Above-Ground Biomass Prediction for Croplands at a Sub-Meter Resolution Using UAV–LiDAR and Machine Learning Methods

Jaime C. Revenga; Katerina Trepekli; Stefan Oehmcke; Rasmus Jensen; Lei Li; Christian Igel; Fabian Cristian Gieseke; Thomas Friborg

doi:10.3390/rs14163912

Above-Ground Biomass Prediction for Croplands at a Sub-Meter Resolution Using UAV–LiDAR and Machine Learning Methods

Jaime C. Revenga^*, Katerina Trepekli, Stefan Oehmcke, Rasmus Jensen, Lei Li, Christian Igel, Fabian Cristian Gieseke, Thomas Friborg

^*Corresponding author af dette arbejde

Publikation: Bidrag til tidsskrift/Konferencebidrag i tidsskrift /Bidrag til avis › Tidsskriftartikel › Forskning › peer review

19 Citationer (Scopus)

Abstract

Current endeavors to enhance the accuracy of in situ above-ground biomass (AGB) prediction for croplands rely on close-range monitoring surveys that use unstaffed aerial vehicles (UAVs) and mounted sensors. In precision agriculture, light detection and ranging (LiDAR) technologies are currently used to monitor crop growth, plant phenotyping, and biomass dynamics at the ecosystem scale. In this study, we utilized a UAV–LiDAR sensor to monitor two crop fields and a set of machine learning (ML) methods to predict real-time AGB over two consecutive years in the region of Mid-Jutland, Denmark. During each crop growing period, UAV surveys were conducted in parallel with AGB destructive sampling every 7–15 days, the AGB samples from which were used as the ground truth data. We evaluated the ability of the ML models to estimate the real-time values of AGB at a sub-meter resolution (0.17–0.52 m²). An extremely randomized trees (ERT) regressor was selected for the regression analysis, based on its predictive performance for the first year’s growing season. The model was retrained using previously identified hyperparameters to predict the AGB of the crops in the second year. The ERT performed AGB estimation using height and reflectance metrics from LiDAR-derived point cloud data and achieved a prediction performance of (Formula presented.) = 0.48 at a spatial resolution of 0.35 m². The prediction performance could be improved significantly by aggregating adjacent predictions ((Formula presented.) = 0.71 and (Formula presented.) = 0.93 at spatial resolutions of 1 m² and 2 m², respectively) as they ultimately converged to the reference biomass values because any individual errors averaged out. The AGB prediction results were examined as function of predictor type, training set size, sampling resolution, phenology, and canopy density. The results demonstrated that when combined with ML regression methods, the UAV–LiDAR method could be used to provide accurate real-time AGB prediction for crop fields at a high resolution, thereby providing a way to map their biochemical constituents.

Originalsprog	Engelsk
Artikelnummer	3912
Tidsskrift	Remote Sensing
Vol/bind	14
Nummer	16
ISSN	2072-4292
DOI	https://doi.org/10.3390/rs14163912
Status	Udgivet - aug. 2022
Udgivet eksternt	Ja

Adgang til dokumentet

10.3390/rs14163912

Andre filer og links

Link to publication in Scopus

Citationsformater

@article{7f08f864d65644b49362682259566d36,

title = "Above-Ground Biomass Prediction for Croplands at a Sub-Meter Resolution Using UAV–LiDAR and Machine Learning Methods",

abstract = "Current endeavors to enhance the accuracy of in situ above-ground biomass (AGB) prediction for croplands rely on close-range monitoring surveys that use unstaffed aerial vehicles (UAVs) and mounted sensors. In precision agriculture, light detection and ranging (LiDAR) technologies are currently used to monitor crop growth, plant phenotyping, and biomass dynamics at the ecosystem scale. In this study, we utilized a UAV–LiDAR sensor to monitor two crop fields and a set of machine learning (ML) methods to predict real-time AGB over two consecutive years in the region of Mid-Jutland, Denmark. During each crop growing period, UAV surveys were conducted in parallel with AGB destructive sampling every 7–15 days, the AGB samples from which were used as the ground truth data. We evaluated the ability of the ML models to estimate the real-time values of AGB at a sub-meter resolution (0.17–0.52 m2). An extremely randomized trees (ERT) regressor was selected for the regression analysis, based on its predictive performance for the first year{\textquoteright}s growing season. The model was retrained using previously identified hyperparameters to predict the AGB of the crops in the second year. The ERT performed AGB estimation using height and reflectance metrics from LiDAR-derived point cloud data and achieved a prediction performance of (Formula presented.) = 0.48 at a spatial resolution of 0.35 m2. The prediction performance could be improved significantly by aggregating adjacent predictions ((Formula presented.) = 0.71 and (Formula presented.) = 0.93 at spatial resolutions of 1 m2 and 2 m2, respectively) as they ultimately converged to the reference biomass values because any individual errors averaged out. The AGB prediction results were examined as function of predictor type, training set size, sampling resolution, phenology, and canopy density. The results demonstrated that when combined with ML regression methods, the UAV–LiDAR method could be used to provide accurate real-time AGB prediction for crop fields at a high resolution, thereby providing a way to map their biochemical constituents.",

keywords = "above-ground biomass, machine learning, precision agriculture, UAV–LiDAR",

author = "Revenga, {Jaime C.} and Katerina Trepekli and Stefan Oehmcke and Rasmus Jensen and Lei Li and Christian Igel and Gieseke, {Fabian Cristian} and Thomas Friborg",

note = "Publisher Copyright: {\textcopyright} 2022 by the authors.",

year = "2022",

month = aug,

doi = "10.3390/rs14163912",

language = "English",

volume = "14",

journal = "Remote Sensing",

issn = "2072-4292",

publisher = "Multidisciplinary Digital Publishing Institute (MDPI)",

number = "16",

}

TY - JOUR

T1 - Above-Ground Biomass Prediction for Croplands at a Sub-Meter Resolution Using UAV–LiDAR and Machine Learning Methods

AU - Revenga, Jaime C.

