Day 4

Detailed paper information

Back to list

Paper title the potential of SMOS L-VOD for post-fire monitoring of dense forests
Authors
  1. Emma Bousquet CESBIO, CNRS Speaker
  2. Arnaud Mialon CESBIO, Université de Toulouse, CNES/CNRS/IRD/UPS
  3. Nemesio Rodriguez CESBIO (CNRS, Université Paul Sabatier, CNES, IRD, INRAE)
  4. Stéphane Mermoz GlobEO
  5. Yann H. Kerr CESBIO, Université de Toulouse, CNES/CNRS/INRAE/IRD/UPS
Form of presentation Poster
Topics
  • A3. Biosphere
    • A3.06 Biomass monitoring
Abstract text For the past decades, wildfires have been increasing in frequency and in severity worldwide. These fires are source of substantial quantities of CO2 released in the atmosphere. They can also lead to the destruction of natural ecosystems and biodiversity. Fires are triggered by various factors that depend on the climate regime and on the vegetation type. Despite the large number of studies conducted on wildfires, post-fire vegetation recovery is still to be better understood, and depends highly on the vegetation type.
In this study, we present pre and post fire climate and vegetation anomalies at global scale, from several remotely sensed observations, such as air temperature (MODIS), precipitation (PERSIANN-CDR), soil moisture (SMOS), and terrestrial water storage (GRACE). Four remotely sensed variables related to vegetation are used and compared, from optical (the enhanced vegetation index (EVI) from MODIS) to microwave wavelengths opacities ranging from 2 to 20 cm: X-band, C-band, and L-band vegetation optical depth (X-VOD, C-VOD, and L-VOD), obtained with AMSR-2 and SMOS satellites. Fires are detected with the MODIS Active Fire product (MOD14A1_M). All datasets are resampled to SMOS grid (~ 25 km) and at a monthly timescale, for the time period June 2010 – December 2020.
We focus our analysis on five particular biomes : grasslands, tropical savannas, needleleaf forests, sparse broadleaf forests, and dense broadleaf forests. Anomalies of all data are computed over the major fires of the ten-year period, at global scale, then time series are readjusted on the fire date and averaged by biome.
We observe a severe drought before the majority of the fire events, and in particular over forests, which generally maintain a steady humidity all year. Pre-fire temperature anomalies are particularly significant in boreal needleleaf forests. In contrast, over savannas and grasslands, the pre-fire drought is slight while an increase in the biomass volume (e. g., available fuel) is supposed to expedite fires. As expected, C- and X-bands are more affected by sparse vegetation fires, as these frequencies are sensitive to the smaller branches and leaves; whereas L-band is particularly impacted over dense broadleaf forest fires, as it is a measurement of coarse woody elements (trunks and stems). For all biomes, the optical-based index (EVI) decreases significantly after fire but recovers quickly, as it observes only herbage and green canopy foliage. The contrasted recovery duration between L-VOD and the other variables over dense forests shows that fires affect coarse woody elements in the long term, while stems and leaves resprout faster. Our study shows the potential of SMOS L-VOD to monitor fire-affected areas as well as post-fire recovery, especially over densely vegetated areas. This study is also the first one to compare multi-frequencies VODs and to observe the impact of fire in L-VOD signal.

Figure - EVI, X-, C-, L-VOD, precipitation, SM, TWS, and temperature anomalies time series, shifted on the fire date, for (a) 520 points in the grassland biome; (b) 232 points in the savanna biome; (c) 701 points in the needleleaf forest biome; (d) 69 points in the sparse broadleaf forest biome; and (e) 48 points in the dense broadleaf forest biome. The missing values are mainly due to snow filtering.