Can canopy height of mixed pastures in integrated crop-livestock systems be estimated using planetscope imagery?
ABSTRACT. Canopy height (CH) is one of the key parameters used to evaluate forage biomass production and support grazing management decisions in intensively managed fields. In this study, we demonstrate the potential of using textural information derived from PlanetScope (PS) imagery to estimate CH of intensively managed mixed pastures in an Integrated Crop-Livestock Systems (ICLS) in the western region of São Paulo State, Brazil. PS images and field data of CH were acquired during the forage growing season of 2019 (from May to November) to calibrate and validate the CH prediction models using the Random Forest (RF) regression algorithm. We used as predictor variables eight second-order texture measures derived from the green, red, near-infrared spectral bands of PS images using the grey level co-occurrence matrix (GLCM) statistical texture approach. Pasture CH varied from 0.12 to 1.20 m with a coefficient of variation equal to 63.34%. Our best RF model was able to predict the spatiotemporal changes in pasture CH with high accuracy (R2 = 0.88) even with the high variability of the pasture CH through the forage growing season, mainly due to forage composition (different proportions of millet and ruzi grass) and grazing activities.
Main Authors: | , , , , , , , , |
---|---|
Other Authors: | |
Format: | Anais e Proceedings de eventos biblioteca |
Language: | Ingles English |
Published: |
2021-11-11
|
Subjects: | Sistemas integrados, Nanossatélites, Medidas de textura, Integração lavoura pecuária, Canopy height, Integrated systems, Nano-satellites, Texture measures, Integrated crop-livestock systems, Pastagem, |
Online Access: | http://www.alice.cnptia.embrapa.br/alice/handle/doc/1136079 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | ABSTRACT. Canopy height (CH) is one of the key parameters used to evaluate forage biomass production and support grazing management decisions in intensively managed fields. In this study, we demonstrate the potential of using textural information derived from PlanetScope (PS) imagery to estimate CH of intensively managed mixed pastures in an Integrated Crop-Livestock Systems (ICLS) in the western region of São Paulo State, Brazil. PS images and field data of CH were acquired during the forage growing season of 2019 (from May to November) to calibrate and validate the CH prediction models using the Random Forest (RF) regression algorithm. We used as predictor variables eight second-order texture measures derived from the green, red, near-infrared spectral bands of PS images using the grey level co-occurrence matrix (GLCM) statistical texture approach. Pasture CH varied from 0.12 to 1.20 m with a coefficient of variation equal to 63.34%. Our best RF model was able to predict the spatiotemporal changes in pasture CH with high accuracy (R2 = 0.88) even with the high variability of the pasture CH through the forage growing season, mainly due to forage composition (different proportions of millet and ruzi grass) and grazing activities. |
---|