Introduction
⌅Rice (Oryza sativa L.) constitutes one of the pillars of food security in Cuba and much of the world. In the Cuban diet, this cereal is irreplaceable: it is estimated that each person consumes around 60 kg annually, one of the highest figures in the region (Monreal, 2025MONREAL, P.: Economista desmiente relato del consumo excesivo de arroz en Cuba, [en línea], Cibercuba, 2025, Disponible en: https://www.cibercuba.com. ). To meet national demand, the country requires about 600,000 tons per year, but in 2024 it barely managed to produce 80,000 tons, representing only 13% of the demand and forcing reliance on costly imports (Infobae, 2025INFOBAE: Cuba agudiza su crisis alimentaria: la producción de arroz cayó a mínimos históricos, [en línea], Infobae, 2025, Disponible en: https://www.infobae.com/america/america-latina/2025/04/02/cuba-agudiza-su-crisis-alimentaria-la-produccion-de-arroz-cayo-a-minimos-historicos/. ). This situation reflects the vulnerability of the Cuban agri-food system and the need to optimize production through advanced technologies.
The rice production crisis in Cuba has been highlighted by various specialists as the result of an inefficient agricultural model dependent on external inputs (Periódico Girón, 2025PERIÓDICO GIRÓN: ¿Qué perspectivas tiene la producción de arroz en Cuba?, [en línea], Girón, 2025, Disponible en: https://giron.cu/2025/03/05/produccion-de-arroz-en-cuba/. ). In this context, the search for technological alternatives to improve crop management becomes a strategic imperative.
Remote sensing has been consolidated as a key tool for modern agriculture. The use of satellites, drones, and remote sensors makes it possible to monitor the physiological status of crops, detect water stress, estimate biomass, and assess plant health quickly and non-destructively. In Cuba, the introduction of these technologies has been progressive, with projects promoted by universities and the Ministry of Science, Technology and Environment (Ruiz et al., 2024RUIZ, P.M.E.; GARCÍA, R.R.; MACHADO, G.N.: “Aplicación de drones en la agricultura internacional y cubana. Revisión”, Revista Ciencias Técnicas Agropecuarias, 33(1), 2024, ISSN: 2071-0054.).
Drones provide high spatial resolution images, useful for characterizing specific plots and detecting local variations in crops. Meanwhile, Sentinel-2 satellites provide multispectral data with broad coverage and adequate temporal frequency for regional monitoring. The combination of both sources of information opens the possibility of establishing multiscale models that integrate local detail and global vision (Ponvert y Lau, 2018PONVERT, D.D.R.; LAU, A.: “Uso de la teledetección en la gestión de desastres “naturales” en la agricultura. Apuntes y reflexiones.”, 2018.).
Spectral indices derived from satellite images, such as NDVI, GNDVI, and NDRE, have proven to be robust indicators of plant vigor and chlorophyll content. However, their spatial resolution (10-20 m) may limit precision in small plots. In contrast, indices calculated from drone RGB orthophotography, such as VARI, ExG, and CIVE, offer a higher level of detail, although they lack the spectral breadth of satellite sensors. The correlation between both sets of indices allows validation of the usefulness of RGB images as a substitute or complement to satellite data. Recent studies have shown that indices such as VARI present significant correlations with NDVI, confirming the possibility of using drones for agricultural monitoring under resource-limited conditions (Famartínez, 2023FAMARTÍNEZ, A.: Sentinel‑2 en R: cálculo de índices y correlación, [en línea], GitHub, 2023, Disponible en: https://github.com. ).This integration is especially relevant in Cuba, where access to high-resolution satellite images may be restricted, but the use of drones has expanded in academic and productive projects.
The objective of this research is to evaluate the correlation between spectral indices derived from Phantom 4 drone RGB orthophotography and Sentinel-2 L2A in rice (Oryza sativa L.) at the seedling stage.
Materials and Methods
⌅Study Area
⌅The study was conducted in plot 6 of the Basic Cooperative Production Unit (UBPC) Las Nuevas, belonging to the Sur del Jíbaro Rice Company, located in the municipality of La Sierpe, Sancti Spíritus Province. This plot is part of one of the main rice-growing areas of the country, characterized by fertile soils, flat topography, and the availability of irrigation infrastructure. The study period corresponds to the dry season of the year 2020 (Figure 1).
