Project Number: 5430-44000-015-23
Project Type: Reimbursable

Start Date: Sep 01, 2008
End Date: Sep 30, 2009

Objective:
1) Develop techniques and provide databases for rapid expansion of Official NIRS spectral databases for corn and soybean proximate analysis, 2) study methodology and provide databases for measurement of specialized factors in corn and soybeans ( presently corn fermentable starch and amino acids; soybean fatty acids), and 3) work with GIPSA personnel to evaluate new NIRS calibration methods (eg Support Vector Machines). Ongoing performance data for new technologies not currently in the Official System will also be shared with GIPSA.

Approach:
We will provide to GIPSA the Iowa State Infratec corn and soybean proximate analysis databases current through the 2007 crop by September 30, 2008. We will collect data for 2008 crop year samples and provide to GIPSA. GIPSA will have unrestricted use of the Iowa State database. Provide 25-30 samples of high moisture corn (25% and above) to GIPSA at harvest for use in calibration programs. Provide 10-20 seed samples of 50 or more that are well characterized (for the above listed factors) bulk samples per grain. We will review with GIPSA ongoing results relative to calibration for corn and soybean subunit factors (e.g., amino acids, fatty acids) and other factors of potential interest to GIPSA. We will provide to GIPSA the Iowa State databases for the corn and soybean specialty factors. We will develop potential NIRS calibrations or calibration validations that could be used if these factors become Official criteria or services. We will review the technologies, NIR and other, from the viewpoint of the requirements of the Official Inspection System. We will work with GIPSA to share knowledge and develop calibrations using techniques not used in the Official system: Support Vector Machines, Locally weighted regression, transfer of bulk grain calibrations to single seed spectra.