Welcome to the Earthdata Forum! Here, the scientific user community and subject matter experts from NASA Distributed Active Archive Centers (DAACs), and other contributors, discuss research needs, data, and data applications.
by africaflores » Wed Sep 12, 2018 12:59 pm America/New_York
Hi,Is it possible and if so what are the steps to update Chl-a algorithm coefficients based on in situ observations in SeaDAS?I just see tools to compare Chl-a values between satellite image and in situ observations. There is also a tool to extract pixel values based on coordinates. With that info, in theory we could do the polynomial regression in excel or R. Thank you in advance for your guidance. Best regards,Africa F.
Africa,It is possible to create custom algorithms, using the Band Maths feature. If you have sufficient in situ observations to relate remote sensing reflectance to chlorophyll concentration, you can define your own algorithm. If you wish to "update" the standard product, that is a wee bit more tricky. If your algorithm follows the same functional form as the chl_ocx (e.g. OC2, OC3, etc) algorithms, you can compute your own coefficients (outside of SeaDAS - its a visualization/data processing tool not an curve fitting tool) and process the L1 data to L2, passing the coefficients to l2gen via two parameters: chloc2_wave and chloc2_coef (or chloc3_wave and chloc3_coef for the OC3 variant). The default values for these are in the msl12_defaults.par file:# algorithm controls# ------------------chloc2_wave=[482,561]chloc2_coef=[0.1977,-1.8117,1.9743,-2.5635,-0.7218]chloc3_wave=[443,482,561]chloc3_coef=[0.2412,-2.0546,1.1776,-0.5538,-0.4570]Sean
by lmckinna » Sat Sep 15, 2018 3:24 am America/New_York
I presume you are doing some sort of regional study and I do have some thoughts regarding your query. Before I give an extended response, it would be useful to get an idea of where you're looking and what your objectives are.
1. Are you working in optically complex and/or shallow conditions (i.e. coastal waters)? 2. Are your in situ Chl data points sampled routinely in the same location/site, or are they randomly distributed? 3. Are you planning to use the re-tuned Chl algorithm do a time-series analysis or just study a handful of snapshots in time?
by africaflores » Sun Sep 16, 2018 12:21 pm America/New_York
Thank you for the interest. Kindly find below answers to your questions.
1. Are you working in optically complex and/or shallow conditions (i.e. coastal waters)? R/ I'm working in an in-land water body in Central America, it is deep, about 300m at its deepest point, but relatively small in surface area (~130 sq km). During dry season waters compare to ocean waters, they are very clear, it's consider an oligotrophic water body during this season. Main constituent that drives the color of the water is phytoplankton. It has experienced sporadic algal blooms caused by cyanobacteria.
2. Are your in situ Chl data points sampled routinely in the same location/site, or are they randomly distributed? R/ There are not consistent in situ Chl data points for this lake, I have some old measurements (one time type of information) that I can only correlate with older Landsat missions and ALI data (I do have overlaps). I would be interested in using l2gen to create atoms correction products, but don't know how to process these other sensors in l2gen, I would like to know if this is possible. BTW, in my previous research I have used other atoms correction methods for this purpose (6S) with specific parametrization over water and tropics.
3. Are you planning to use the re-tuned Chl algorithm do a time-series analysis or just study a handful of snapshots in time? R/ Our interest is to re-tuned Chl algorithm for an specific water body, in case enough in-situ data exists. I have already done that using a variety of GIS and python packages and I'm interested in knowing at which extent that process can be done in SeaDAS.
by lmckinna » Wed Sep 19, 2018 12:56 am America/New_York
Hmmm, as I'm sure you're aware applying ocean color data with ~1km pixel resolution to lakes can be problematic. My suggestion was going to be to empirically scale the standard chlorophyll algorithm for MODIS using in situ data rather than retuning the OC algorithm coefficients. However, if you don't have measurements with coincident MODIS data that won't be possible. In addition, I presume as you are referring to Landsat and ALI because the lake is too small for MODIS/SeaWiFS. Unfortunately, I don't believe that ALI data processing is supported in SeaDAS.
Perhaps you might consider contacting someone from the GloboLakes project: http://www.globolakes.ac.uk/, they may have a better appreciation of the problem you're trying to solve and might be able to provide you with some further guidance.