Hi Karnan,The ensemble capability should correctly handle the missing flags. Using the datasets you attached to your previous email I am not seeing a problem with the flags being used in the average.
NOAA/PMEL TMAP FERRET v7.1 (optimized) Linux 2.6.32-642.13.1.el6.x86_64 64-bit - 02/15/17 20-Jun-17 09:30 yes? ensemble/TITLE="kd" KD490_aggr = 1.nc,2.nc,3.nc,4.nc yes? shade MODISA_L3M_KD_2014_KD_490[d=KD490_aggr,m=@ave]I've attached the plot. Opening the files first and then defining the ensemble is equivalent and does not give a different result. What version of Ferret are you running? I believe this may not have been handled correctly when the ENSEMBLE capability was first introduced.
Ansley On 6/19/2017 10:20 PM, Karnan C wrote:
Thank you Dr Russ, The suggested ensemble (DEFINE/DATA) method is working well. yes? use 1.nc , 2.nc , 3.nc , 4.nc yes? DEFINE DATA/AGGR/E/TITLE="kd" KD490_aggr= 1, 2, 3, 4 yes? shade MODISA_L3M_KD_2014_KD_490[d=KD490_aggr,m=@ave] Here the fill value (-32767) also took role in deriving average, though my actual data range is 0.01 to 5. Then I used the following scripts to change the fill value for each file yes? set var/bad=0 MODISA_L3m_KD_2014_Kd_490 Then I followed above method (ensemble). Now I got the spatial average. Thank you very much. 2017-06-16 11:43 GMT+05:30 Russ Fiedler <russell.fiedler@xxxxxxxx>:Hi, Try making an ensemble. This will make virtual dataset using the E/M dimension ENSEMBLE/title="KD490 Ensemble" kd490_ens=1,2,3,4 show data kd490_ens shade MODISA_L3M_KD_2014_KD_490[d=kd490_ens,m=@ave] Look up DEFINE/DATA in the documentation for more info. Russ On 16/06/17 15:24, Karnan C wrote:Hi, I tried to derive spatial average of KD490 from four different files (1.nc, 2.nc, 3.nc and 4.nc downloaded from giovanni database) \\I used the following stpes yes? use 1.nc , 2.nc , 3.nc , 4.nc yes? sh da currently SET data sets: 1> ./1.nc name title I J K L M N MODISA_L3M_KD_2014_KD_490 Diffuse attenuation coefficient 1:156 1:96 ... ... ... ... 2> ./2.nc name title I J K L M N MODISA_L3M_KD_2014_KD_490 Diffuse attenuation coefficient 1:156 1:96 ... ... ... ... 3> ./3.nc name title I J K L M N MODISA_L3M_KD_2014_KD_490 Diffuse attenuation coefficient 1:156 1:96 ... ... ... ... 4> ./4.nc (default) name title I J K L M N MODISA_L3M_KD_2014_KD_490 Diffuse attenuation coefficient 1:156 1:96 ... ... ... ... yes? fill MODISA_L3M_KD_2014_KD_490[d=1] yes? frame/file=Figure1.gif //Figure1 in the attachement yes? fill MODISA_L3M_KD_2014_KD_490[d=2] yes? frame/file=Figure2.gif //Figure2 in the attachement yes? fill MODISA_L3M_KD_2014_KD_490[d=3] yes? frame/file=Figure3.gif //Figure3 in the attachement yes? fill MODISA_L3M_KD_2014_KD_490[d=4] yes? frame/file=Figure4.gif //Figure4 in the attachement yes? fill (MODISA_L3M_KD_2014_KD_490[d=1]+MODISA_L3M_KD_2014_KD_490[d=2]+MODISA_L3M_KD_2014_KD_490[d=3]+MODISA_L3M_KD_2014_KD_490[d=4])/4 yes? frame/file=Figure5.gif //Figure5 in the attachement //In the image, most of the regions were empty while taking the average (the method read from ferret user group). //It was because of the missing data in any of the file (eg. 4.nc), led the empty regions, though other files having values. I should take the average from these files. How can I solve this? Is it possible to replace the 'no data (....)' as 'zero (0)'? or Is it possible to feed K and L axis with some values like L value = 1,2,3 and 4 for each .nc files, respectively? May it be useful to take average like "yes? fill MODISA_L3M_KD_2014_KD_490 [l=1:4@AVE]". The sample data files attached here. Thanks. -- C. Karnan, CSIR - SRF National Institute of Oceanography (CSIR) India.
Attachment:
ensemble.gif
Description: GIF image