Chl_a Averages

Please enter here to ask a question about any NASA Science related topics!
Post Reply
r.fernandes
Posts: 27
Joined: Wed Jun 06, 2012 6:55 pm America/New_York

Chl_a Averages

by r.fernandes » Wed Feb 08, 2017 12:38 pm America/New_York

Hi! I'm new to seaDAS 7 and still learning my way so i'm sorry if this is too basic but i need to do 8-day and monthly average of L2_LAC images.
I'm using L2bin on each image (or images when i have more than one for the same day). After i'm using L3mapgen to map them.
I'm going the right way or should i be using other processing tools?

I read all the tutorials and watched the videos but i'm still with some doubts about the way to do this.

I'm using seaDAS 7.3.2 on macOS Sierra.

Thanks in advance!

Tags:

gnwiii
Posts: 606
Joined: Fri Jan 29, 2021 5:51 pm America/New_York
Answers: 1

Chl_a Averages

by gnwiii » Wed Feb 08, 2017 1:14 pm America/New_York

It sounds like you may be planning to average the daily mapped files to get 8-day and monthly mapped files.   You can streamline things a bit by using l3bin to combine daily binned files into 8-day binned files, and then combining three 8-day files with a few daily files to get monthly binned files.   

There are many organizations that don't use binning, preferring to map each level-2 files and then make averages.  This has the advantage that each level-2 file produces a "pretty" map, but also tends to give more weight to pixels near the pass edge (where the ground "footprint" is much larger than at the center of a pass).  Binning a few level-2 files at high resolution will have missing data "speckles"  near pass edges which may obscure patterns in the image .  Mapping fills in these missing data areas, but edge pixels tend to be lower quality (due to longer atmospheric path length), so low quality pixels are over-represented in the "pretty" mapped images.  Binning ensures that each level-2 pixel gets the same weight.

r.fernandes
Posts: 27
Joined: Wed Jun 06, 2012 6:55 pm America/New_York

Chl_a Averages

by r.fernandes » Wed Feb 08, 2017 1:27 pm America/New_York

The region i'm studying is open ocean but close to the Azores islands in the middle of the Atlantic, so many days there is a lot of cloud coverage.
I started by using "Reproject" on the L2 images and then "Level-3 Binning" on the reprojected images to do the averages, but the images looked too "nice" when the cloud coverage is so high, even the 8day averages were almost cloud free. So after reading a little bit more i started l2bin then l3mapgen.
So should i use l2bin on each image, then l3bin to average them and only in the end l3mapgen?
I'm sorry but im really green in all this!

Thank you for the fast answer!

gnwiii
Posts: 606
Joined: Fri Jan 29, 2021 5:51 pm America/New_York
Answers: 1

Chl_a Averages

by gnwiii » Wed Feb 08, 2017 5:52 pm America/New_York

Depending on how you want to use the data, you may want to adjust the list of flags masked by l2bin or l2mapgen (or gpt Mosaic).  

In principle, you can use l2bin with a whole month of level-2 files.  In practice that may hit resource limits.  Most use cases will have files binned for daily and 8-day or quarter-monthly periods.  L3bin exists to combine existign binned files for longer periods such as a month.  The results from binning a months worth of level-2 files (with l2bin) or some combination of level-3 binned files adding up to a month (with l3bin) will be essentially the same (due to numerical rounding errors when calculations are done in different orders).

r.fernandes
Posts: 27
Joined: Wed Jun 06, 2012 6:55 pm America/New_York

Chl_a Averages

by r.fernandes » Thu Feb 09, 2017 6:53 am America/New_York

Thank you that helped a lot!

Another problem, i did the l2bin for each day, then l3bin to do the 8-day bin although an error started to show in l3bin.

The log shows:
execution exception: java.io.IOException: l3bin failed with exit code 1.
Check log for more details.
L3BIN 4.20 (Apr 27 2016 10:27:28)
8 input files
0 /Users/Mogli/Desktop/Imagens/A2015244
1 /Users/Mogli/Desktop/Imagens/A2015245
2 /Users/Mogli/Desktop/Imagens/A2015246
3 /Users/Mogli/Desktop/Imagens/A2015247
4 /Users/Mogli/Desktop/Imagens/A2015248
5 /Users/Mogli/Desktop/Imagens/A2015249
6 /Users/Mogli/Desktop/Imagens/A2015250
7 /Users/Mogli/Desktop/Imagens/A2015251
irow:     0 of    17280 Thu Feb  9 10:40:01 2017
irow:   500 of    17280 Thu Feb  9 10:40:01 2017
irow:  1000 of    17280 Thu Feb  9 10:40:01 2017
irow:  1500 of    17280 Thu Feb  9 10:40:01 2017
irow:  2000 of    17280 Thu Feb  9 10:40:02 2017
irow:  2500 of    17280 Thu Feb  9 10:40:02 2017
irow:  3000 of    17280 Thu Feb  9 10:40:02 2017
irow:  3500 of    17280 Thu Feb  9 10:40:02 2017
irow:  4000 of    17280 Thu Feb  9 10:40:02 2017
irow:  4500 of    17280 Thu Feb  9 10:40:03 2017
irow:  5000 of    17280 Thu Feb  9 10:40:03 2017
irow:  5500 of    17280 Thu Feb  9 10:40:03 2017
irow:  6000 of    17280 Thu Feb  9 10:40:03 2017
irow:  6500 of    17280 Thu Feb  9 10:40:04 2017
irow:  7000 of    17280 Thu Feb  9 10:40:04 2017
irow:  7500 of    17280 Thu Feb  9 10:40:04 2017
irow:  8000 of    17280 Thu Feb  9 10:40:04 2017
irow:  8500 of    17280 Thu Feb  9 10:40:04 2017
irow:  9000 of    17280 Thu Feb  9 10:40:05 2017
irow:  9500 of    17280 Thu Feb  9 10:40:05 2017
irow: 10000 of    17280 Thu Feb  9 10:40:05 2017
irow: 10500 of    17280 Thu Feb  9 10:40:05 2017
irow: 11000 of    17280 Thu Feb  9 10:40:06 2017
irow: 11500 of    17280 Thu Feb  9 10:40:06 2017
irow: 12000 of    17280 Thu Feb  9 10:40:06 2017
irow: 12500 of    17280 Thu Feb  9 10:40:06 2017
irow: 13000 of    17280 Thu Feb  9 10:40:07 2017
irow: 13500 of    17280 Thu Feb  9 10:40:07 2017
irow: 14000 of    17280 Thu Feb  9 10:40:07 2017
irow: 14500 of    17280 Thu Feb  9 10:40:07 2017
irow: 15000 of    17280 Thu Feb  9 10:40:08 2017
irow: 15500 of    17280 Thu Feb  9 10:40:08 2017
irow: 16000 of    17280 Thu Feb  9 10:40:08 2017
irow: 16500 of    17280 Thu Feb  9 10:40:08 2017
irow: 17000 of    17280 Thu Feb  9 10:40:09 2017
Error reading output file binlist records

What am i doing wrong?

Also if i l2bin all the images for the 8days (where some days have more than one image) together will it be a a good 8day average or do i need to keep doing the daily averages and then use them for the 8day average? Will the outputs be the same like the example you said?

Thank you so much for all the help!

Using seaDAS 7.3.2 macOS Sierra, all processors updated last week.

gnwiii
Posts: 606
Joined: Fri Jan 29, 2021 5:51 pm America/New_York
Answers: 1

Chl_a Averages

by gnwiii » Thu Feb 09, 2017 7:28 am America/New_York

You may see more helpful error messages by running l3bin in a terminal. 

For details of the binning process, see IOCCG Report 4.  Level-3 binned files store sums, counts, etc. needed to compute averages, but the actual averages are computed following the algorithm described in Report 4 (and included references) by l3bmapgen or l3bindump.  There is one potentially important detail that you don't get -- the distribution over time of the pixels that have data.  For, say, a monthly average in a region with lots of cloudy days, it is possible that the cloud-free pixels that contribute to a monthly binned file occurred early or late in the month, so if there is a time trend in the data the monthly mean will be under- or over-estimated.

r.fernandes
Posts: 27
Joined: Wed Jun 06, 2012 6:55 pm America/New_York

Chl_a Averages

by r.fernandes » Fri Feb 10, 2017 4:57 pm America/New_York

Hi,

Thank you so much for your help!
Everything is running smothly now. I was going to ask exactly what you just explained about the average. I was thinking that the pixels needed to be mapped first for the average to be accurate. Again thank you for explaining it!

I'm also working with SST and when i try to run l3mapgen on the averages it gives an error. I did exactly the same procedure that i used with the Chlr. L2bin-->l3bin-->l3mapgen. Do i need to do anything different?

gnwiii
Posts: 606
Joined: Fri Jan 29, 2021 5:51 pm America/New_York
Answers: 1

Chl_a Averages

by gnwiii » Sat Feb 11, 2017 7:24 am America/New_York

The same basic outline applies to SST. Usually when some workflow that was runing properly gives an error when you use different data there is a type on the changes or a resource problem (no space left on device). You need to provide details of the error.  It is generally easier to post output from a terminal session -- the GUI makes it much harder to write down the exact steps you used and tends to hide key details.

r.fernandes
Posts: 27
Joined: Wed Jun 06, 2012 6:55 pm America/New_York

Chl_a Averages

by r.fernandes » Fri Feb 17, 2017 11:30 am America/New_York

I'm so sorry for the late late response!

i just needed to add sst to the products on the l3mapgen. Everything is running perfectly now, thank you for you help!

Post Reply