Remote sensing-based mapping of soil carbon and microbial functional traits

Overview

This solution leverages airborne visible to shortwave infrared (VSWIR) imaging spectroscopy to map soil organic carbon stocks, nutrient profiles, and microbial functional traits across landscapes. By analyzing foliar chemistry captured at high spatial and spectral resolution, the approach provides an indirect method for predicting the vertical distribution and spatial heterogeneity of soil organic compounds without exhaustive ground sampling. The technology addresses a critical gap in understanding the connections between critical zone architecture and the carbon cycle, offering scalable insights into soil respiration, hydrological carbon export, and carbon-atmosphere exchange.

Potential applications include carbon cycle modeling, ecosystem monitoring, land management planning, and environmental assessment for agriculture, forestry, and climate research. The method builds on validated expertise in mapping vegetation type, species, and foliar chemistry from airborne sensors, as well as a novel approach for mapping contaminant distributions through remotely sensed bioaccumulation.

Technical specifications

Key capabilities:

• High-resolution mapping at 3.7-meter spatial resolution using Planet's VSWIR data
• Deep learning model with custom architecture to map carbon, nutrient, and microbial functional trait distributions across VSWIR mosaics
• ACORN 6.5 software for VSWIR data processing and mosaicking
• Contemporaneous ground sampling of foliar, soil, and microbial biology samples collected within 72 hours of VSWIR data acquisition
• Sample stabilization and processing for a variety of assays to validate remotely sensed predictions
• Extensive dataset generation linking foliar chemistry to vertical distribution of soil organic compounds and nutrients

Technology readiness level

The research is at an advanced validation stage. The investigator has already developed and applied novel methods for mapping contaminant distributions through remotely sensed bioaccumulation, demonstrating proof of concept for the broader applicability of these techniques. The proposed one-year research program will produce a comprehensive dataset and validate the deep learning approach for carbon, nutrient, and microbial trait mapping. Future work will refine the methodology and expand applicability across diverse landscape types.