Methodology
⌅For the development of the study, the following materials and tools were used:
-
QGIS v3.40 Bratislava software (OSGeo, 2024OSGEO: QGIS v3.40 Bratislava: User Guide, 2024.).
-
Sentinel-2 L2A satellite image (resolution of 10 to 20 meters, acquired on October 4, 2020) (Copernicus, 2024COPERNICUS: Sentinel‑2 mission overview, [en línea], Copernicus Programme, 2024, Disponible en: https://www.copernicus.eu/en/copernicus-services/sentinel-2. ).
-
RGB orthophotography captured with Phantom 4 drone (resolution of 5 cm/pixel, acquired on October 5, 2020), Figure 2.
-
QGIS plugins: Semi-Automatic Classification Plugin (SCP), Raster Calculator, Field Calculator.
-
Hardware: workstation with processing capacity for multispectral images.
Image Preprocessing
⌅Sentinel-2 images were subjected to atmospheric correction (level L2A), clipping of the study area, and reprojection to the UTM coordinate system. The RGB orthophoto was orthorectified and mosaicked to generate a single raster in QGIS.
Spectral Indices Calculation
⌅The following spectral indices were calculated using the QGIS Raster Calculator:
Integration and Analysis in QGIS
⌅Raster layers were integrated into the QGIS environment through spatial overlay. A systematic sampling grid was generated according to the spatial resolution of the data sources. For Sentinel‑2 imagery (10-20 m/pixel), a spacing of 20 meters was applied, ensuring that each point represented an independent pixel and avoiding spatial redundancy. For RGB drone orthophotographs (0.05-0.20 m/pixel), a spacing of 5-10 meters was used, allowing the internal variability of the crop to be captured. In this case, 415 points were sampled, from which spectral index values were extracted for correlation analysis. This definition followed FAO (2023)FAO: “Digital agriculture transformation and remote sensing applications”, En: Organización de las Naciones Unidas para la Alimentación y la Agricultura, Roma, Italia, 2023, DOI: . https://doi.org/10.4060/cc5170en. methodological recommendations for agricultural remote sensing and recent precision agriculture studies, ensuring reproducibility of the analysis.
Field Validation
⌅The results were contrasted with direct field observations of the physiological status of rice at the seedling stage.
Statistical Analysis
⌅Statistical analysis was performed using Statgraphics Centurion v15.0. Pearson’s correlation coefficient (r) was applied to evaluate the linear relationship between RGB spectral indices and those derived from Sentinel-2. The r values were interpreted according to Cohen (2013)COHEN, J.: Statistical power analysis for the behavioral sciences, Ed. routledge, DOI: https://doi.org/10.4324/9780203771587 p., 2013, ISBN: 0-203-77158-3.. The results were validated with field observations of the physiological status of the rice crop.
Cohen (2013)COHEN, J.: Statistical power analysis for the behavioral sciences, Ed. routledge, DOI: https://doi.org/10.4324/9780203771587 p., 2013, ISBN: 0-203-77158-3. Scale:
Methodological Flow Diagram
⌅The methodological flow followed in the study is summarized in the following steps:
-
Phantom 4 drone capture (RGB): A flight was carried out with the Phantom 4 drone on October 5, 2020 to obtain a high-resolution RGB orthophoto (5 cm/pixel) of plot 6.
-
Sentinel-2 L2A image: A Sentinel-2 L2A image corresponding to October 4, 2020 was downloaded, with spectral bands at 10 m and 20 m resolution.
-
Spectral indices calculation: RGB indices (VARI, ExG, CIVE, GLI) and Sentinel-2 indices (NDVI, GNDVI, NDRE) were calculated to characterize the crop status (Figure 3).
Technical Considerations
⌅QGIS v3.40 Bratislava offers robust tools for raster image processing, spectral index calculation, and spatial analysis, facilitating reproducibility of the workflow (OSGeo, 2024OSGEO: QGIS v3.40 Bratislava: User Guide, 2024.). Meanwhile, Statgraphics provides a user-friendly interface for statistical correlation analysis, allowing precise interpretation of relationships between spectral variables. The combination of both platforms enables efficient and accessible multiscale analysis for agricultural monitoring.
