IOTA2 withS1&2 over Indian agro-systems
Dear IOTA2 developers team,
I thank you in advance for helping me on this issue.
I am trying to characterize rice area extent using S1&2 during Indian monsoon period (Thick cloud cover from mid-June to end of October), when you can barely expect a couple of cloud free S2. Rice is generally sown in July and harvested in October.
I have been preparing with the help of some of your team members the attached config files to launch iota2 on Hal.
step1: Amalthee for S1&2; with @thierionv
step2: Launch of iota2 init to init with only S2; with @ArthurV module load iota2 mkdir /work/CESBIO/users/ferrants/S2_i2 Iota2Cluster.py -config /work/CESBIO/users/ferrants/SARclassif/ConfigS2init.cfg -config_ressources /work/CESBIO/users/ferrants/ressources_iota2.cfg -scheduler_type PBS -nb_parallel_tasks 13
step3: Launch of iota2 init to validation with S1 and S2; with @thierionv Iota2Cluster.py -config /work/CESBIO/users/ferrants/SARclassif/ConfigS1and2.cfg -config_ressources /work/CESBIO/users/ferrants/ressources_iota2.cfg -scheduler_type PBS -nb_parallel_tasks 13
This command stops without explicit error message after a varying time even after being launched with -restart (help on this of @ArthurV ).
IOTA2 has been used with S2 only on monsoon and dry periods (2016 to 2019) to do this seasonal rice mapping (Claire pascal work) but F-scores for monsoon maps are too low to be acceptable, and should be improved with the input of S1, as previously shown in Ferrant et al., 2017 (https://doi.org/10.3390/rs9111119).
I wish you a pleasant day! Sylvain