JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

Predicting Soil Organic Matter, Available Nitrogen, Available Phosphorus and Available Potassium in a Black Soil Using a Nearby Hyperspectral Sensor System

  • 0Heilongjiang Academy of Black Soil Conservation and Utilization, Heilongjiang Academy of Agricultural Sciences, Harbin 150086, China.
Sensors (basel, Switzerland) +

|

May 11, 2024

|

May 11, 2024

View abstract on PubMed

Summary

This summary is machine-generated.

Hyperspectral imaging can assess soil properties like organic matter and nutrients. Soil moisture impacts accuracy, with dry conditions generally yielding better results for most properties, aiding field-based soil analysis.

Area Of Science

  • Agricultural Science
  • Soil Science
  • Remote Sensing

Background

  • Black soils are crucial for agriculture and food security due to high soil organic matter (SOM).
  • SOM significantly impacts farmland sustainability and plant nutrition.
  • Hyperspectral imaging (HSI) shows promise for laboratory soil nutrient detection, but field application is hindered by soil moisture (SM) variability.

Purpose Of The Study

  • To evaluate the feasibility of using handheld spectrometers for predicting soil organic matter (SOM), available nitrogen (AN), available phosphorus (AP), and available potassium (AK) under varying soil moisture conditions.
  • To compare the predictive performance of models using air-dried soil samples versus samples with 20%, 30%, and 40% soil moisture.
  • To identify optimal conditions and feature wavelengths for accurate, on-site soil property assessment.

Main Methods

  • Collected spectral data outdoors using a handheld spectrometer.
  • Developed Partial Least Squares Regression (PLSR) models to predict SOM, AN, AP, and AK.
  • Compared model performance across different soil moisture levels (air-dried, ~20%, ~30%, ~40% SM).
  • Utilized Variable Importance in Projection (VIP) to identify key wavelengths.

Main Results

  • The best predictive model for SOM, AN, and AK was achieved with air-dried soil samples.
  • Available phosphorus (AP) was best predicted using soil with approximately 30% moisture.
  • Model performance generally decreased as soil moisture increased, with AP being an exception.
  • Recommended feature wavelengths for each soil property were identified.

Conclusions

  • Field-based hyperspectral analysis is feasible for predicting key soil properties, but soil moisture management is critical.
  • Dry soil conditions are preferable for predicting SOM, AN, and AK using HSI.
  • Specific moisture levels may optimize prediction for certain nutrients like AP.
  • Identifying key wavelengths can guide the development of cost-effective, portable hyperspectral sensors for rapid, on-site soil assessment.

Keywords: