Evaluation of UAV-Based RGB and Multispectral Vegetation Indices for Precision Agriculture in Palm Tree Cultivation

Alavikunhu Panthakkan; Hussain Al-Ahmad; K. Sherin; Saeed Al Mansoori; S M Anzar

REVIEW 4 major objections 6 minor 12 references

Evaluation of UAV-Based RGB and Multispectral Vegetation Indices for Precision Agriculture in Palm Tree Cultivation

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read RGB drone imagery can match multispectral sensors for palm health monitoring.

desk verdict The paper's claim that RGB indices match multispectral for palm health is unsupported: the aggregate percentages are artifacts of hand-set thresholds, with no pixel-level agreement or ground truth. read the letter →

arxiv 2505.07840 v1 pith:GJVG6UIB submitted 2025-05-06 eess.IV cs.CV

Alavikunhu Panthakkan , S M Anzar , K. Sherin , Saeed Al Mansoori , Hussain Al-Ahmad This is my paper · ORCID

classification eess.IVcs.CV

keywords precisionagricultureUAVremotesensingvegetationindicesNDVISAVIVARIMGRVIpalmtreecultivation

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that vegetation indices computed from ordinary RGB drone photographs—VARI and MGRVI—classify palm-tree health as reliably as the costlier multispectral indices NDVI and SAVI. On a palm cultivation plot in arid Dubai it finds that all four indices place roughly the same share of the field into non-vegetation, stressed, moderate, and dense categories, with the stressed class around 61% of the vegetated area and the dense class near 0.2%. A sympathetic reader would take this as evidence that routine crop health monitoring could be done with cheap RGB cameras instead of multispectral sensors, lowering the barrier to precision farming in resource-limited regions.

What carries the argument

The argument is carried by four vegetation-index formulas paired with hand-set threshold ranges. NDVI = (NIR − R)/(NIR + R) and SAVI = (NIR − R)(1 + L)/(NIR + R + L) with L = 0.5 use near-infrared plus red reflectance; VARI = (G − R)/(G + R − B) and MGRVI = (G² − R²)/(G² + R²) use only visible light. Each index map is partitioned by thresholds, for example stressed vegetation is 0.1 < NDVI ≤ 0.3 but 0.08 < VARI ≤ 0.22, into non-vegetation, stressed, moderate, and dense classes. The near-identical class percentages from those partitions are the evidence that the two sensor classes perform alike.

What would settle it

Reclassify the same drone images after replacing the paper's index-specific thresholds with thresholds calibrated to field-scored palm health or a single shared percentile rule; if VARI and MGRVI then assign different trees to stressed versus moderate classes on more than a few percent of the canopy compared with NDVI and SAVI, the claimed comparability would fail.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that RGB-only indices reproduce the multispectral picture of a palm plot: NDVI and SAVI assign 76.15% and 75.59% of the scene to non-vegetation and split the remaining vegetation into roughly 61% stressed, 38% moderate, and 0.2% dense classes, while VARI and MGRVI assign 76.24% and 76.58% to non-vegetation and split vegetation into 61–62% stressed, about 38% moderate, and about 0.2% dense. Because the RGB shares match the multispectral shares to within a percent or two, the paper claims that RGB-based indices offer comparable performance for stress detection and health classification, making them a cost-effective alternative for large-scale agricultural monitoring.

Load-bearing premise

The whole comparison rests on the assumption that the threshold ranges used to split each index into stressed, moderate, and dense classes are independently valid; if those thresholds were tuned to make RGB and multispectral percentages agree, the similarity would be built in.

Editorial extensions

If this is right

If the claim holds, routine palm health screening can be done with RGB-only drone surveys, avoiding the higher purchase and maintenance cost of multispectral sensors.
If the claim holds, the stressed, moderate, and dense class shares can serve as a low-cost early-warning signal for irrigation decisions in arid palm cultivation.
If the claim holds, agriculture agencies could map stress across many plots with standard consumer drones rather than specialized NIR cameras.
If the claim holds, the same index-plus-threshold workflow transfers to other visible-light drone images of palms, producing a consistent three-way health map.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Going beyond the paper, the same threshold-comparison protocol could be run on olive or citrus orchards to see whether RGB indices track multispectral shares when canopy geometry differs from palms.
Going beyond the paper, the roughly 61% stressed share suggests a predominantly water-limited grove, but linking index classes to measured soil moisture or leaf water potential would be needed to turn that pattern into an irrigation recommendation.
Going beyond the paper, if the thresholds prove stable across seasons and sites, the cost advantage of RGB could be quantified as a per-hectare saving; the paper does not price that saving.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take on arXiv:2505.07840. It's a UAV palm-monitoring study comparing RGB indices (VARI, MGRVI) with multispectral ones (NDVI, SAVI). The headline claim is that RGB matches multispectral for classifying palm health. The evidence does not support that claim. The paper reports nearly identical class-area percentages across the four indices, but those percentages are direct functions of thresholds that are reported without derivation and look adjusted to match. No pixel-level comparison, no ground truth. So the central result is essentially circular.

What's new: a new site (palm grove in Dubai) with real UAV imagery. The index definitions are standard and correctly stated. The authors do include a caveat in the Discussion that multispectral may be better in dense vegetation, which slightly softens the abstract's 'without compromising accuracy.'

