local Authority ( LA) for the 2009-2020 period.

LA	Sum	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Aberdeen City	125	13	13	16	17	14	17	10
Aberdeenshire	281	16	12	12	11	12	19	17
Angus	54	9	24	15	7	13	17	15
Argyll & Bute	355	19	12	12	13	14	14	15
Clackmannanshire	79	14	14	11	15	10	14	22
Dumfries & Galloway	177	17	10	12	14	14	14	20
Dundee City	150	17	13	21	15	13	11	9
East Ayrshire	609	13	14	14	13	13	18	16
East Dunbartonshire	81	14	14	10	9	21	21	12
East Lothian	154	12	16	15	13	12	19	14
East Renfrewshire	35	20	20	6	6	17	11	20
Edinburgh City	200	15	12	13	9	11	21	20
Falkirk	357	16	16	11	13	12	16	16
Fife	414	15	11	13	14	12	16	19
Glasgow City	198	13	11	14	11	16	16	20
Highland	1852	14	14	14	14	15	17	12
Inverclyde	245	12	12	16	17	13	14	16
Midlothian	202	14	7	12	14	17	17	18
Moray	323	14	14	11	13	16	14	18
Na h-Eileanan Siar	644	19	16	12	13	11	17	11
North Ayrshire	187	16	12	11	12	11	23	15
North Lanarkshire	690	17	11	12	13	11	18	17
Orkney Islands	101	12	8	15	14	10	18	24
Perth & Kinross	182	14	10	17	5	14	23	16
Renfrewshire	195	16	13	14	12	17	15	13
Scottish Borders	257	11	14	11	12	18	16	19
Shetland Islands	83	12	8	10	11	18	20	20
South Ayrshire	87	17	2	16	17	7	21	20
South Lanarkshire	368	15	13	14	15	12	17	16
Stirling	211	16	13	11	15	9	19	17
West Dunbartonshire	145	12	12	16	14	10	16	20
West Lothian	684	16	14	11	12	14	16	17

Regarding bigger wildfires, fire incidents occurring during weekdays outnumbered wildfires occurring at the weekend in 25 of the 31 LAs, with three (3) LAs having the same wildfire occurrence between weekdays and weekends and another three (3) LAs having more wildfires recorded in weekends than during the week (Table 5.32). However, these six (6) LAs where weekend wildfires were the same or outnumbered fires during weekdays had only 26 (out of a total 1,325 fires) incidents of bigger wildfires. Overall, weekday wildfires accounted for 60% and more of total wildfires in most LAs. Looking at LAs with the greater numbers of bigger wildfires (>30 incidents), weekday wildfires accounted for 80%, 75% and 75% of the total number of bigger wildfires in the Argyll and Bute, Highland and Na h-Eileanan Siar areas, respectively.

Table 5.33 Total counts and daily proportions (in %) of IRS wildfires with damage area greater than 1,000 m ² per Local Authority ( LA) for the 2009-2020 period.

LA	Sum	Monday	Tuesday	Wednesday	Thursday	Friday	Saturday	Sunday
Aberdeen City	3	0	0	33	33	0	33	0
Aberdeenshire	27	22	11	19	15	4	22	7
Angus	4	25	0	0	0	50	0	25
Argyll & Bute	59	24	15	12	10	19	10	10
Clackmannanshire	7	29	29	14	14	0	14	0
Dumfries & Galloway	33	24	12	12	9	6	15	21
Dundee City	2	0	0	0	0	50	0	50
East Ayrshire	30	13	23	7	0	17	23	17
East Dunbartonshire	4	25	0	25	0	50	0	0
East Lothian	5	0	0	40	20	0	20	20
Edinburgh City	2	0	0	0	50	0	0	50
Falkirk	7	57	0	0	0	14	29	0
Fife	17	12	0	12	29	12	18	18
Glasgow City	2	0	0	0	0	0	50	50
Highland	656	17	16	14	13	14	15	11
Inverclyde	8	38	13	13	13	13	13	0
Midlothian	4	0	0	25	50	0	25	0
Moray	25	24	4	8	24	12	12	16
Na h-Eileanan Siar	251	22	19	14	8	12	14	11
North Ayrshire	2	0	0	0	50	0	50	0
North Lanarkshire	26	19	12	8	12	4	31	15
Orkney Islands	11	18	9	18	0	0	18	36
Perth & Kinross	13	23	15	0	23	15	8	15
Renfrewshire	3	0	33	0	67	0	0	0
Scottish Borders	22	23	9	5	32	14	18	0
Shetland Islands	11	0	18	18	18	0	36	9
South Ayrshire	7	14	14	14	0	0	29	29
South Lanarkshire	16	19	6	13	31	13	6	13
Stirling	15	33	13	0	7	7	20	20
West Dunbartonshire	14	14	14	7	29	21	0	14
West Lothian	39	18	10	8	15	10	26	13