Results and Discussion
⌅Field Photographs
⌅This section presents visual evidence of the physiological status of rice crops at the seedling stage (Figure 4), as well as the conditions of plots flooded by gravity irrigation in plot 6 of UBPC Las Nuevas (Figure 5). The images allow visual validation of crop uniformity, seedling density, and absence of water stress during the dry season of 2020 (Figure 6).
This index presents thematic maps generated from spectral data obtained in plot 6 of UBPC Las Nuevas, Sur del Jíbaro Rice Company, during the dry season of 2020. The maps allow spatial visualization of the distribution of different spectral indices that characterize the status of the rice crop at the seedling stage (Figure 7).
CIVE is an empirical index that combines RGB bands with specific coefficients: CIVE=0.441R−0.811G+0.385B+18.78745
⌅
-
Due to this formula, CIVE values depend directly on the scale of RGB values (for example, 0-255 if the image is in 8-bit format).
-
Therefore, CIVE values may vary widely, from negative (when green predominates) to positive (when red or blue predominates), and are not constrained between −1 and 1.
RGB vs Sentinel-2 Correlation Report
⌅This report presents the Pearson correlation analysis between spectral indices derived from RGB drone images (VARI, ExG, CIVE) and multispectral indices obtained from Sentinel-2 imagery (NDVI, GNDVI, NDRE). The objective is to evaluate the linear relationship between these pairs of indices to validate the use of RGB indices as reliable proxies in multiscale agricultural monitoring of rice crops.
Correlation Analysis Results
⌅The Pearson correlation coefficients obtained indicate a significant correlation between RGB indices and satellite indices (Table 1). In particular, the correlation between VARI and NDVI is high, suggesting that the VARI index may serve as a strong proxy for vegetation vigor. The correlations between ExG-GNDVI and CIVE-NDRE are moderate, further supporting their usefulness in agricultural monitoring (Figures 9, 10, 11). Sampled values for each index in the pixel grid were extracted using the QGIS Simple Raster Values tool.
| Pair of Indices | Pearson Coefficient (r) | Interpretation |
|---|---|---|
| VARI vs NDVI | 0.741 | Very strong correlation |
| ExG vs GNDVI | 0.672 | Moderate correlation |
| CIVE vs NDRE | −0.635 | Moderate correlation |
Scatter Plots
⌅
Pattern description: The scatter plot shows a positive and consistent relationship between VARI (RGB) and NDVI (Sentinel-2), with points concentrated along an upward trend. The Pearson coefficient obtained (r ≈ 0.74) indicates a very strong correlation, consistent with the sensitivity of both indices to vegetation vigor and canopy greenness. This behavior supports the use of RGB orthophotos as a proxy for vigor when NIR data is not available.
Pattern description: The scatter plot of ExG (RGB) vs GNDVI (Sentinel-2) shows a positive relationship with moderate dispersion and a clear linear trend. The Pearson coefficient (r ≈ 0.67) indicates a moderate correlation, consistent with the sensitivity of GNDVI to chlorophyll content (green band) and the ability of ExG to emphasize the green component in RGB images.
Pattern description: The scatter plot of CIVE (RGB) vs NDRE (Sentinel-2) shows an inverse relationship with a downward trend; the Pearson coefficient (r ≈ −0.64) indicates a moderate negative correlation. This reflects that CIVE-by combining RGB channels to extract vegetation-does not directly capture the sensitivity of the Red Edge to chlorophyll and nitrogen, but is inversely related to physiological variations detected by NDRE.
Map Interpretation
⌅
-
The NDVI map showed predominantly high values in plot 6, indicating elevated and homogeneous vegetation vigor, characteristic of the seedling stage without water stress. These results are consistent with studies in Peru, where Sentinel-2 NDVI has been used to predict rice yield and monitor vigor across different phenological phases (Quille et al., 2024QUILLE, M.J.A.; RUIZ, L.A.; RAMOS-FERNÁNDEZ, L.: “Multiseasonal analysis of rice crop yield prediction with Sentinel-2 time series and UAV imagery in Lambayeque (Peru)”, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 48: 421-426, 2024b, ISSN: 1682-1750, DOI: https://doi.org/10.5194/isprs-archives-XLVIII-3-421-2024. ).
