The use of a cosmic ray probe as a proxy of green vegetation biomass

Abstract

A preliminary study is undertaken to determine whether the fast neutron counts of a cosmic ray probe can be used as proxy estimate of green biomass over its 40 hectare measurement area. The study was conducted using the Normalized Difference Vegetation Index (NDVI) product from NASA MODIS satellite imagery and pressure corrected fast neutron counts of a cosmic ray probe located at Tullochgorum in north-eastern Tasmania between October 2010 and December 2013. Machine learning based regression models, namely, Support Vector Regression (SVR), Generalized Linear Model (GLM), Regression Decision Tree, Multi Layer Perceptron (MLP) network and Radial Basis Function (RBF) network were employed to estimate the NDVI from the dependent variable of fast neutron counts across a range of input configurations. Results from this study showed the relationship between wet soil and increased vegetation density and greenness could be used to provide some form of proxy estimate of green biomass at a weekly time resolution. The model with the highest accuracy was an MLP network (Pearson's correlation coefficient of 0.86) with inputs composed of the previous 12 weeks of averaged fast neutron counts.