High-resolution Canada domain disturbance forcings suitable for land surface modeling applications

Spatially explicit fire and harvest data are useful for driving land surface model (LSM) simulations of the carbon cycle. From 1985-present, numerous Canadian disturbance datasets exist. However, before the launch of Landsat-4 (1984), few are available. We create spatially explicit LSM disturbance drivers for Canada for 1740-2018. We catalog and harmonize spatial and aspatial datasets and develop a novel algorithm that reconstructs disturbance far back in time using stand age. Based on possible historical scenarios, we reconstruct 283-394 Mha of fire and 3.42 Mha of harvest in total Canada-wide from 1740-1918. After 1918, when spatial records are available, we supplement them by reconstructing 25.79-60.30 Mha of fire and 24.75 Mha of harvest. After 1984, we exclusively use spatially explicit records. We verify the algorithm by comparing the inputs and resultant drivers and examine diagnostic metrics to disentangle the contribution of spatial, aspatial, and stand-age data. The resulting drivers primarily capture stand-replacing disturbance on forested land. Our forcings and algorithm will improve the representation of disturbance-mediated impacts on Canada's terrestrial carbon cycle and possibly in other regions.