A Novel and Efficient Spatiotemporal Oxygen Production Estimation Based on Land Vegetation Using PNN and Convolutional Autoencoder

Abstract

Oxygen is a sensitive indicator of atmospheric compositional changes and is also a primary requirement for human life. With deforestation on the rise, dwindling level of oxygen concentration in the air is a concern for mankind. Our paper aims to estimate the oxygen production levels of a particular area of land captured by satellite imagery. Our algorithm aims to identify forests and agricultural land patches and analyzes the oxygen production in a very efficient manner. Further, the algorithm takes in images of leaves from these areas and processes them to identify the species, chlorophyll content, and nitrogen levels in the plant using feature selection and probabilistic neural networks (PNN). We have computed the oxygen level of the patch of land and these calculations were performed on images spanning over few years allowing us to calculate the changes in oxygen production. This not only helps to map the carbon footprint, but also acts as a curative measure for global warming.