Welcome to the NOAA Institutional Repository | Neural network technique for gap-filling satellite ocean color observations - :12149 | National Weather Service (NWS)
Stacks Logo
Advanced Search
Select up to three search categories and corresponding keywords using the fields to the right. Refer to the Help section for more detailed instructions.
 
 
Help
Clear All Simple Search
Advanced Search
Neural network technique for gap-filling satellite ocean color observations
  • Published Date:
    2016
Filetype[PDF - 1005.50 KB]


Details:
  • DOI:
    doi:10.7289/V5/NCEP-ON-483
  • Corporate Authors:
    National Centers for Environmental Prediction (U.S.)
  • Series:
    Office note (National Centers for Environmental Prediction (U.S.)) ; 483
  • Document Type:
  • Description:
    Integrating/assimilating satellite ocean color (OC) fields (chlorophyll-a, Kd490, KdPAR) in NOAA's operational ocean models requires scientifically consistent and robust techniques to address temporal and spatial gaps in data, especially gaps longer than a few days. In this work, we introduce one possible approach based on a Neural Network (NN) gap-filling technique, linking OC variability, which is primarily driven by biological processes, with the physical processes of the upper ocean. A NN method for correlating satellite OC fields with other assimilated satellite and in situ observations: a) instigates fewer assimilation errors (since the inputs to the NN are already being assimilated) and b) reduces reliance on sparse in situ OC observations. In this study, satellite-derived surface variables (sea-surface temperature (SST), sea-surface height (SSH), and sea-surface salinity (SSS) fields) and gridded ARGO salinity and temperature profiles, from 0 to 75m depth, are employed as signatures of upper-ocean dynamics. Chlorophyll-a (Chl-a) fields from NOAA's operational Visible Imaging Infrared Radiometer Suite (VIIRS) are used for NN developments, as well as NOAA SSH and SST fields and NASA Aquarius mission SSS fields. The OC data correlations with the satellite SSH/SST/SSS fields are spatially and temporally dependent. The NN technique is trained using data for two years (2012 and 2013) and tested on the remaining year (2014). Results are assessed using the root-mean-square error (RMSE) and cross-correlations between observed OC fields and NN output. To reduce the impact of noise in the data while obtaining a stable computation of the NN Jacobian for sensitivity studies, an ensemble of NN with different weights is constructed. The results for the ensemble mean are compared with those for a single NN. The long-term objectives with respect to these NN efforts are to significantly expand the utility and use of satellite ocean color fields (Chl-a, Kd490, Kdpar) by using neural network techniques to 1) develop complete and consistent global ocean color fields using satellite observations, addressing gaps, such as those resulting from swath ground track separation, obscured areas, and satellite data loss and 2) statistically predict satellite-derived ocean color fields for numerical prediction applications. This new gap-filling capability will enable the assimilation of near-real-time (NRT) ocean color data into NOAA's operational numerical modeling to address a bio-physical feedback process that is particularly important to ocean atmosphere coupled modeling. The assimilation of ocean color data also drives/constrains modeled physical-biogeochemical processes that underlie ecological forecasting. Efforts to incorporate biogeochemical components into NCEP operational global ocean models have already commenced. While this note details preliminary work, a subsequent optimal configuration of the NN technique will be assessed using extensive experimentation, sensitivity tests, statistical metrics, using ocean modeling for validation. [doi:10.7289/V5/NCEP-ON-483 (http://doi.org/10.7289/V5/NCEP-ON-483)]

  • Supporting Files:
    No Additional Files