-
The VARI map, derived from RGB drone orthophotos, reflected similar patterns to NDVI, with areas of greater greenness in the center of the plot, suggesting a strong correlation between both indices. Previous research has demonstrated that VARI can be a reliable proxy for NDVI in rice and wheat crops, especially when using low-cost UAVs (García et al., 2018GARCÍA, C.D.A.; RAMÓN, V.J.A.; ALZATE, V.D.F.; PALACIOS, G.J.R.: “Dynamics of the indices NDVI and GNDVI in a rice growing in its reproduction phase from multi-spectral aerial images taken by drones”, En: International Conference of ICT for Adapting Agriculture to Climate Change, Ed. Springer, pp. 106-119, 2018, DOI: https://doi.org/10.1007/978-3-030-04447-3_7. ; MDPI., 2019MDPI.: “Assessing correlation of high‑resolution NDVI with fertilizer application and rice yield”, Remote Sensing, 11(2): 112, 2019, ISSN: https://www.mdpi.com/2072-4292/11/2/112. ).
-
The NDRE map revealed subtle variations in chlorophyll content and initial vigor, with higher values in the central and eastern areas of the plot, possibly associated with greater leaf density. These findings are consistent with studies in Iran and Indonesia, where Sentinel-2 NDRE has been correlated with chlorophyll and nitrogen content in rice, showing sensitivity in early growth (Wijayanto et al., 2024WIJAYANTO, Y.; SAFITRI, M.; PURNAMASARI, I.; BUDIMAN, S.A.; SAPUTRA, T.W.; REGAR, A.F.; RISTIYANA, S.: “Estimating the contents of Chlorophyll, Nitrogen, and Yields on Rice Through Sentinel-2 Vegetation Indices in Heterogeneous Land Management”, The Indonesian Journal of Geography, 56(3): 357-367, 2024, ISSN: 0024-9521, DOI: https://doi.org/10.22146/ijg.87159.; Ehsanpour et al., 2025EHSANPOUR, A.; GHANBARI, P.E.; AMINI, S.: “Monitoring chlorophyll in rice fields using Sentinel 2 and UAV images and applying machine learning”, Sepehr Journal, 2025, ISSN: 2588-3860, DOI: https://doi.org/10.22131/sepehr.2025.2045093.3102. ).
Index Interpretation
⌅
-
NDVI showed high and homogeneous values in plot 6, indicating vegetation vigor. This behavior has been widely documented in studies of rice and other crops, where NDVI is used as a robust indicator of biomass and vigor (Tucker, 1979TUCKER, C.J.: “Red and photographic infrared linear combinations for monitoring vegetation”, Remote sensing of Environment, 8(2): 127-150, 1979, ISSN: 0034-4257, DOI: https://doi.org/10.1016/0034-4257(79)90013-0. ; Quille et al., 2024aQUILLE, M.J.A.; RUIZ, L.A.; RAMOS, F.L.: “Multiseasonal analysis of rice crop yield prediction with Sentinel-2 time series and UAV imagery in Lambayeque (Peru)”, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 48: 421-426, 2024a, ISSN: 1682-1750, DOI: https://doi.org/10.5194/isprs-archives-XLVIII-3-421-2024. ).
-
NDRE revealed subtle variations in chlorophyll in central and eastern areas. Research in Iran and Indonesia has demonstrated that NDRE is sensitive to chlorophyll and nitrogen content in rice, especially in early stages (Wijayanto et al., 2024WIJAYANTO, Y.; SAFITRI, M.; PURNAMASARI, I.; BUDIMAN, S.A.; SAPUTRA, T.W.; REGAR, A.F.; RISTIYANA, S.: “Estimating the contents of Chlorophyll, Nitrogen, and Yields on Rice Through Sentinel-2 Vegetation Indices in Heterogeneous Land Management”, The Indonesian Journal of Geography, 56(3): 357-367, 2024, ISSN: 0024-9521, DOI: https://doi.org/10.22146/ijg.87159.; Ehsanpour et al., 2025EHSANPOUR, A.; GHANBARI, P.E.; AMINI, S.: “Monitoring chlorophyll in rice fields using Sentinel 2 and UAV images and applying machine learning”, Sepehr Journal, 2025, ISSN: 2588-3860, DOI: https://doi.org/10.22131/sepehr.2025.2045093.3102. ).
