Skip to main content
Dryad

Microwave and millimeter wave signals reflectance of soil carbon content

An, Di1; Chen, YangQuan1

Research facility: University of California, Merced
Published May 24, 2023; Updated Apr 15, 2024 on Dryad. https://doi.org/10.6071/M3M092

Cite this dataset

An, Di; Chen, YangQuan (2024). Microwave and millimeter wave signals reflectance of soil carbon content [Dataset]. Dryad. https://doi.org/10.6071/M3M092

Abstract

Agricultural and forestry biomass can be converted to biochar through pyrolysis gasification, making it a significant carbon source for soil. Applying biochar to soil is a carbon-negative process that helps combat climate change, sustain soil biodiversity, and regulate water cycling. However, quantifying soil carbon content conventionally is time-consuming, labor-intensive, imprecise, and expensive, making it difficult to accurately measure in-field soil carbon's effect on storage water and nutrients. To address this challenge, for the first time, we report on extensive lab tests demonstrating non-intrusive methods for sensing soil carbon and related smart biochar applications, such as differentiating between biochar types from various biomass feedstock species, monitoring soil moisture, and biochar water retention capacity using portable microwave and millimeter wave sensors and machine learning. The datasets provide details on the microwave and millimeter wave reflectance signals. We validated our quantification method using supervised machine learning algorithms by collecting real soil mixed with known biochar contents in the field.

README: Microwave and Millimeter wave signals reflectance of Soil Carbon Content

Reference Information

Authors: Di An, YangQuan Chen

Data Created: 04/10/2023

Data Modified: 05/08/2023

Suggested Citations: Dataset: An, Di; Chen, YangQuan (2023), Microwave and millimeter wave signals reflectance of soil carbon content, Dryad, Dataset, https://doi.org/10.6071/M3M092

Corresponding publication: An, D., & Chen, Y. (2023). Non-intrusive soil carbon content quantification methods using machine learning algorithms: A comparison of microwave and millimeter wave radar sensors. Journal of Automation and Intelligence, 2(3), 152-166.

Data and File Overview

There are a total of 10 datasets: 5 for microwave and 5 for millimeterwave, respectively. The name meaning as follows:
"amount_biochar_microwave", "final_soilContent_low_millimeterwave" = Different Soil Carbon Amount Recognition
"type_biochar_microwave", "final_biochar_low_millimeterwave" = Different Types of Soil Carbon Classification
"soil_moisture_microwave", "final_moisture_low_millimeterwave" = Soil Moisture Study and Comparison
"Mos_C_microwave", "final_moistureCC_low_millimeterwave" = Soil Moisture Influenced by Carbon Content Added
"multiple_containers_microwave", "millimeterwave_mul_containers" = Sensing Robustness for Soil Carbon Content Percentage

Setup

  • Unpacking instructions: n/a
  • Relationships between files/folders: they are in different experiments sections
  • Recommended software/tools: > Matlab 2020

File/Folder Details-Microwave

Details for: amount_biochar_microwave.mat

  • Description: This is for classifying different amount of biochar mixed with soil using microwave sensor.
  • Format(s): .mat
  • Size(s): 6,149 KB
  • Dimensions:
    • Data: 50 rows x 8192 cols
    • label: 50 rows x 1 cols
  • Variables:
    • amount_biochar:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • d_2: raw data
    • y: Total 5 classes, each label has 10 sampling, therefore totally it has 50 labels.

Details for: type_biochar_microwave.mat

  • Description: This is for classifying different types of biochar using microwave sensor.
  • Format(s): .mat
  • Size(s): 6,140 KB
  • Dimensions:
    • Data: 50 rows x 8192 cols
    • label: 50 rows x 1 cols
  • Variables:
    • type_biochar:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • d_2: raw data
    • y: Total 5 classes, each label has 10 sampling, therefore totally it has 50 labels.

Details for: Mos_C_microwave.mat

  • Description: This is for classifying different soil moisture affected by carbon content added
  • Format(s): .mat
  • Size(s): 6,150 KB
  • Dimensions:
    • Data: 50 rows x 8192 cols
    • label: 50 rows x 1 cols
  • Variables:
    • mos_carbon:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • d_2: raw data
    • y: Total 5 classes, each label has 10 sampling, therefore totally it has 50 labels.

Details for: multiple_contains_microwave.mat

  • Description: This is for investigating robustness sensing for soil carbon content percentage
  • Format(s): .mat
  • Size(s): 61,322 KB
  • Dimensions:
    • Data: 500 rows x 8192 cols
    • label: 500 rows x 1 cols
  • Variables:
    • mulitle_container:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • d_2: raw data
    • y: Total 5 classes, each label has 10 sampling, therefore totally it has 50 labels.

