Assessing the impact of spatial resolution of UAS-based remote sensing and spectral resolution of proximal sensing on crop nitrogen retrieval accuracy

Authors

• Kianoosh Hassani, Oklahoma State University
• Hamed Gholizadeh, Oklahoma State University
• Saleh Taghvaeian, University of Nebraska-LincolnFollow
• Victoria Natalie, Oklahoma State University
• Jonathan Carpenter, Oklahoma State University
• Jamey Jacob, Oklahoma State University

Document Type

Article

Date of this Version

7-6-2023

Citation

INTERNATIONAL JOURNAL OF REMOTE SENSING 2023, VOL. 44, NO. 14, 4441–4464 https://doi.org/10.1080/01431161.2023.2237162

Comments

Open access.

Abstract

Foliar nitrogen (N) plays a central role in photosynthetic machinery of plants, regulating their growth rates. However, field-based methods for monitoring plant N concentration are costly and limited in their ability to cover large spatial extents. In this study, we had two objectives: (1) assess the capability of unoccupied aerial system (UAS) and non-imaging spectroscopic data in estimating sorghum and corn N concentration and (2) determine the impact of spatial and spectral resolution of reflectance data on estimating sorghum and corn N concentration. We used a UAS and an ASD spectroradiometer to collect canopy- and leaf-level spectral data from sorghum and corn at experimental plots located in Stillwater, Oklahoma, U.S. We also collected foliage samples in the field and measured foliar N concentration in the lab for model validation. To assess the impact of spectral scale on estimating N concentration, we resampled our leaf-level ASD data to generate datasets with coarser spectral resolutions. To determine the impact of spatial scale on estimating N concentration, we resampled our UAS data to simulate five datasets with varying spatial resolutions ranging from 5 cm to 1 m. Finally, we used a suite of vegetation indices (VIs) and machine learning algorithms (MLAs) to estimate N concentration. Results from leaf-level ASD spectral data showed that the resampled data matching the spectral resolution of our UAS-based data at five spectral bands ranging from 360 to 900 nm provided sufficient spectral information to estimate plot-level sorghum and corn N concentration. Regarding spatial resolution, canopy-level UAS data resampled at multiple pixel sizes, ranging from 1 cm to 1 m were consistently capable of estimating N concentration. Overall, our findings indicate the possibility of developing monitoring instruments with optimal spectral and spatial resolution for estimating N concentration in crops.