-
The strongest correlation is between VARI and NDVI, validating the use of RGB orthophotos for early monitoring. Studies in Colombia and Mexico have confirmed that VARI can be a reliable proxy for NDVI in rice and maize, even with low-cost RGB cameras (García et al., 2018GARCÍA, C.D.A.; RAMÓN, V.J.A.; ALZATE, V.D.F.; PALACIOS, G.J.R.: “Dynamics of the indices NDVI and GNDVI in a rice growing in its reproduction phase from multi-spectral aerial images taken by drones”, En: International Conference of ICT for Adapting Agriculture to Climate Change, Ed. Springer, pp. 106-119, 2018, DOI: https://doi.org/10.1007/978-3-030-04447-3_7. ; MDPI., 2019MDPI.: “Assessing correlation of high‑resolution NDVI with fertilizer application and rice yield”, Remote Sensing, 11(2): 112, 2019, ISSN: https://www.mdpi.com/2072-4292/11/2/112. ).
-
VARI is confirmed as the most robust RGB index for estimating vegetation vigor, consistent with international findings in wheat and rice crops (Ruiz et al., 2024RUIZ, P.M.E.; GARCÍA, R.R.; MACHADO, G.N.: “Aplicación de drones en la agricultura internacional y cubana. Revisión”, Revista Ciencias Técnicas Agropecuarias, 33(1), 2024, ISSN: 2071-0054.).
-
The moderate correlation of CIVE with NDRE suggests that RGB can capture chlorophyll variations, though with less precision than satellite indices. This result aligns with studies highlighting the usefulness of RGB indices for vegetation discrimination in the absence of NIR bands (Famartínez, 2023FAMARTÍNEZ, A.: Sentinel‑2 en R: cálculo de índices y correlación, [en línea], GitHub, 2023, Disponible en: https://github.com. ).
-
NDRE, although less correlated, provides additional sensitivity to chlorophyll and vigor, reinforcing its value as a complement to NDVI and VARI in agricultural monitoring (Huete, 1988HUETE, A.R.: “A soil-adjusted vegetation index (SAVI)”, Remote sensing of environment, 25(3): 295-309, 1988, ISSN: 0034-4257, DOI: https://doi.org/10.1016/0034-4257(88)90106-X. ).
-
RGB indices show significant correlations with satellite indices, confirming their utility as proxies at the seedling stage, as reported in precision agriculture projects in Latin America (FEDEARROZ, 2018FEDEARROZ: Hacia una agricultura de precisión en el cultivo del arroz. Federación Nacional de Arroceros de Colombia, [en línea], 2018, Disponible en: https://fedearroz-website-aws.s3.amazonaws.com/media/documents/cartilla_precision_2025.pdf. ; EcoTV Panamá., 2025ECOTV PANAMÁ.: Panamá revoluciona el cultivo de arroz con blockchain y agricultura de precisión, [en línea], Ecotvpanama, 2025, Disponible en: https://www.ecotvpanama.com. ).
-
The scatter plots confirm that RGB indices can be used as proxies for satellite indices at the seedling stage, validating drone-satellite integration for multiscale monitoring (AgroLatam, 2025AGROLATAM: Agricultura de precisión en América Latina: avances en drones y digitalización, [en línea], 2025, Disponible en: https://www.agrolatam.com. ).
Correlation Interpretation
⌅VARI-NDVI
⌅
-
Agronomic interpretation: The association between VARI and NDVI suggests that greenness captured in the visible spectrum (VARI) adequately reproduces the variability of vigor detected by NDVI with NIR, especially at the seedling stage where coverage is homogeneous and soil exposure is limited. UAV studies have shown that high-resolution NDVI correlates with fertilization management and yield in rice and wheat, reinforcing the validity of RGB proxies in low-cost monitoring schemes.