5.3.3 Fuel type

Broad Habitats (BH) classified from IRS Level 3 Property Types (Table 5.34) for wildfires from the full dataset show that grasslands were the most frequent fuel type in almost all LAs, where they ranged from 24% to 79% of total number of wildfires in the Angus and Dundee City, respectively. The main exceptions were in the Highland and Na h-Eileanan Siar areas where wildfires on heathlands/shrubland accounted for 56% and 50% of total wildfires, respectively. Wildfires on broadleaved and conifer woodlands accounted for 50% of all wildfires that occurred in the Falkirk area and for 44% of all wildfires in the Scottish Borders, while woodland/forest fires accounted for only 1% of total numbers of wildfires in the Shetland Islands and Glasgow City areas. In addition, wildfires on loose refuse were most frequent in the Orkney Islands, where they accounted for 25% of all recorded fires, and in the Midlothian and Angus areas (19% of all wildfires for both areas). Regarding BHs classified from LCS88 classes (Table 5.35), seminatural or improved grasslands were also the most frequent fuel types in most LAs, but there was several LAs where wildfires on either mixed or conifer forest were the most frequent (Aberdeenshire, Clackmannanshire, East Lothian, Falkirk, Fife, Midlothian, Moray, Scottish Borders), while wildfires on shrublands were the most frequent in the Highland area.

On the other hand, analysis of the subset of bigger wildfires using BHs classified from IRS Level 3 Property Types (Table 5.36) showed that shrublands were the most frequent fuel type for 22 LAs, with grasslands being the most frequent fuel type in another eight (8) LAs and conifers in one LA (Moray). Looking at BHs based on LCS88 classes (Table 5.37), bigger wildfires on grasslands (either seminatural or improved) outnumbered wildfires on bogs and peatlands and shrubland combined and woodland fires (broadleaves, mixed and conifers) in 18 LAs. Wildfires on bogs and peatland and on shrubland were more frequent in the Angus, Highland and Na h-Eileanan Siar LAs (75%, 56% and 53% of all wildfires, respectively), while combined woodland fires were more frequent in Falkirk, Moray and Midlothian LAs (71%, 68% and 50% of all wildfires, respectively).

Table 5.34 Total count of IRS wildfires and proportions (in %) of Broad Habitat ( BH) types classified from Level 3 Property Types per Local Authority ( LA).

LA	Sum	Shrubland	Grasslands	Broadleaves	Conifers	Arable	Freshwater	Linear features	Built-up	Refuse	Other
Aberdeen City	125	26	38	18	3	1	1	0	0	13	0
Aberdeenshire	281	22	31	19	6	1	0	4	0	14	3
Angus	54	20	24	17	2	0	0	11	2	19	6
Argyll & Bute	355	35	36	8	4	1	0	4	0	3	8
Clackmannanshire	79	14	41	33	1	5	0	1	1	4	0
Dumfries & Galloway	177	28	29	10	7	2	0	12	0	9	3
Dundee City	150	7	79	7	1	0	0	0	2	5	0
East Ayrshire	609	14	57	7	3	0	0	4	0	9	6
East Dunbartonshire	81	21	30	11	2	1	0	5	0	12	17
East Lothian	154	11	34	21	8	1	0	6	1	6	12
East Renfrewshire	35	11	51	6	3	6	0	0	0	9	14
Edinburgh City	200	22	32	19	7	2	2	6	4	6	3
Falkirk	357	7	29	35	15	1	0	4	2	5	3
Fife	414	11	37	21	10	2	0	3	1	12	2
Glasgow City	198	10	75	1	0	1	1	2	0	8	4
Highland	1852	56	23	7	4	1	0	5	0	3	2
Inverclyde	245	40	46	4	1	0	0	2	0	4	4
Midlothian	202	10	27	24	6	1	0	8	3	19	1
Moray	323	9	41	21	8	1	0	9	0	10	1
Na h-Eileanan Siar	644	50	35	4	2	1	0	1	0	5	1
North Ayrshire	187	13	43	2	2	2	0	10	1	9	20
North Lanarkshire	690	21	52	6	1	1	0	4	1	11	3
Orkney Islands	101	17	37	1	2	4	0	7	2	25	6
Perth & Kinross	182	16	32	23	4	4	1	7	1	7	7
Renfrewshire	195	10	46	13	3	2	1	4	1	14	8
Scottish Borders	257	20	16	28	16	2	0	6	1	6	5
Shetland Islands	83	29	35	0	1	0	0	8	2	17	7
South Ayrshire	87	8	32	15	6	2	0	3	0	7	26
South Lanarkshire	368	14	37	10	3	1	0	5	0	18	12
Stirling	211	26	28	23	9	1	1	1	1	7	1
West Dunbartonshire	145	23	52	9	2	0	0	1	3	4	6
West Lothian	684	24	36	13	10	0	0	4	2	10	3

Table 5.35 Total count of IRS wildfires and proportions (in %) of Broad Habitat ( BH) types classified from LCS88 classes per Local Authority ( LA).

