High spatial resolution gap free biophysical variables for Earth Observation at continental scales with Google Earth Engine
Abstract
Earth vegetation plays an essential role in the study of global climate change influencing terrestrial mass, energy fluxes and variability through plant transpiration and photosynthesis. Within this context, biophysical variables such as the Leaf Area Index (LAI) and the Fraction of Absorbed Photosynthetic Radiation (FAPAR) are key inputs in a wide range of ecological, meteorological and agricultural applications and models. Therefore we propose a methodology based on a hybrid method which inverts a radiative transfer model (PROSAIL) and a cloud optimized fusion algorithm (HISTARFM). This combines MODIS and Landsat information to produce 30 m resolution gap-free biophysical variables (FAPAR, LAI, FVC, CWC) at continental scales. To invert PROSAIL, we have tested two machine learning approaches: Artificial Neural Networks (ANN) and Gaussian Processes (GPs). In addition, HISTARFM provides realistic uncertainty estimates along with the fused reflectances. This information has been propagated through the models to obtain both accurate biophysical estimates and realistic uncertainties. They go beyond the classical and oversimplified quality control flags provided with most of the state of the art operational products. We have taken into account the aleatoric uncertainty (data error) that needs to be propagated through the nonlinear function implemented by the machine learning methods used, and also the epistemic uncertainty (model error). This study is carried out over the contiguous US (CONUS) area with Google Earth Engine (GEE), a cloud computing platform specifically designed for geospatial analysis at the petabyte scale. The proposed retrieval methodology combined with the immense GEE computational power allows one to obtain high spatial resolution biophysical products to achieve the high spatial detail needed to adequately monitor croplands and heterogeneous vegetated landscapes at very broad scales.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMB040...07M
- Keywords:
-
- 0410 Biodiversity;
- BIOGEOSCIENCES;
- 0439 Ecosystems;
- structure and dynamics;
- BIOGEOSCIENCES;
- 0466 Modeling;
- BIOGEOSCIENCES;
- 0480 Remote sensing;
- BIOGEOSCIENCES