TREE-WIND

1- NAM
   - Nowcast is downloaded by N_S1_gridded.sh for both meshes
   - Forecast is downloaded by F_S1_gridded.sh after GEFS and SREF download completion

2- GEFS
   - FORECAST, downloaded after nowcast simulations completion and before any forecast simulation
     March2019: the serial download version is not up to date.
                GEFS forecast hour starts at the same time as NAM
     index is downloaded to get the issued cycles in the day.
     the last line is the latest cycle.
     March2019: no need no line for checking with the old cycle. old cycle checking is redundant here
                as the code is activated when there is a new NAM, so there should be a new SREF/GEFS

3- SREF
   - FORECAST, downloaded after nowcast simulations completion and before any forecast simulation
     March2019: the serial download version not exists          
                SREF forecast hour starts at the same time as NAM. since it is 3 hours off, we use an old cycle.
     index is downloaded to get the issued cycles in the day.
     the last line is the latest cycle.
     6 hours is subtracted to get the cycle issued before the latest
     This old one is used starting at 03.
     for 03 cycle, today date is updated to yesterday to get the 21 cycle.
     March2019: no need no line for checking with the old cycle. old cycle checking is redundant here
                as the code is activated when there is a new NAM, so there should be a new SREF/GEFS

NOTE    
     GEFS and SREF headers start six hours before the NAM header 
     IS IT Okay? Seems no for 1st inpect
   
     and why NAM is like 6 hours ahead of GEFS? at least in naming cycle?
   
     Keep this line if everything is fine after troubleshooting
     NAM cycle is named as GEFS cycle + 6 hour
     SREF cycle starts 3 hours after GEFS.
  
      
     1st inspect: cycle 00 of march 4. New method shows 6 hours lag. 
                  headers shows 6 hour lag as well.
                  An old cycle in which all were in phase inspected. Headers had the same start time indicating old method was right
                  The cycle that showed the 6 hour diff and made me apply these changes inspected. headers were 6 hour diff.
                  why? GEFS is downloaded much earlier than NAM, what about SREF? SREF was downloaded after NAM.
     2nd inspect: cycle 12 of march 4


     Everything get back to what it used to be, GEFS starts at 06 and SREF starts at 03
     3rd inspect: cycle 00 of march 5


     4rd inspect: cycle 12 of march 5
               

     possible conclusion:
                         NAM is issued sooner than any cycle. When capturing cycle XX for nam, cycle XX-1 is available for GEFS
                         Thus, should get the XX-1 of GEFS and start from 06.

                         probable issues: 
                                        old method in terms of when each is downloaded. GEFS during nowcast, then NAM, then SREF
                                        NAM nowcast is downloaded and the cycle number is saved into current cycle.
                                        Then, nowcast sim is run and GEFS is downloaded. GEFS has issued for this cycle at the same time
                                        as nam, or in the meantime, GEFS cycle has been updated. For any reason, GEFS and NAM has the same
                                        cycle, therfore, starting downloading GEFS from 006 causes a lag of 6 hours.
                                        It is possible that even NAM has been updated, but the system download for the same cycle as nowcast
                                        by reading from current.cycle file. SREF and GEFS should do the same, I guess.