LA	Sum	Bogs & peatlands	Shrubland	Montane	Cliffs	Grasslands	Improved grass	Broadleaves	Mixed woodland	Conifers	Arable	Fresh water	Built- up	No Data
Aberdeen City	125	0	3	0	0	54	11	2	2	16	5	0	2	5
Aberdeenshire	281	2	7	0	0	16	5	8	13	37	10	0	0	0
Angus	54	0	13	0	0	35	9	2	17	11	2	0	7	4
Argyll & Bute	355	10	14	0	0	31	17	8	8	12	0	0	1	0
Clackmannanshire	79	0	0	0	0	18	6	8	42	10	11	0	5	0
Dumfries & Galloway	177	8	3	0	0	37	12	10	8	17	3	1	1	0
Dundee City	150	0	0	0	0	48	4	3	12	0	19	0	15	0
East Ayrshire	609	4	0	0	0	36	19	5	10	13	0	0	12	0
East Dunbartonshire	81	2	4	0	0	20	41	7	17	4	1	0	4	0
East Lothian	154	0	0	0	0	25	1	2	13	26	31	0	3	0
East Renfrewshire	35	6	0	0	0	23	63	0	3	3	0	0	3	0
Edinburgh City	200	0	12	0	0	15	11	6	10	14	21	2	12	0
Falkirk	357	2	0	0	0	11	14	2	15	51	1	0	3	1
Fife	414	1	3	0	0	20	9	8	15	28	11	1	2	1
Glasgow City	198	2	0	0	0	12	41	4	1	0	1	1	39	1
Highland	1852	10	32	0	0	14	15	7	4	16	0	0	0	0
Inverclyde	245	1	11	0	0	66	13	0	4	2	1	0	2	0
Midlothian	202	0	3	0	0	7	8	2	49	12	16	0	1	0
Moray	323	2	2	0	0	8	6	3	7	64	6	0	1	1
Na h-Eileanan Siar	644	18	23	0	0	1	45	1	3	0	0	1	5	3
North Ayrshire	187	3	1	0	0	26	48	2	3	3	1	0	9	4
North Lanarkshire	690	7	1	0	0	24	37	0	11	3	2	0	13	2
Orkney Islands	101	4	5	0	0	3	71	0	0	0	0	0	11	6
Perth & Kinross	182	1	5	0	0	26	7	12	25	17	7	0	1	0
Renfrewshire	195	3	0	0	0	33	39	2	8	3	6	0	8	0
Scottish Borders	257	0	6	0	0	21	19	4	11	35	2	0	2	0
Shetland Islands	83	11	16	0	1	10	41	0	0	0	0	0	20	1
South Ayrshire	87	0	1	0	0	23	45	7	11	8	2	0	2	0
South Lanarkshire	368	2	0	0	0	12	52	4	12	5	0	0	3	10
Stirling	211	2	2	0	0	27	9	8	28	15	5	0	3	1
West Dunbartonshire	145	1	9	0	0	30	19	1	20	4	0	0	14	1
West Lothian	684	16	0	0	0	18	28	0	5	21	4	0	5	2

Table 5.36 Total count of IRS wildfires with damage area greater than 1,000 m ² and proportions (in %) of Broad Habitat ( BH) types classified from Level 3 Property Types per Local Authority ( LA).

