Argentina

Specific Project Objectives & Deliverables

Intercomparisson of Methodologies for cropland Mapping (Waldner et al, 2016)

Five methodologies developed for different JECAM sites (Argentina, Belgium, China, Russia and Ukraine) were tested over 5 locations (Argentina, Brazil, China, Russia and Ukraine). In situ (in field collected) data was collected by JECAM site partners and used for training and validation of classifiers.  

Results showed quite high overall accuracy for all methods (in general more than 90%). Higher variability was observed among sites than among methods. Russia and Ukraine showed the highest accuracy values, while Brazil the lowers. Some problems were observed to classify no cropland areas due to similarity with cropland fileds and spatial distribution of land uses. 

A test was performing reducing the sample size for training up to 10 % and showed low reduction in overall accuracy when reduction was higher than 40 %.

©2015 Joint Experiment for Crop Assessment and Monitoring
© HER MAJESTY THE QUEEN IN RIGHT OF CANADA SA MAJESTE LA REINE DU CHEF DU CANADA (2015)