Exploring the impact of varied background on quantification of soil carbon content using microwave and millimeter wave signal reflectance

The quantification of soil carbon content is paramount to the advancement of soil carbon management practices, serving as the bedrock for the development and implementation of various carbon-negative and carbon-neutral technologies. These technologies are crucial in the battle against climate change and in enhancing soil health to boost agricultural productivity.  Therefore, we employ an innovative sensor technology comprised of a compact array, including 18 pairs of radar transmitters (TX) and receivers (RX) within a microwave radar array sensor, alongside a configuration of 20 TX and 20 RX pairs in a millimeter wave radar array sensor. For this kind of setup to work, the sensors need to be carefully tested to make sure they can capture changes in the soil's carbon content over time and space.

This dataset presents an exhaustive series of raw data aimed at evaluating the efficacy of these two sensor technologies in detecting soil carbon content. A distinctive aspect of this study is the examination of sensor performance across four divergent backgrounds—glass, metal, sponge, and wood surfaces—to ascertain the influence of background material on radar wave reflectance and, consequently, measurement accuracy. Furthermore, the investigation extends to the analysis of sensing distance as a critical parameter for classification accuracy, with the microwave radar array sensor tested across distances ranging from 1 inch to 4 inches and the millimeter wave radar sensor evaluated at intervals of 2.5 inches, 5 inches, 7.5 inches, and 10 inches. By providing a comprehensive dataset on the performance evaluation of cutting-edge sensor technologies, this study aims to facilitate further advancements in the field of soil carbon management, ultimately supporting the global effort to combat climate change through innovative and sustainable agricultural practices.

Methods
Data Collection
Microwave Sensor:

Frequency and Sampling: The microwave sensor operated at a transmission frequency of 109.2 GHz, with each sample undergoing 20 iterations of data collection to mitigate random noise interference.

Antenna Configuration: The sensor configuration included 4 transmitters and 14 receivers, yielding 40 unique antenna pair combinations for comprehensive signal capture. The antenna pair combinations are in the readme file.

Signal Features: Each signal acquisition resulted in 8192 distinct signal features, facilitating detailed analysis.

Millimeter Wave Sensor:

Frequency Range and Sampling: The millimeter wave sensor’s operational frequency spanned from 62 GHz to 69 GHz, adopting an identical sampling strategy (20 times per sample) to ensure consistency in error reduction.

Resolution and Sampling Points: Across the specified frequency range, 150 sampling points were established, with a resolution bandwidth (RBW) of 100 kHz chosen to balance sweep time and noise reduction.

Antenna and Resolution Profiles: The sensor comprised 20 transmitters and 20 receivers, with data collected under three resolution profiles (high, medium, and low) utilizing configurations of 4, 10, and all 20 transmitters, respectively.

Signal Features: The millimeter wave sensor captured 1024 features per antenna pair, enabling precise soil carbon content analysis.

Sample Preparation and Testing Environment:

Soil and Biochar Mixture: Samples consisted of soil collected from almond tree fields, cleared of coarse matter like stones, and mixed with biochar produced from 100% pinewood. The mixtures varied in biochar content across six classes, ranging from 0% to 100%.

Backgrounds and Sensing Distance: Data collection was conducted against four different backgrounds (glass, metal, sponge, and wood) to assess the impact of environmental interference on signal accuracy. The microwave sensor testing distances spanned from 1 to 4 inches, while the millimeter wave sensor was evaluated at distances of 2.5, 5, 7.5, and 10 inches from the sample.