LA	Sum	Shrubland	Grasslands	Broadleaves	Conifers	Arable	Linear features	Refuse	Other
Aberdeen City	3	100	0	0	0	0	0	0	0
Aberdeenshire	27	56	30	4	7	0	0	0	4
Angus	4	100	0	0	0	0	0	0	0
Argyll & Bute	59	64	25	2	8	0	0	0	0
Clackmannanshire	7	57	43	0	0	0	0	0	0
Dumfries & Galloway	33	61	27	6	6	0	0	0	0
Dundee City	2	50	50	0	0	0	0	0	0
East Ayrshire	30	23	60	0	17	0	0	0	0
East Dunbartonshire	4	50	25	0	25	0	0	0	0
East Lothian	5	80	20	0	0	0	0	0	0
Edinburgh City	2	0	100	0	0	0	0	0	0
Falkirk	7	57	29	0	14	0	0	0	0
Fife	17	6	47	12	29	0	0	0	6
Glasgow City	2	0	100	0	0	0	0	0	0
Highland	656	73	17	3	4	0	1	0	1
Inverclyde	8	75	13	0	13	0	0	0	0
Midlothian	4	25	25	25	25	0	0	0	0
Moray	25	24	24	16	32	0	0	4	0
Na h-Eileanan Siar	251	65	28	2	4	0	0	0	0
North Ayrshire	2	0	100	0	0	0	0	0	0
North Lanarkshire	26	50	42	4	4	0	0	0	0
Orkney Islands	11	36	45	0	18	0	0	0	0
Perth & Kinross	13	54	15	8	15	8	0	0	0
Renfrewshire	3	33	67	0	0	0	0	0	0
Scottish Borders	22	55	23	14	5	0	5	0	0
Shetland Islands	11	100	0	0	0	0	0	0	0
South Ayrshire	7	43	43	14	0	0	0	0	0
South Lanarkshire	16	63	38	0	0	0	0	0	0
Stirling	15	60	13	13	13	0	0	0	0
West Dunbartonshire	14	57	21	14	7	0	0	0	0
West Lothian	39	44	33	0	18	0	0	3	3

Table 5.37 Total count of IRS wildfires with damage area greater than 1,000 m ² and proportions (in %) of Broad Habitat ( BH) types classified from LCS88 classes per Local Authority ( LA).

LA	Sum	Bogs & peatlands	Shrubland	Montane	Cliffs	Grasslands	Improved grass	Broadleaves	Mixed woodland	Conifers	Arable	Fresh water	Built- up	No Data
Aberdeen City	3	0	0	0	0	67	33	0	0	0	0	0	0	0
Aberdeenshire	27	11	33	0	0	22	4	0	7	22	0	0	0	0
Angus	4	0	75	0	0	0	0	0	0	25	0	0	0	0
Argyll & Bute	59	10	24	0	0	29	10	8	8	10	0	0	0	0
Clackmannanshire	7	0	0	0	0	71	29	0	0	0	0	0	0	0
Dumfries & Galloway	33	18	12	0	0	39	6	9	0	15	0	0	0	0
Dundee City	2	0	0	0	0	0	0	0	0	0	50	0	50	0
East Ayrshire	30	20	3	0	0	47	3	0	0	20	0	0	7	0
East Dunbartonshire	4	0	0	0	0	75	0	25	0	0	0	0	0	0
East Lothian	5	0	0	0	0	100	0	0	0	0	0	0	0	0
Edinburgh City	2	0	0	0	0	0	0	0	50	0	0	0	50	0
Falkirk	7	29	0	0	0	0	0	0	0	71	0	0	0	0
Fife	17	0	6	0	0	29	12	6	6	35	6	0	0	0
Glasgow City	2	0	0	0	0	50	50	0	0	0	0	0	0	0
Highland	656	13	43	0	0	14	14	8	2	6	0	0	0	0
Inverclyde	8	0	0	0	0	100	0	0	0	0	0	0	0	0
Midlothian	4	0	25	0	0	25	0	0	0	50	0	0	0	0
Moray	25	4	8	0	0	20	0	0	0	68	0	0	0	0
Na h-Eileanan Siar	251	27	26	0	0	1	40	0	1	0	0	1	3	2
North Ayrshire	2	0	0	0	0	100	0	0	0	0	0	0	0	0
North Lanarkshire	26	23	8	0	0	23	35	0	8	4	0	0	0	0
Orkney Islands	11	18	0	0	0	9	55	0	0	0	0	0	0	18
Perth & Kinross	13	8	23	0	0	38	8	0	15	8	0	0	0	0
Renfrewshire	3	0	0	0	0	33	0	0	33	0	0	0	33	0
Scottish Borders	22	0	23	0	0	41	14	0	0	23	0	0	0	0
Shetland Islands	11	9	36	0	0	9	36	0	0	0	0	0	9	0
South Ayrshire	7	0	14	0	0	14	43	14	0	14	0	0	0	0
South Lanarkshire	16	19	0	0	0	38	25	0	0	0	0	0	0	19
Stirling	15	20	0	0	0	60	7	0	13	0	0	0	0	0
West Dunbartonshire	14	0	21	0	0	71	0	0	0	7	0	0	0	0
West Lothian	39	46	0	0	0	10	21	0	0	21	0	0	0	3

5.3.4 Ignition motives

Wildfires caused by deliberate ignitions were the most frequent in 23 of the 32 LAs in Scotland (Table 5.38), accounting for 53% of total wildfires in the Stirling area and up to 95% in the Renfrewshire area and 97% in the Inverclyde areas. On the other hand, wildfires caused by accidental ignitions were the most frequent in the Argyll and Bute, Na h-Eileanan Siar, Dumfries and Galloway, Aberdeenshire, Perth and Kinross, Shetland Islands, Orkney Islands, Moray and Highland areas, where they accounted for 45% to 60% of all wildfires recorded. This was expected based on the finding (Section 5.2.5) that wildfires in the least accessible and least populated areas are more likely to be accidental, as most of the more remote and very remote areas in Scotland are concentrated in these LAs.

Table 5.38 Total count of IRS wildfires and proportions (in %) of recorded motive per Local Authority ( LA).

LA	Sum	Accidental	Deliberate	Not known
Aberdeen City	125	22	70	7
Aberdeenshire	281	55	42	3
Angus	54	24	59	17
Argyll & Bute	355	60	33	7
Clackmannanshire	79	9	77	14
Dumfries & Galloway	177	56	38	6
Dundee City	150	1	90	9
East Ayrshire	609	4	90	5
East Dunbartonshire	81	5	89	6
East Lothian	154	17	60	23
East Renfrewshire	35	3	94	3
Edinburgh City	200	10	70	20
Falkirk	357	6	71	23
Fife	414	14	68	18
Glasgow City	198	2	94	4
Highland	1852	45	17	38
Inverclyde	245	2	97	2
Midlothian	202	7	78	15
Moray	323	46	42	12
Na h-Eileanan Siar	644	59	10	31
North Ayrshire	187	8	91	1
North Lanarkshire	690	4	92	4
Orkney Islands	101	50	26	24
Perth & Kinross	182	51	40	10
Renfrewshire	195	3	95	2
Scottish Borders	257	20	67	13
Shetland Islands	83	51	19	30
South Ayrshire	87	25	70	5
South Lanarkshire	368	14	83	3
Stirling	211	18	53	29
West Dunbartonshire	145	8	90	2
West Lothian	684	7	79	14

Looking at the subset of bigger wildfires, those caused by deliberate ignitions were the most frequent in 14 of the 31 LAs (Table 5.39), accounting for all wildfires in the Inverclyde, Renfrewshire, Dundee City and North Ayrshire LAs. On the other hand, wildfires caused by accidental ignitions were the most frequent in eight (8) LAs that include those where most bigger wildfires were recorded. In particular, accidental wildfires in the Aberdeenshire, Dumfries & Galloway, Argyll and Bute, Na h-Eileanan Siar and Highland areas accounted for 85%, 73%, 63%, 59% and 47% of all bigger wildfires in these LAs, respectively.

Table 5.39 Total count of IRS wildfires with damage area greater than 1,000 m ² and proportions (in %) of recorded motive per Local Authority ( LA).

LA	Sum	Accidental	Deliberate	Not known
Aberdeen City	3	67	33	0
Aberdeenshire	27	85	7	7
Angus	4	50	25	25
Argyll & Bute	59	63	27	10
Clackmannanshire	7	0	71	29
Dumfries & Galloway	33	73	18	9
Dundee City	2	0	100	0
East Ayrshire	30	10	87	3
East Dunbartonshire	4	0	75	25
East Lothian	5	0	80	20
Edinburgh City	2	0	0	100
Falkirk	7	0	29	71
Fife	17	18	35	47
Glasgow City	2	50	50	0
Highland	656	47	11	43
Inverclyde	8	0	100	0
Midlothian	4	50	25	25
Moray	25	40	16	44
Na h-Eileanan Siar	251	59	7	34
North Ayrshire	2	0	100	0
North Lanarkshire	26	15	81	4
Orkney Islands	11	73	0	27
Perth & Kinross	13	77	23	0
Renfrewshire	3	0	100	0
Scottish Borders	22	23	50	27
Shetland Islands	11	73	9	18
South Ayrshire	7	43	29	29
South Lanarkshire	16	44	56	0
Stirling	15	40	20	40
West Dunbartonshire	14	21	79	0
West Lothian	39	3	69	28

5.3.5 Burnt area size

Most wildfires in all LAs were small fires with burnt areas of less than 100 m², except for Na h-Eileanan Siar where bigger fires of more than 1,000 m² were the most frequent (39%) (Table 5.40). A significant proportion of bigger wildfires (35%) was also recorded in the Highland area. The greater numbers of small wildfires occurred in Renfrewshire, Falkirk, Aberdeen City, Glasgow City and Edinburgh City areas, where they accounted for more than 84% of all recorded wildfires. On the other hand, fewer small wildfires occurred in the Na h-Eileanan Siar, Highland and Argyll and Bute areas, where they accounted for 35%, 44% and 54% of all wildfires, respectively. Again, this finding is consistent with the previous analysis (Section 5.2.5) which showed that small fires are more likely to occur in easily accessible urban areas, while bigger fires tend to occur in more remote and less accessible areas.

