This data publication contains a set of files in which different variables related to fire burned severity (Canada Landsat Burned Severity, CanLaBS) were computed for all events in Canada between 1985 and 2015 as detected by the Canada Landsat Disturbance (CanLaD (Guindon et al. 2017 and 2018) product. Details on the creation of this product are available in Guindon et al. 2020 (https://doi.org/10.1139/cjfr-2020-0353) and in supplementary materials accompanying the publication. The current document is therefore a complement to the article and supplementary materials. The supplementary materials are referenced in the publication (cjfr-2020-0353suppla, cjfr-2020-0353supplb etc.).

This is the first Canada-wide product that aims to promote nationwide research on fire severity by making available the data used in the article. The data is in the form of grids composed of pixels at a resolution of 30m. To simplify the distribution and manipulation of the data and considering that two or three fire occurrences within a given location is rare (respectively 2.3% and less than 0.01%), only the most recent fire data are considered in the final product. For these very rare cases, from 2015 to 1985, the most recent burned areas overlap the older data. Overlapping fire count can be found in layer “CanLaBS_Nbdisturb_v0”, multiple fire events in same areas have values equal to or greater than two.

Landsat radiometric values for calculating the NBR index were derived from summer Landsat mosaics (July and August), for years 1984 to 2015 (Guindon et al. 2018). These mosaics were developed from individual USGS Landsat scenes with surface reflectance correction (Masek et al., 2006; Vermote et al., 2006). For each annual compound, the pixel with the less atmospheric opacity was selected. An algorithm was also developed to remove clouds that were not detected by the cloud masks provided with the USGS data.

Here is a general description of the layers provided and a more technical description can be found in Table 1 (see "Ressources" section below):

1. NBR and dNBR. All these values are multiplied by 1000. The value of dNBR represents the value obtained for NBRpre - NBRpost. It is calculated for each pixel that was classified as a fire in CanLaD, according to the corrected year (see cjfr-2020-0353suppla).

2. Year of fire. The fire years detected in CanLaD (Guindon et al. 2018) was corrected using different fire databases, this layer contains the correct year. (see cjfr-2020-0353suppla)

3. Julian Days of the Fire, based on various high-resolution products. However, this variable is only available from 1989 onwards.

4. Presence of salvage logging one year after the fire. Classification of satellite images detecting scarified soils (see cjfr-2020-0353suppld).

5. Pre-fire forest attributes: Pre-fire forest attributes values were calculated for median mosaics, from 1985 to 2000. These attributes values were derived from NFI (national forest inventory) photo-plot attributes and were spatialized. Pre-fire attribute values were created to stratify the analyses (see cjfr-2020-0353supplc). The predicted variables are as follows:

• Canopy density in percent.

• Predicted living biomass in tonnes per hectare.

• Percentage coniferous biomass proportion of total biomass.

• Percentage hardwood biomass proportion of total biomass.

• Percentage unknown species biomass proportion of total biomass. Note, as unknown species are found especially in northern areas, they are considered coniferous for the purpose of the article.

6. Missing remote sensing data, one year after the fire. The estimation of burned severity needs NBR data (NBRpost) in the next year after fire occurrences. NBRpost is available for 91% of the cases, but for the remaining 9%, no data were available due to the presence of clouds. For these cases, satellite data from the years following the fire were used with a regression radiometry correction. This gives values to missing data for year following the fire. This layer flags the areas that have derived data. The values of 1= one year after the fire (no regression), 2= two years after the fire (regression), 3= three years after the fire (regression) and 4= four years after the fire (no regression, set as missing data). (see cjfr-2020-0353supplb).

7. Areas with more than one fire disturbance between 1985 and 2015 (1=one single disturbance, 2=two or more, 3=three or more).

## Data citation:

1. Guindon, L., Villemaire P., Manka F., Dorion H. , Skakun R., St-Amant R., Gauthier S. : Canada Landsat Burned Severity (CanLaBS): a Canada-wide Landsat-based 30-m resolution product of burned severity since 1985

https://doi.org/10.23687/b1f61b7e-4ba6-4244-bc79-c1174f2f92cd

2. The creation, the validation and the limits of the CanLaBS product are describe in the text and supplementary material: Guindon, L., Gauthier, S., Manka, F., Parisien, MA, Whitman, E., Bernier, P., Beaudoin, A., Villemaire P., Skakun R. Trends in wildfire burn severity across Canada, 1985 to 2015

https://doi.org/10.1139/cjfr-2020-0353

## References cited:

1. Guindon, L., Villemaire, P., St-Amant, R., Bernier, P.Y., Beaudoin, A., Caron, F., Bonucelli, M., and

Dorion, H. 2017. Canada Landsat Disturbance (CanLaD): a Canada-wide Landsat-based 30m

resolution product of fire and harvest detection and attribution since 1984.

https://doi.org/10.23687/add1346b-f632-4eb9-a83d-a662b38655ad

2. Guindon, L., Bernier, P., Gauthier, S., Stinson, G., Villemaire, P., & Beaudoin, A. (2018). Missing forest cover gains in boreal forests explained. Ecosphere, 9(1), e02094.

https://doi.org//10.1002/ecs2.2094

3. Masek, J.G., Vermote, E.F., Saleous N.E., Wolfe, R., Hall, F.G., Huemmrich, K.F., Gao, F., Kutler, J., and Lim, T-K. (2006). A Landsat surface reflectance dataset for North America, 1990–2000. IEEE Geoscience and Remote Sensing Letters 3(1):68-72.

http://dx.doi.org/10.1109/LGRS.2005.857030.

4. Vermote, E., Justice, C., Claverie, M., & Franch, B. (2016). Preliminary analysis of the performance of the Landsat 8/OLI land surface reflectance product. Remote Sensing of Environment.

http://dx.doi.org/10.1016/j.rse.2016.04.008.