-
Technical interpretation: VARI, calculated from RGB images, consistently reproduces vegetation vigor captured by NDVI, which includes the NIR band. This validates the use of RGB drones as proxies under resource-limited conditions.
-
Guan et al. (2019)GUAN, S.; FUKAMI, K.; MATSUNAKA, H.; OKAMI, M.; TANAKA, R.; NAKANO, H.; SAKAI, T.; NAKANO, K.; OHDAN, H.; TAKAHASHI, K.: “Assessing correlation of high-resolution NDVI with fertilizer application level and yield of rice and wheat crops using small UAVs”, Remote Sensing, 11(2): 112, 2019, ISSN: 2072-4292, DOI: https://doi.org/10.3390/rs11020112. demonstrated that UAV-derived NDVI correlates with yield and fertilization in rice and wheat. Goigochea et al. (2024)GOIGOCHEA, D.; JUSTINO, P.M.; VEGA, H.S.S.; SANCHEZ, O.M.; LOBATO, G.R.H.; SANTILLAN, G.M.D.; GANOZA, R.J.J.; ORE, A.Z.L.; AGURTO, P.A.I.: “Yield prediction models for rice varieties using UAV multispectral imagery in the Amazon lowlands of Peru”, AgriEngineering, 6(3): 2955-2969, 2024, ISSN: 2624-7402, DOI: https://doi.org/10.3390/agriengineering6030170. confirmed that multitemporal indices such as NDVI predict rice yield with high accuracy.
ExG-GNDVI
⌅
-
Agronomic interpretation: The correspondence between ExG and GNDVI is expected in early rice stages, where vegetation-soil contrast and chlorophyll variability are strongly reflected in the green channel. UAV multispectral studies in rice have reported useful relationships between green-based indices and yield metrics, supporting the use of ExG as an operational indicator when spatial resolution and cost are prioritized.
-
Technical interpretation: ExG emphasizes the green component in RGB images, while GNDVI is sensitive to chlorophyll content. The correlation confirms that both indices capture nutritional variations in early rice stages.
-
García et al. (2018)GARCÍA, C.D.A.; RAMÓN, V.J.A.; ALZATE, V.D.F.; PALACIOS, G.J.R.: “Dynamics of the indices NDVI and GNDVI in a rice growing in its reproduction phase from multi-spectral aerial images taken by drones”, En: International Conference of ICT for Adapting Agriculture to Climate Change, Ed. Springer, pp. 106-119, 2018, DOI: https://doi.org/10.1007/978-3-030-04447-3_7. reported significant correlations between GNDVI and vigor in rice using UAVs. Goigochea et al. (2024)GOIGOCHEA, D.; JUSTINO, P.M.; VEGA, H.S.S.; SANCHEZ, O.M.; LOBATO, G.R.H.; SANTILLAN, G.M.D.; GANOZA, R.J.J.; ORE, A.Z.L.; AGURTO, P.A.I.: “Yield prediction models for rice varieties using UAV multispectral imagery in the Amazon lowlands of Peru”, AgriEngineering, 6(3): 2955-2969, 2024, ISSN: 2624-7402, DOI: https://doi.org/10.3390/agriengineering6030170. validated that green-based indices predict yield in the Peruvian Amazon.
CIVE-NDRE
⌅
-
Agronomic interpretation: NDRE is more sensitive to subtle changes in chlorophyll and nutritional status than RGB indices, especially in early stages and under slight variations in leaf density. The negative relationship suggests that increases in NDRE (higher chlorophyll) are associated with lower CIVE values under certain lighting and contrast conditions, consistent with studies recommending NDRE for physiological monitoring and water management in rice.
-
Technical interpretation: NDRE, derived from the red edge, is highly sensitive to chlorophyll and nitrogen, while CIVE combines RGB channels to extract vegetation. The inverse relationship reflects that increases in chlorophyll (high NDRE) are associated with lower CIVE values, due to differences in spectral sensitivity.