Table 5.40 Total count of IRS wildfires and proportions (in %) of recorded Damage Area ( DA) group: None; 0-100 m ²; 100-1,000 m ² and >1,000 m ² per Local Authority ( LA).

LA	Sum	None	DA0_100	DA100_1000	DA>1000
Aberdeen City	125	4	85	9	2
Aberdeenshire	281	7	71	13	10
Angus	54	17	65	11	7
Argyll & Bute	355	7	54	23	17
Clackmannanshire	79	4	71	16	9
Dumfries & Galloway	177	6	57	19	19
Dundee City	150	5	81	13	1
East Ayrshire	609	9	69	17	5
East Dunbartonshire	81	4	83	9	5
East Lothian	154	5	81	11	3
East Renfrewshire	35	0	80	20	0
Edinburgh City	200	4	84	12	1
Falkirk	357	3	87	8	2
Fife	414	3	81	12	4
Glasgow City	198	9	84	7	1
Highland	1852	3	44	17	35
Inverclyde	245	2	75	20	3
Midlothian	202	13	77	8	2
Moray	323	2	78	12	8
Na h-Eileanan Siar	644	3	35	22	39
North Ayrshire	187	10	80	9	1
North Lanarkshire	690	5	79	13	4
Orkney Islands	101	10	65	14	11
Perth & Kinross	182	3	75	14	7
Renfrewshire	195	3	89	7	2
Scottish Borders	257	7	72	12	9
Shetland Islands	83	10	64	13	13
South Ayrshire	87	11	75	6	8
South Lanarkshire	368	7	77	12	4
Stirling	211	3	68	22	7
West Dunbartonshire	145	1	70	20	10
West Lothian	684	2	76	16	6

5.4 Local Authority Cluster Groups

5.4.1 Cluster analysis

As mentioned in Section 5.1, we used hierarchical cluster analysis to group Local Authorities (LAs) based on similarities of their wildfire characteristics, aiming to facilitate an easier comparison and understanding of wildfire occurrence patterns in different LAs. For this purpose, we compiled a dataset of wildfire characteristics recorded in the IRS for all identified wildfires occurring in each of the 32 LAs that consisted of the following values:

• Proportions (in %) of wildfires belonging to main Level 3 Property Types: grasslands, heathlands, refuse, scrub land, broadleaves, and conifers.
• Proportions (in %) of wildfires occurring in winter (December, January, February), spring (March, April, May), summer (June, July, August) and autumn (September, October, November) seasons.
• Proportions (%) of wildfires with burnt area estimates for Outdoor Damage Area classes of 0-100 m²; 100-1,000 m² and more than 1,000 m².
• Proportions (in %) of wildfires with an accidental or a deliberate ignition motive.

We calculated and used proportions because cluster analysis requires using numeric variables. We used Ward's method of hierarchical cluster analysis implemented in the R statistical software using the dataset with all wildfire characteristics (Clusters A) and without the proportions of ignition motive (Clusters B); this was done in consultation with the SFRS aiming to assess the effect that motive of ignition, which is prone to a certain degree of subjectivity because of personal beliefs or experience, has on the groupings of LAs.

Figure 5.13 shows the result of the cluster analysis in the form of two dendrograms based on similarity distance between LAs. The analysis using the dataset with all wildfire characteristics produced two main cluster groupings (Figure 5.13a): the smaller Cluster group A2 consisting of six (6) LAs and the bigger Cluster that was divided into Cluster sub-groups A1 and A3, consisting of nineteen (19) and seven (7) LAs, respectively. Regarding the number of wildfires within Cluster groups, Cluster groups A1, A2 and A3 contained 4,441, 3,212 and 2,072 wildfires, respectively.

Removing the motive of ignition from the analysis resulted in two main cluster groupings that each was split into two sub-groups (Figure 5.13b): Cluster sub-groups B1 and B3, consisting of eleven (11) and seven (7) LAs, respectively, and Cluster sub-groups B2 and B4, consisting of twelve (12) and two (2) LAs, respectively. Cluster groups B1, B2, B3 and B4 contained 2,434, 3,578, 1,217 and 2,496, respectively.

Table 5.41 Cluster groupings of each Local Authority ( LA).

