Satellite imaging spectroscopy for mapping the spatial distribution of crop residues

In a study published in European Journal of Remote SensingIn this study, the researchers proposed a two-step classification method for mapping non-photosynthetic plants or crop residues using spectroscopy and machine learning techniques from PRISMA Space Imaging Spectroscopy.

Stady: Mapping the spatial distribution of crop residues using PRISMA satellite imaging spectroscopy. Image Credit: Hryshchyshen Serhii /

Non-photosynthetic vegetation and crop residues in agriculture

Non-photosynthetic vegetation refers to the vegetation that cannot use photosynthesis for its growth and development. It consists mainly of plant droppings, dead or dying plants, and crop residues.

The remains of dying crops and plants are essential in carbon farming and sustainable agriculture. Its importance stems from its roles in the carbon, nitrogen and water cycles and soil conservation.

The presence of crop residue in the fields protects the soil from erosion, regulates temperatures and moisture levels, enhances soil organic carbon, reduces soil compaction caused by agricultural machinery, and improves soil structure.

The relative abundance of crop residues indicates agricultural management practices as they are related to crop rotation, tillage and harvesting techniques.

Ultra-spectrum remote sensing of non-photosynthetic crop or plant residues

Aerial and satellite remote sensing systems can identify and evaluate NPV faster, accurately, and objectively than traditional field operator-based reports. By determining the availability and quantity of NPV, remote sensing can help monitor and control the implementation and effectiveness of agricultural conservation policies.

Many remote sensing technologies for non-photosynthetic plants in agricultural environments rely on spectral differences between crop residues, bare soil, and green vegetation to perform class detection.

In particular, they use multispectral broadband data to estimate spectral indices for NPV mapping.

PRISMA mission

Previous research on non-photosynthetic vegetation using hyperspectral data relied on terrestrial and atmospheric data due to the limited availability of spaceborne sensors. Therefore, they were unable to differentiate NPV from different types of bare soil, especially in the visible and near-infrared spectral range.

However, the necessary continuous bands can be provided by imaging spectrometers to detect and distinguish NPV from bare soil.

The PRISMA satellite mission is one of the most recent Earth Observation Space Imaging (EO) spectroscopy missions, and its first application is in agriculture.

PRISMA is a precursor to the next generation of hyperspectral missions; It targets specific EO applications related to environmental conservation, monitoring, and sustainable development.

However, PRISMA’s 30 km coverage and long return interval may not be ideal for monitoring operations. Despite these limitations, acquisitions must be frequent enough to do in-season research.

Spectral and machine learning techniques for mapping crop conditions

this is study It aims to build a framework for classifying distinct field conditions using hyperspectral satellite data to generate a reliable quantitative NPV map, assess crop residues, and monitor field condition over time.

A two-step classification method has been developed for mapping crop conditions based on PRISMA data from Space Imaging Spectroscopy.

First, four diagnostic spectral periods were modeled using Gaussian exponential optimization so that the hyperspectral space is limited to the model parameters. Second, a pixel-level decision tree (DT) is trained to classify crop residues, bare soil, established dead plants, and emerging green plants.

The plot-level results are then aggregated to provide objective maps for monitoring and evaluating agricultural practices at the farm level.

Important results of the study

Crop residue mapping is accurate at pixel and parcel levels (overall accuracy > 90%; K > 0.9). The results indicated that PRISMA data are sufficient for plot-level mapping of crop conditions.

The accuracy of the predictions was assessed and analyzed based on the change in the trajectories of target conditions over time, which is also important for tracking soil conservation activities. This investigation of the tracks showed that the classification results are consistent with independent data and validated the effectiveness of the method in mapping field conditions.

It also shows that the use of machine learning classifiers in conjunction with spectroscopic methods constrained by known optical properties of the investigated surfaces results in a mapping methodology that is reliable, transferable and usable with time-series data.

Future limitations and recommendations

In this studyClassification supervision was performed once. The decision tree model was used over a time series that includes several seasons, different types of crops, and different environmental variables. The train sample was selected on the basis of the image.

Although useful in situations without an additional spectrum data set, this strategy has several drawbacks.

A reference spectral library that includes a wide range of surface spectra recorded under controlled settings should be available to widely adopt a decision tree model for various field situations and quantitative examination.

The machine learning model can be trained using these libraries to improve the stratification rules by taking into account different crops and agricultural conditions. The performance of several narrow-band remote sensors with various spectral designs can be tested using the same spectral libraries.

Multiple machine learning algorithms should be used to improve the classification of PRISMA images to improve the classification of NPV and crop residues in agricultural landscapes.


Monica Pepe, Loredana Pompilio, Luigi Rangetti, Francesco Notini and Mirco Boschetti (2022) Mapping the spatial distribution of crop residues using PRISMA satellite imaging spectroscopy. European Journal of Remote Sensing.

Disclaimer: The opinions expressed here are those of the author expressed in his personal capacity and do not necessarily represent the opinions of Limited T/A AZoNetwork is the owner and operator of this website. This disclaimer forms part of the terms and conditions of use of this website.