Details for: soil_moisture_microwave.mat

  • Description: This is for classifying different soil moisture levels using microwave sensor.
  • Format(s): .mat
  • Size(s): 6,148 KB
  • Dimensions:
    • Data: 50 rows x 8192 cols
    • label: 50 rows x 1 cols
  • Variables:
    • moisture_soil:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • d_2: raw data
    • y: Total 5 classes, each label has 10 sampling, therefore totally it has 50 labels.

File/Folder Details-Millimeter wave

Details for: final_soilContent_low_millimeterwave.mat

  • Description: This is for classifying different amount of biochar mixed with soil using millimeter wave sensor.
  • Format(s): .mat
  • Size(s): 2,134 KB
  • Dimensions:
    • Data: 140 rows x 1024 cols
    • label: 140 rows x 1 cols
  • Variables:
    • soilCarbon_low:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • new_soilContent_low: raw data
    • y: Total 7 classes, each label has 20 sampling, therefore totally it has 140 labels.

Details for: final_biochar_low_millimeterwave.mat

  • Description: This is for classifying different types of biochar using millimeter wave sensor.
  • Format(s): .mat
  • Size(s): 1,857 KB
  • Dimensions:
    • Data: 120 rows x 1024 cols
    • label: 120 rows x 1 cols
  • Variables:
    • newBiochar:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • new_biochar_low: raw data
    • y: Total 6 classes, each label has 20 sampling, therefore totally it has 120 labels.

Details for: final_moistureCC_low_millimeterwave.mat

  • Description: This is for classifying different soil moisture affected by carbon content added using millimeter wave sensor.
  • Format(s): .mat
  • Size(s): 2,475 KB
  • Dimensions:
    • Data: 160 rows x 1024 cols
    • label: 160 rows x 1 cols
  • Variables:
    • new_Mosi:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • new_moistureC_low: raw data
    • y: Total 8 classes, each label has 20 sampling, therefore totally it has 160 labels.

Details for: mm_multi_containers.mat

  • Description: This is for investigating robustness sensing for soil carbon content percentage using millimeter wave sensor.
  • Format(s): .mat
  • Size(s): 3,279 KB
  • Dimensions:
    • Data: 500 rows x 300 cols
    • label: 500 rows x 1 cols
  • Variables:
    • mm_mult_container:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • mul_contain_mm: raw data in double types, can be used directly.
    • multi_containers_mm: raw data in complex number, easy for doing PDP plots.
    • y: Total 5 classes, each label has 100 sampling, therefore totally it has 500 labels.

Details for: final_moisture_low_millimeterwave.mat

  • Description: This is for classifying different soil moisture levels using microwave sensor.
  • Format(s): .mat
  • Size(s): 2,166 KB
  • Dimensions:
    • Data: 140 rows x 8192 cols
    • label: 140 rows x 1 cols
  • Variables:
    • new_Mosi:
      • type: table
      • description: combined sampling data with label together. Can be used in Classifier App directly.
    • new_moisture_low: raw data
    • y: Total 7 classes, each label has 20 sampling, therefore totally it has 140 labels.

Methods

The microwave sensor operated at a sampling frequency of 10.24 GHz, and each sample was measured 10 times to mitigate random error. Similarly, the millimeter-wave sensor, functioning within the 62 GHz to 69 GHz frequency range, replicated the microwave sensor's sampling instances to diminish error probabilities. We established 150 sampling points within the specified frequency band, setting the resolution bandwidth (RBW) at 100 kHz to balance sweep time efficiency and noise reduction. Additionally, samples were consistently positioned less than 2 centimeters beneath the direct millimeter-wave radar sensor for measurement consistency.

Soil samples were obtained from almond orchards and carefully filtered to remove large particles like stones. The biochar used was derived from a variety of sources, including 100% orchard prunings, walnut shells, pure wood, almond shells, and pinewood, designated as samples 1 through 5, respectively. In subsequent experiments, the soil was evenly spread to a weight of 50 grams, and the biochar was allocated at 2 grams, constituting 20% of the soil's weight. The materials were thoroughly combined prior to testing. For the millimeter-wave experimental setup, we prepared new soil and biochar mixtures with variable ratios ranging from 0% to 100% biochar content.

Additionally, the microwave data encapsulated a total of 8192 attributes per signal, whereas the millimeter-wave data comprised 1024 attributes.

Usage notes

Please use Matlab or Numpy to load mat files. 

Funding

California Strategic Growth Council, Award: CCR20014

United States Department of Commerce

National Science Foundation, Award: 1856112, Chemical, Bioengineering, Environmental and Transport Systems