ID	LA	Clusters A	Clusters B	ID	LA	Clusters A	Clusters B
1	Aberdeen City	1	1	17	Inverclyde	3	2
2	Aberdeenshire	1	1	18	Midlothian	1	3
3	Angus	1	1	19	Moray	1	1
4	Argyll & Bute	2	2	20	Na h-Eileanan Siar	2	4
5	Clackmannanshire	1	2	21	North Ayrshire	1	3
6	Dumfries & Galloway	2	2	22	North Lanarkshire	3	2
7	Dundee City	3	2	23	Orkney Islands	2	3
8	East Ayrshire	3	2	24	Perth & Kinross	1	1
9	East Dunbartonshire	1	3	25	Renfrewshire	1	3
10	East Lothian	1	1	26	Scottish Borders	1	1
11	East Renfrewshire	3	2	27	Shetland Islands	2	3
12	Edinburgh City	1	1	28	South Ayrshire	1	1
13	Falkirk	1	1	29	South Lanarkshire	1	3
14	Fife	1	1	30	Stirling	1	2
15	Glasgow City	3	2	31	West Dunbartonshire	3	2
16	Highland	2	4	32	West Lothian	1	2

Looking at Table 5.41 and Figure 5.14, it becomes evident that Cluster group A2 consists of the LAs that include more remote rural areas, while Cluster groups A1 and A3 consist of LAs where most urban and accessible rural areas are found. Similarly, Cluster groups B2 and B4 consist of the LAs with more rural areas and Cluster groups B1 and B3 consist of LAs with more urban and accessible rural areas. The main difference between Cluster A and B groups is that the Highland and Na h-Eileanan Siar LAs form a separate group (Cluster B4) consisting of bigger wildfires on bogs and shrublands, while those two (2) LAs are included in the Cluster A2 group. There are also differences in cluster groupings in the more urban areas in the Central Belt, while there are similarities in groupings in the eastern parts of Scotland.

To understand differences in wildfire occurrence patterns within the identified cluster groups, we conducted the same analysis as in Sections 5.2 and 5.3 looking at patterns of seasonality, fuel type, motives of ignition, burnt area sizes along with accessibility aspects based on the Urban-Rural classification.

5.4.2 Temporal – Seasonality

Figure 5.15 shows overall limited inter-annual variability with regards to the relative proportions of wildfire incidents belonging to the different cluster groups. There seem to be fewer than average Cluster group A2 and B4 wildfires between 2015 to 2017 proportionally to Cluster groups A1 and A3 and B1, B2, B3 wildfires, respectively. This indicated that fewer than expected remote wildfires were recorded for these years.

Looking at seasonality at a monthly basis, most Cluster group A3 wildfires occurred in April, and this was the same but to a lesser extent for the Cluster group A2 wildfires (Figure 5.16a). On the other hand, most wildfires in Cluster group A1 occurred almost equally in April and May. Similarly, most wildfires in Cluster groups B2 and B4 occurred in April, but there was a greater proportions of early spring wildfires (in March) than late spring wildfires (in May) in Cluster group B4 compared to group B2 (Figure 5.16b). Most wildfires in Cluster groups B1 and B3 occurred almost equally in April and May, but their combined proportion was greater in Cluster group B3 than in B1, with the latter also having the greater proportion of summer wildfires compared to the other three (3) groups.

5.4.3 Fuel type

Figure 5.17 shows clear patterns related to fuel type composition between cluster groups. Cluster group A2 includes a greater number of wildfires on shrub and on bogs and peatlands compared to Cluster groups A1 and A3 (Figure 5.17a). This was expected since Cluster group A2 includes LAs with a more remote and rural character. Based on Broad Habitats (BH) classified from both IRS Level 3 Property Types and LCS88 classes, Cluster group A3 is dominated by grassland wildfires and has the greater proportion of wildfires on built-up areas. Grassland fires are also frequent within Cluster group A1, but this group also has the greater proportion of woodland fires (mainly for mixed woodlands and conifer forests) and of fires on arable land (Figure 5.17b).

Similar patterns of fuel type composition emerge for the Cluster B groups (Figure 5.18). More than half of wildfires included in Cluster group B4 were on shrub using BHs based on Level 3 Property types, while wildfires on bogs and peatlands and on shrubland were the dominant in group B4 based on LCS88 classes. This was expected based on information given in Tables 5.36 and 5.37 for the Highland and Na h-Eileanan Siar areas. Based on BHs classified from both IRS Level 3 Property Types and LCS88 classes, Cluster group B2 is dominated by grassland wildfires and has the greater proportion of wildfires on built-up areas. Grassland fires are also frequent within Cluster group B1, but this group also has the greater proportion of woodland fires (mainly for conifer forests), while Cluster group B3 had the greater proportion of wildfires on improved grassland and on mixed woodland (Figure 5.18b).

5.4.4 Ignition motives

Figure 5.19a shows that one of the main parameters causing the split between Cluster (sub) groups A1 and A3 is the motive of ignition. While both Cluster groups A1 and A3 consist of wildfires in more urban or accessible rural areas, Cluster group A3 consists almost exclusively of wildfires caused by deliberate ignitions, compared to Cluster group A2, where deliberate wildfires are also the most frequent ones but there are also important proportions of wildfires with an accidental or unknown motive of ignition. On the other hand, accidental ignitions followed by unknown ignitions were the most frequent motives for the wildfires belonging to Cluster group A2.