-
Nageswara et al. (2023)NAGESWARA, R.D.; BADRI, J.; SUNDARAM, R.: “Sentinel-2 Derived Indices to Understand Rice Vegetation Dynamics at Pixel Level”, En: National Conference on CONTROL INSTRUMENTATION SYSTEM CONFERENCE, Ed. Springer, pp. 153-165, 2023, DOI: https://doi.org/10.1007/978-981-97-5157-0_13. confirmaron la utilidad de NDRE para detectar variaciones fisiológicas confirmed the usefulness of NDRE for detecting physiological variations in rice. Pedroso de Lima et al. (2021)PEDROSO DE LIMA, I.; JORGE, R.G.; DE LIMA, J.L.P.: “Remote sensing monitoring of rice fields: Towards assessing water saving irrigation management practices”, Frontiers in Remote Sensing, 2: 762093, 2021, ISSN: 2673-6187, DOI: https://doi.org/10.3389/frsen.2021.762093. highlighted NDRE as key for evaluating irrigation and nutrition practices in rice.
Field Validation
⌅
-
Direct observations in plot 6 confirmed the absence of water stress, consistent with the high NDVI and VARI values that reflected homogeneous vegetation vigor across the plot.
-
The homogeneity of the seedling stage observed in the field corresponded with the spectral patterns of the thematic maps, where NDVI and VARI showed high and consistent values in the center and edges of the plot.
-
Subtle NDRE variations in central and eastern areas were related to slight differences in leaf density observed in the field, confirming its sensitivity to chlorophyll content.
-
This correspondence between field observations and spectral results validates the use of RGB indices as reliable proxies for satellite indices, in line with international studies reporting similar correlations in rice and other crops (García et al., 2018GARCÍA, C.D.A.; RAMÓN, V.J.A.; ALZATE, V.D.F.; PALACIOS, G.J.R.: “Dynamics of the indices NDVI and GNDVI in a rice growing in its reproduction phase from multi-spectral aerial images taken by drones”, En: International Conference of ICT for Adapting Agriculture to Climate Change, Ed. Springer, pp. 106-119, 2018, DOI: https://doi.org/10.1007/978-3-030-04447-3_7. ; Quille et al., 2024QUILLE, M.J.A.; RUIZ, L.A.; RAMOS-FERNÁNDEZ, L.: “Multiseasonal analysis of rice crop yield prediction with Sentinel-2 time series and UAV imagery in Lambayeque (Peru)”, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 48: 421-426, 2024b, ISSN: 1682-1750, DOI: https://doi.org/10.5194/isprs-archives-XLVIII-3-421-2024. ; Wijayanto et al., 2024WIJAYANTO, Y.; SAFITRI, M.; PURNAMASARI, I.; BUDIMAN, S.A.; SAPUTRA, T.W.; REGAR, A.F.; RISTIYANA, S.: “Estimating the contents of Chlorophyll, Nitrogen, and Yields on Rice Through Sentinel-2 Vegetation Indices in Heterogeneous Land Management”, The Indonesian Journal of Geography, 56(3): 357-367, 2024, ISSN: 0024-9521, DOI: https://doi.org/10.22146/ijg.87159.).
Conclusions
⌅
-
RGB indices derived from drone imagery showed significant correlations with Sentinel‑2 multispectral indices, validating their use as reliable proxies under resource‑limited conditions.
-
The correlation between VARI and NDVI reached r ≈ 0.74, which corresponds to a very strong correlation according to Cohen’s (1988) scale.
-
The integration of drone and satellite imagery constitutes an effective strategy for multiscale agricultural monitoring, enabling characterization of rice physiological status at early stages.
-
Limitations of RGB cameras were identified, related to their reduced sensitivity to specific spectral variations, suggesting the need to complement them with multispectral sensors in future studies.
Recomendations
⌅
-
Implement the combined use of RGB drone and Sentinel-2 imagery for multiscale agricultural monitoring under resource-limited conditions.
-
Extend correlation validation to different phenological stages of rice and to other strategic crops in the region.
-
Complement RGB cameras with multispectral or hyperspectral sensors in future studies to improve sensitivity to specific spectral variations.
-
Integrate results into local precision agriculture programs, strengthening decision-making on irrigation and fertilization.
-
Promote training of technicians and producers in the use of remote sensing tools (QGIS, drones, Sentinel-2) to ensure sustainability and replicability.