Crop stages and biophysical variables retrieval using VENµS observations

The VENµS optical Earth observation mission is a unique opportunity to evaluate the contribution of dense temporal time series at high resolution to crop monitoring. VENµS’ 5m-2day spatio-temporal resolution permits to characterize subtle changes in crop development.

To take advantage of these observations we aim to assimilate VENµS and Sentinel2 imaging into the SAFYE-CO2 crop model on the SUDOUE site in southern France. However, two steps are necessary to convert the VENµS reflectance data into suitable inputs for our model.

Firstly, we retrieve biophysical variables: LAI and FAPAR. These variables are obtained with the PROSAIL model through a LUT approach. This task also emphasizes the contribution of the B7 to B10 red-edge bands to the retrieval of LAI and FAPAR. Our preliminary results present a high correlation with Sentinel-2S SNAP LAI (RMSE=0.53 R2=0.85).

Secondly, crop calendars are retrieved per plot and main crop type. Our algorithms exploit both the temporal and spectral dimensions of the signal to detect events such as rapeseed flowering and harvest. The features of the VENµS mission increase the probability to detect such events, however this also increases the complexity of the signal compared to other optical missions. The detection of harvest exhibits a high accuracy when compared to expert readings of NDVI time series (80%).

This study will allow us to give an insight into the impact of observational density on the retrieval of crop variables such as yield and carbon storage and on the retrieval of variables and parameters used in crop monitoring.