The level 4 FINCH-Science-Mission for FINCH requires our ability to successfully estimate the amount of crop residue (AKA non-photosynthetic vegetation or NPV) contained within agricultural areas before and after tillage events. Using hyperspectral data, the process of crop residue estimation falls under the umbrella term of Spectral Mixture Analysis (SMA) (See https://drive.google.com/file/d/1Z17w0xR-NNGIogGzgIy-R_kDDA1L2T_W/view). There are dozens of approaches to crop residue estimation using SMA. A long term analysis of different methods and databases will determine the ideal methods and scenarios for NPV unmixing using the FINCH satellite. An overview of our architecture can be found here Abundance Estimation Pipeline Architecture. Our spectral library for analysis can be located here Build Preliminary Spectral Library.
Results go here, results go up top
FAE: Fractional Abundance Estimation
SOO RMSE: Single Objective Optimization, Root Mean Squared Error
SFA: Spectral Feature Analysis
RF: Random Forest
BF: Brute Force
NN: Neural Network
MLP: Multi Layer Perceptron
CNN: Convolutional Neural Network
NIF: Neural Implicit Flow
EC: Endmember Classification
KNN: K-Nearest Neighbors