AU - Trepekli, Katerina

AU - Oehmcke, Stefan

AU - Jensen, Rasmus

AU - Li, Lei

AU - Igel, Christian

AU - Gieseke, Fabian Cristian

AU - Friborg, Thomas

PY - 2022/8

Y1 - 2022/8

N2 - Current endeavors to enhance the accuracy of in situ above-ground biomass (AGB) prediction for croplands rely on close-range monitoring surveys that use unstaffed aerial vehicles (UAVs) and mounted sensors. In precision agriculture, light detection and ranging (LiDAR) technologies are currently used to monitor crop growth, plant phenotyping, and biomass dynamics at the ecosystem scale. In this study, we utilized a UAV–LiDAR sensor to monitor two crop fields and a set of machine learning (ML) methods to predict real-time AGB over two consecutive years in the region of Mid-Jutland, Denmark. During each crop growing period, UAV surveys were conducted in parallel with AGB destructive sampling every 7–15 days, the AGB samples from which were used as the ground truth data. We evaluated the ability of the ML models to estimate the real-time values of AGB at a sub-meter resolution (0.17–0.52 m2). An extremely randomized trees (ERT) regressor was selected for the regression analysis, based on its predictive performance for the first year’s growing season. The model was retrained using previously identified hyperparameters to predict the AGB of the crops in the second year. The ERT performed AGB estimation using height and reflectance metrics from LiDAR-derived point cloud data and achieved a prediction performance of (Formula presented.) = 0.48 at a spatial resolution of 0.35 m2. The prediction performance could be improved significantly by aggregating adjacent predictions ((Formula presented.) = 0.71 and (Formula presented.) = 0.93 at spatial resolutions of 1 m2 and 2 m2, respectively) as they ultimately converged to the reference biomass values because any individual errors averaged out. The AGB prediction results were examined as function of predictor type, training set size, sampling resolution, phenology, and canopy density. The results demonstrated that when combined with ML regression methods, the UAV–LiDAR method could be used to provide accurate real-time AGB prediction for crop fields at a high resolution, thereby providing a way to map their biochemical constituents.

AB - Current endeavors to enhance the accuracy of in situ above-ground biomass (AGB) prediction for croplands rely on close-range monitoring surveys that use unstaffed aerial vehicles (UAVs) and mounted sensors. In precision agriculture, light detection and ranging (LiDAR) technologies are currently used to monitor crop growth, plant phenotyping, and biomass dynamics at the ecosystem scale. In this study, we utilized a UAV–LiDAR sensor to monitor two crop fields and a set of machine learning (ML) methods to predict real-time AGB over two consecutive years in the region of Mid-Jutland, Denmark. During each crop growing period, UAV surveys were conducted in parallel with AGB destructive sampling every 7–15 days, the AGB samples from which were used as the ground truth data. We evaluated the ability of the ML models to estimate the real-time values of AGB at a sub-meter resolution (0.17–0.52 m2). An extremely randomized trees (ERT) regressor was selected for the regression analysis, based on its predictive performance for the first year’s growing season. The model was retrained using previously identified hyperparameters to predict the AGB of the crops in the second year. The ERT performed AGB estimation using height and reflectance metrics from LiDAR-derived point cloud data and achieved a prediction performance of (Formula presented.) = 0.48 at a spatial resolution of 0.35 m2. The prediction performance could be improved significantly by aggregating adjacent predictions ((Formula presented.) = 0.71 and (Formula presented.) = 0.93 at spatial resolutions of 1 m2 and 2 m2, respectively) as they ultimately converged to the reference biomass values because any individual errors averaged out. The AGB prediction results were examined as function of predictor type, training set size, sampling resolution, phenology, and canopy density. The results demonstrated that when combined with ML regression methods, the UAV–LiDAR method could be used to provide accurate real-time AGB prediction for crop fields at a high resolution, thereby providing a way to map their biochemical constituents.

KW - above-ground biomass

KW - machine learning

KW - precision agriculture

KW - UAV–LiDAR

UR - http://www.scopus.com/inward/record.url?scp=85137837857&partnerID=8YFLogxK

U2 - 10.3390/rs14163912

DO - 10.3390/rs14163912

M3 - Journal article

AN - SCOPUS:85137837857

SN - 2072-4292

VL - 14

JO - Remote Sensing

JF - Remote Sensing

IS - 16

M1 - 3912

ER -

Above-Ground Biomass Prediction for Croplands at a Sub-Meter Resolution Using UAV–LiDAR and Machine Learning Methods

Abstract

Adgang til dokumentet

Andre filer og links

Fingeraftryk

Citationsformater