I want to download MODIS chlorophyll and SST data within California Gulf region from 2000-2012 (mainly focus on 2006-2008). I am looking at the change of these two indices with time for given pixel within California Gulf. I have tried download one zip file and used ENVI to open it directly as an "image". It seems everything is loaded, such as metadata, lat/long/calibration info, and hdf image, etc.
Here is the questions:
1. How can I download at once the data matching the requirements? (same location but different time quickly and conveniently)
2. I have problem in finding the location by searching "HTTP-based data distribution server" for the file names has no indication of location.
3. Can I just get the CH or SST image only within the file downloaded, how can I extract hdf files from this combined file easily?
Thanks a lot.
What data are you using (individual level-2 passes or level3, binned or mapped)? Have you looked at Giovanni? It uses binned data, so doesn't get to individual pixel level of detail, but in practice the biggest drawback is that Giovanni uses rectangular areas. This tends to increase variability over, say, a circular region of the same surface area. Giovanni provides a quick look at the data, but for more careful investigation I prefer extracting values from the daily level-3 binned files using a version of l3bindump.c (source included in SeaDAS) modified to write a format that can be inported into a stats program. For use with time-series analysis, the daily data are generally averaged over some time period longer than a day, but at the daily level you can see issues such as large spikes associated with extreme values on cloud edges and drop those days from the analysis.
Thank you for reply. Could I ask more questions: what is the difference about binned and mapped and how to process these data downloaded?
I initially tried downloading L2 because they've got CH and SST wihin file. But I just found those mapped data sorted by type SST and CHL. It is not necessary to get daily data (8days or monthly also work) and what I need is to observe the CHL and SST change within Gulf of california through time (2000-2012).
I have downloaded one bz2 file and extract, but I have no idea how to open it (with ENVI). Could you help me with this? I was also wondering if they are global data so I have to subset the gulf of california?
Below is the path and file name:http://oceandata.sci.gsfc.nasa.gov
/ MODISA / Mapped /Monthly /9km /SST
when extracted, I got this file, how to open the image within this file?
Thanks a lot.
The L3m files are global, using hdf4 but with metadata specific to SeaDAS. ENVI supports some SeaDAS formats, but I don't think it handles L3m metadata, so you have treat them as generic HDF4 file (not too bad, as mapped data are just an array). You will need to apply scale factors and navigation.
Otherwise you would need to install SeaDAS or BEAM. I suggest you start with Giovanni, which will very quickly allow you to create a time series for the spatial mean over a specified rectangle to get a feel for the data, then decide if you want to invest the time to install the software.
The L3b refers to binned data, which has statistical advantages (you can get means and variances for the data in each bin). There is a IOCCG report devoted to binning. To work with binned data you really do need specialized software, e.g., SeaDAS or BEAM as the binning scheme is not widely used.
Hi, Thank you so much for the detail explanations. I just downloaded some binned and mapped data today and some of them could be viewed thru ENVI, they look fine to me.
Also, as you suggested, I have searched Giovanni, it does provide with important information that I want! I will learn more by starting with Giovanni. Thanks a lot!
Can I ask you more questions?
So far I have downloaded L3M SST and CHl, 4km & monthly data through 2002 to 2012. I could view them with ENVI. Now I am trying the layer stack them and then subset all the layers to gulf of california. But when I click the image, it shows the input file is not georeferenced. Is this the problem related to what you indicated about L3m files? Do you have any idea about this and how to fix it, such as to get them georeferenced, stack to as bands? Thanks a lot.
You need to manually provide the georeference information. I do this using gdal's VRT files to create geotiffs as everything can done via shells scripts to process masses of files with little effort, and the geotiffs can be used with BEAM/SeaDAS 7 and other apps but ENVI may be be able to georeference your stack without manually processing the individual files.
Would you be kind to instruct me how to use the gdal VRT files to process with masses of images? And how to set up those projection information properly for my objectives? Would it be possible to share me with the scripts? I really have a hard time to keeping doing for each of the images.
Also, the last sentence about ENVI, do you mean ENVI could process georeference easily with itself? But I tried with ENVI and failed to figure out this projection problem. I don't think I will use BEAM/SeaDas or other software, and I hope just ENVI could do these straightforward stuff.
For VRT files, I follow http://www.gdal.org/gdal_vrttut.html
to get a
for one test file. The SRS for the cylindrical equidistant projection is:
is needed to pass the SRS quotes via XML. You also have to work out geotransform values for your images.
One you have a working example, it is not hard to create a shell script that adusts the filename to create
files for a list of input files, and use gdal_translate to get geotiffs that work with visat and ENVI. If you only use ENVI you may be able to use ENVI .hdr Labelled Raster. I checked with a colleague who uses ENVI much more than I do -- he writes IDL code to load one file with georeferencing, then runs the same code for his list of files. We aren't using l3map files, so you will need to work out the details, but it should not be hard to get one file georeferenced in a way that uses a command line (shell or IDL) so you can apply the same commands to multiple files.