Search

Fuels within the immediate vicinity of a house (e.g., within 30–60 m), referred to as the 'home-ignition zone', are important determinants of structure damage during wildfires. Methods for mapping home-ignition zone fuels using earth observing satellites are lacking, limiting the capacity to quantify the spatial and temporal dynamics of urban fuel hazard and wildfire risk. Here, we (i) develop a methodology to map the fractional cover of five common fuel types (i.e., bare, buildings, herbaceous, woody and water cover) at the scale of the home-ignition zone (45 m radius area) using Landsat, Sentinel 1, and PALSAR imagery, and (ii) demonstrate that these maps improve prediction of house loss in wildfires in western Canada. A fractional cover database was created across thirteen cities via manual digitisation of aerial photos and used to train and validate random forest models consisting of combinations of Landsat, Sentinel 1 and PALSAR indices. The best models for the fractional cover of each fuel type explained most of the variation in the data (R2 = 0.66–0.96) and had high accuracy (Mean absolute error = 0.02–0.11) with low bias (Mean error = −0.005–0.008). The combination of optical (Landsat) and synthetic aperture radar (Sentinel 1 and/or PALSAR) improved model accuracy relative to models containing only multi-spectral indices. A house loss database was then derived for five major fires in western Canada and used to examine the effect of mapped fractional cover on house loss rates for 227 neighbourhood blocks. House loss rates were greater in blocks with a low amount of defensible space (i.e., areas of low fuel hazard; bare, herbaceous, or water cover) surrounding houses compared to those with extensive defensible space. This finding is consistent with house loss research globally and exposure assessment guidelines for communities in North America. Our fuel cover mapping methodology will facilitate improvements to models of house loss through the inclusion of important home-ignition zone fuel properties, allowing for spatially and temporally dynamic (i.e., annual) estimates of house loss likelihood and wildfire risk to communities. • Fuel fractional cover can be accurately mapped with Landsat, PALSAR and Sentinel 1. • Fusing optical and radar imagery improved the predictive accuracy of cover models. • PALSAR and Sentinel 1 help distinguish cover types with similar spectral signatures. • Fuel cover maps improved the prediction of house loss during fire events. [ABSTRACT FROM AUTHOR]

Atmospheric delay (AD) is the main source of error in time-series interferometric synthetic aperture radar (InSAR) deformation estimation over large areas. In this study, we propose a bidirectional gated recurrent unit (BiGRU) model to correct random and seasonal ADs adaptively. The BiGRU model decomposes InSAR time-series measurements into non-seasonal and seasonal components by adopting a branched network structure and extracting component-wise features separately. To remove seasonal ADs and meanwhile preserve true seasonal deformation, a dense and fully connected layer with weighted feature learning was designed. Five typical time-series deformation patterns were simulated for model training, and its robustness was evaluated using synthetic data. We applied the trained model to two city clusters in China (Guangdong and Jiangxi-Hunan) using 178 Sentinle-1 images. The results showed that BiGRU with moderate Generic Atmospheric Correction Online Service (GACOS) and spatiotemporal filtering (pGA_Fi_BiGRU) reduced the standard deviation of InSAR time-series measurements by 64.3% in the Guangdong region and by 53.5% in the Jiangxi-Hunan region compared with the raw data. Compared with the traditional combined GACOS and spatiotemporal filtering processing methods, the pGA_Fi_BiGRU improved the AD reduction performance by 4.7% and by 8.5% in Guangdong and Jiangxi-Hunan, respectively. The InSAR time-series deformation after pGA_Fi_BiGRU processing removed residual ADs and preserved true deformation, which agreed well with the geodetic leveling and Global Navigation Satellite System data. The first overall subsidence velocity of the Irrawaddy Delta city cluster in Myanmar was then mapped, followed by time-series deformation estimation using pGA_Fi_BiGRU. Representative time-series deformation due to groundwater extraction, coastal erosion, and accretion were properly derived, suggesting that the proposed model can be generalized to other city clusters with different atmospheric noise and geophysical dynamics. • Land subsidence of two city clusters was measured using Sentinel-1 images. • The pGA_Fi_BiGRU model was proposed to correct the seasonal ADs. • The performance enhancement in AD reduction varied across different cities. • Generalizability was demonstrated using the third city cluster. [ABSTRACT FROM AUTHOR]