Purushothaman, N K and Premsagar, A and Raj, M and Majeed, Israr and Reddy, N N and Sinha, L K and Das, B S (2025) Influence of Bidirectional Reflectance Distribution Function in Estimating Basic Soil Properties Using Airborne Hyperspectral Data. IEEE Transactions on Geoscience and Remote Sensing, 63. pp. 1-11. ISSN 0196-2892
![[img]](http://oar.icrisat.org/style/images/fileicons/application_pdf.png) |
PDF
- Published Version
Available under License ["licenses_description_cc_attribution" not defined]. Download (1MB) |
Abstract
Recent studies on hyperspectral remote sensing (HSR) have shown that the estimation accuracy of different vegetation characteristics improves when the HSR data are corrected for the bidirectional reflectance distribution function (BRDF) effects. Similar studies involving soil parameters are limited. Here, we used the BRDF-corrected HSR data collected using the airborne visible-infrared imaging spectrometer-next generation (AVIRIS-NG) sensor to estimate soil parameters over a 138-km2 agricultural catchment. Surface soil samples were collected from 173 ground reference locations (GRLs) from this catchment to measure clay and sand contents, pH, electrical conductivity (EC), and soil organic carbon (SOC) contents. The BRDF correction was applied using the flexible BRDF (FlexBRDF) algorithm, and a polynomial unmixing approach was used to extract soil spectra from the corrected image. The BRDF correction successfully removed the shading effects and produced smooth transitions along the overlapping regions when multiple AVIRIS-NG images were mosaicked. Upon unmixing, soil spectra could be extracted at 140 GRLs when BRDF-corrected spectra were used, while uncorrected spectra produced soil spectra only for 114 GRLs. Chemometric models were validated using 109 common GRLs to compare estimation accuracy across laboratory-measured soil spectra ( SSLab ) and those obtained from unmixing of BRDF-corrected and uncorrected spectra. The coefficient of determination ( R2 ) values in the validation datasets ranged from 0.40 to 0.83 for both the BRDF-corrected and SSLab data, while the uncorrected spectra showed poor estimation accuracy (R2: 0.25–0.56). The resulting root-mean-squared error (RMSE) was reduced by 10% and 47% for the BRDF-corrected soil spectra compared to their uncorrected data. The BRDF-corrected and unmixed soil spectra were used to map soil properties at ~5-m spatial resolution for the entire catchment. Low SOC contents in the resulting maps adjoining the Ganges river flowing through our study site captured the topsoil loss typically observed from river banks. Thus, the BRDF-corrected HSR data not only improved the accuracy of soil estimates but also showed potential to identify vulnerable areas needing precision management measures with high spatial resolution.
| Item Type: | Article |
|---|---|
| Divisions: | Global Research Program - Resilient Farm and Food Systems |
| CRP: | UNSPECIFIED |
| Uncontrolled Keywords: | Soil, Soil properties, Accuracy, Reflectivity, Estimation, Soil measurements, Rivers, Hyperspectral imaging, Vegetation mapping, Land surface, Memory-based learning (MBL), nonlinear polynomial unmixing, precision agriculture, soil erosion |
| Subjects: | Others > Remote Sensing Others > Soil |
| Depositing User: | Mr Nagaraju T |
| Date Deposited: | 14 Oct 2025 04:26 |
| Last Modified: | 14 Oct 2025 04:26 |
| URI: | http://oar.icrisat.org/id/eprint/13360 |
| Official URL: | https://ieeexplore.ieee.org/abstract/document/1100... |
| Projects: | UNSPECIFIED |
| Funders: | UNSPECIFIED |
| Acknowledgement: | Bhabani Sankar Das thankfully acknowledges the erstwhile Defence Terrain Research Laboratory, Defence Research and Development Organization (DRDO), New Delhi, for facilitating soil collection at their study site. Naveen Kumar Purushothaman acknowledges the Prime Minister’s Research Fellowship for his Ph.D. research. |
| Links: |
Actions (login required)
 |
View Item |
Altmetric
Altmetric