Day 4

Detailed paper information

Back to list

Paper title Multitemporal comparisons between GEDI Lidar products and SMOS L-VOD retrieved by the latest version of level 2 algorithm
Authors
  1. Cristina Vittucci Tor Vergata University of Rome Speaker
  2. Leila Guerriero Università di Roma Tor Vergata
  3. Paolo Ferrazzoli Università degli Studi di Roma "Tor Vergata"
  4. Philippe Richaume CESBIO, Université de Toulouse, CNES/CNRS/INRAE/IRD/UPS
  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 The key parameters provided by the Soil Moisture and Ocean Salinity (SMOS) mission over land are soil moisture (SM) and L-band vegetation optical depth (L-VOD). Although the retrieval of SM was the low hanging fruit of the mission, the information about vegetation has reached maturity causing a growing interest in testing the L-VOD product and using it in applications. Previous studies investigated the correlation between L-VOD and vegetation properties, such as vegetation height and forest biomass, made available by data bases.
In this paper, L-band vegetation optical depth (L-VOD) retrieved by SMOS is compared against vegetation parameters (RH100 and PAI) retrieved by Global Ecosystem Dynamics Investigation (GEDI) lidar instrument, recently launched by NASA. L-VOD was retrieved using the recent v700 version of SMOS level 2 algorithm. In order to manage the different spatial resolutions, GEDI parameters were averaged within SMOS pixels and a threshold to the minimum number of GEDI samples per SMOS pixel was applied. The investigation is multitemporal, since spatial correlations between monthly averages are investigated from May 2019 to April 2020, and a temporal extension to a two year interval is in progress.
The analysis was initially done for four large continents. For Africa and South America, mostly covered by tropical vegetation, the Pearson correlation coefficients between L-VOD and RH100 are higher than 0.8 in all months of the year. Conversely, seasonal effects are observed in North America and Asia, producing a lower correlation coefficient in colder months. RMS differences between L-VOD’s retrieved by SMOS and the ones obtained using a linear regression over RH100 are lower than 0.2 for all cases, and close to 0.1 for most cases. Using PAI in place of RH100 slightly lower spatial correlations are generally achieved.
The analysis was repeated considering three latitude belts: Northern, Tropical, and Southern. In the tropical belt the coefficients of L-VOD versus RH100 regression are stable and the Pearson correlation coefficient is higher than 0.88 for all months of the year. For Northern vegetation the regression slope and the Pearson correlation coefficient are stable from May to September, but decrease in the winter season. Lower Pearson correlation coefficients (about 0.7) are found in the Southern belt, due to reduced dynamic ranges of L-VOD and vegetation height.
All correlation coefficients between v700 L-VOD and RH100 are better with respect to L-VOD from previous level 2 versions. Overall, the obtained results confirm the good potential of L-VOD to monitor vegetation height in different environments. The synergic use of GEDI and SMOS L-VOD data sets can improve the accuracy and/or the timeliness in monitoring vegetation changes occurring at yearly or monthly time scales, such as deforestation, re-growth and desertification.