Ignition motive was not used in the analysis for Cluster B groups. As expected, accidental wildfires were the dominant ones in Cluster group B4 (Highland and Na h-Eileanan Siar areas), which also had the smaller proportion of wildfires caused by deliberate ignitions between the different cluster groups (Figure 5.19b). Cluster group B1 had the second greater proportion of accidental wildfires, but deliberate ignitions were the most frequent in this group and in Cluster groups B2 and B3, both of which had similar proportion of wildfires with accidental, deliberate and unknown motives.

5.4.5 Burnt area size

Figure 5.20a shows that most of the smaller wildfires with estimated burnt areas less than 100 m² are found in Cluster groups A1 and A3, while the bigger wildfires with burnt areas between 100 m² and 1,000 m² and more than 1,000 m² are more frequent in Cluster group A2. Looking at Cluster B groups (Figure 5.20b), most of the bigger wildfires (>1,000 m²) are found in Cluster group B4, while Cluster groups B2 and B4 have almost equal proportions of wildfires with burnt areas between 100 m² and 1,000 m². Cluster groups B1 and B3 have similar proportions of wildfires falling in the different damage area classes.

5.4.6 Urban – Rural Classification

Based on Figure 5.21a, most wildfires in Cluster group A2 occurred in very remote areas (66%), while an almost equal number of wildfires occurred in remote and accessible rural areas (around 16% of all wildfires for both areas). Cluster groups A1 and A3 showed similar patterns with regards to the number of wildfires within the different folds of the Urban Rural classification. This was expected because wildfires belonging to both clusters occurred mostly within accessible rural areas; the difference being that these wildfires were slightly more frequent in Cluster group A1 than in group A3 (82% compared to 69%). On the other hand, around 10% of wildfires in Cluster group A3 occurred in urban areas, which was the greatest proportion between the three Cluster A groups.

Regarding Cluster B groups (Figure 5.21b), most wildfires in Cluster group B4 occurred in very remote areas (67%), while, as in the case of Cluster group A2, an almost equal number of wildfires occurred in remote and accessible rural areas (around 16% of all wildfires for both areas). Cluster groups B1 and B3 had similar proportions of wildfires in accessible rural areas (78% and 74%, respectively), but a greater proportion of wildfires in very remote areas was found in Cluster group B3 than in group B1 (14% compared to 3%) and a greater proportion of wildfires in remote areas occurred in Cluster group B1 than in group B3 (13% compared to 3%). Cluster group B2 shows a pattern that is intermediate to Clusters groups B1 and B3 and Cluster group B4, with 67% of wildfires in this group occurring in accessible rural areas, 16% of wildfires occurring in remote areas and 8% of wildfires occurring in very remote areas.

5.4.7 Conclusions

The results of the cluster analysis revealed clear groupings of LAs that were based on similarities regarding wildfire occurrence patterns and characteristics:

• LAs where bigger wildfires, in terms of larger burnt area sizes, caused by accidental ignitions are more frequent and more likely to occur that mainly affect remote seminatural habitats (i.e., bogs and peatland and heathlands/shrubland) (Cluster group A2). When motive of ignition is ignored, then the bigger and very remote wildfires affecting bogs and peatlands and heathlands are mainly found in Cluster group B4 (Highland and Na h-Eileanan Siar areas), while other wildfires mainly on shrubs and seminatural grasslands in accessible and remote rural areas are found in Cluster group B2, which consists of LAs located in the central and south-western parts of Scotland.
• LAs where small or smaller wildfires, mainly on grassland and in proximity of urban areas or in relatively accessible rural areas, are more frequent and that are caused by either a mix of ignition motives (Cluster group A1) or predominantly by deliberate ignitions (Cluster group A3). Again, when motive is removed from the analysis, LAs with smaller wildfires in relatively accessible rural areas and located at central and eastern Scotland are split mainly based on fuel type composition between those where woodland fires are dominant (Cluster group B1) and those where fires on (mainly improved) grassland are dominant (Cluster group B3).

We need to stress that since this is a data-driven approach, its outcomes depend purely on the variables used to conduct the analysis. For example, splitting between Cluster groups A1 and A3 seemed to be greatly influenced by the motive of ignition, while removing this variable changed the outcome of the analysis to a certain degree and resulted in different cluster groupings. However, this approach provides an objective way to identify LAs with similarities in their wildfire occurrence patterns, which could be useful for making decisions about prioritising areas of certain wildfire patterns or characteristics to improve preparedness or enhancing the wildfire suppression capability of certain regional SFRS regional stations.