The soft spots are not minor. Tables II and IV are the whole argument. The stress threshold for NDVI is 0.1–0.3; for VARI it's 0.08–0.22; dense thresholds are all near 0.6. Yet all four indices yield roughly 61% stressed, 38% moderate, 0.2% dense. That is only possible if thresholds were chosen to produce similar proportions, or the distributions are so different that the thresholds happen to align—either way, you cannot infer comparable classification from aggregate percentages. Two binary maps with identical areas can disagree at every pixel. The paper never reports a confusion matrix, spatial overlap, or accuracy against field-measured health. It also omits basic acquisition details (platform, altitude, resolution, number of trees).

The literature review is fine and cites prior RGB-vs-multispectral comparisons, so the novelty is low beyond a local case study.

Would I send this to peer review? I would not. The evidence presented is not enough to referee meaningfully; it needs a redesign or at least a per-pixel comparison and threshold justification. As is, it's more of a cautionary example than a citable result.

Referee Report

4 major / 6 minor

Summary. The paper evaluates four vegetation indices computed from UAV imagery of a palm tree cultivation area in Dubai: NDVI and SAVI from multispectral data, and VARI and MGRVI from RGB data. The indices are thresholded into non-vegetation, stressed, moderate, and dense vegetation classes, and the percentage of area in each class is reported in Tables I–IV. On the basis of similar class-area percentages between the RGB-based and multispectral-based indices, the authors claim that RGB-based indices offer performance comparable to multispectral indices and represent a cost-effective alternative for precision agriculture. The paper includes vegetation maps, pie charts, and a brief discussion of the comparative results. The central claim is that RGB imaging can replace multispectral imaging for routine palm health monitoring without compromising accuracy.

Significance. If the central claim were established, the paper would address a practically important question: whether low-cost RGB cameras can substitute for more expensive multispectral sensors in arid-region precision agriculture. The study targets a real palm cultivation area and includes both multispectral and RGB indices, which is a sensible starting point. However, the reported evidence is not sufficient to support the claim. The comparison rests entirely on aggregate area percentages in broad value ranges; no pixel-level agreement, confusion matrix, statistical test, or ground-truth plant-health labels are provided. Moreover, the classification thresholds are asserted without derivation and differ across indices, making the apparent agreement between RGB and multispectral percentages potentially an artifact of threshold selection. The paper is therefore better read as a feasibility demonstration of computing these indices, not as a validated comparison of their accuracy. The significance of the intended application is real, but the measurement of 'comparable performance' is missing.

major comments (4)

[§IV.A–B and Tables I–IV] The conclusion that RGB indices are comparable to multispectral indices rests solely on aggregate percentages of pixels assigned to broad classes. Two classifications can have identical class-area fractions while disagreeing at every pixel, so the reported percentages do not establish comparable classification performance. A quantitative pixel-level comparison is required, such as a confusion matrix between index-based maps, Cohen's kappa, Dice overlap, or per-class agreement statistics. Without such measures, the central claim is unsupported by the reported evidence.
[§IV and Tables II and IV] The thresholds used to define stress, moderate, and dense classes are presented without any derivation or justification, and they differ between indices. For example, NDVI stress is 0.1 < NDVI ≤ 0.3, while VARI stress is 0.08 < VARI ≤ 0.22, and the dense thresholds also differ. Since the conclusion is based on the similarity of class percentages, thresholds that were selected to align those percentages would make the comparison circular. The authors need to justify each threshold independently, for example from literature, from a data-driven procedure such as Otsu thresholding, or from calibration against ground-truth plant health, and should include a sensitivity analysis showing how the class percentages vary with threshold choices.
[§IV.C and Abstract/Conclusion] No ground-truth information is used anywhere in the evaluation. The 'stress', 'moderate', and 'dense' categories are defined solely by thresholded index values, so the paper never measures whether either the RGB or the multispectral classification is actually correct. 'Comparable performance' is an accuracy claim, and accuracy cannot be assessed without independent validation data such as field-assessed palm health, leaf chlorophyll measurements, or expert-labeled reference maps. Without such validation, the most that can be claimed is that the two index families produce similar area fractions under the chosen thresholds.
[§IV.C and Abstract] The Discussion states that multispectral indices 'may offer better accuracy' in areas with dense vegetation, and the Conclusion repeats that multispectral indices offer 'greater sensitivity.' This is in tension with the Abstract's assertion that RGB indices are comparable 'without compromising accuracy' and that RGB provides 'similar results in vegetation classification and stress detection.' If dense vegetation is poorly captured by RGB indices, the blanket conclusion should be qualified, and the scope of the claimed comparability should be stated precisely. The current wording overstates the findings.

minor comments (6)

[§IV.A and Figures 1–4] The text in §IV.A refers to 'Figures 1(a) and 1(b)' as the NDVI and SAVI maps, but Figure 1 is the methodology flowchart; the actual vegetation maps appear to be Figures 3(a) and 3(b). The same cross-referencing problem affects the RGB figures. All figure callouts should be corrected.
[Table IV] Table IV contains the inverted character '¡' in the range notation, e.g., '0.22 ¡ V ARI ≤ 0.59' and '0.2 ¡ MGRVI ≤ 0.59'; these should be '<'.
[§III.B] The database and study-area section does not report sensor specifications, flight altitude, image resolution, date and time of acquisition, solar angle, or radiometric calibration procedures. Such details are necessary for reproducibility and for interpreting comparisons between RGB and multispectral data.
[References] References [8] and [9] are incomplete: [8] lacks full author and volume/page information consistent with a 2022 article, and [9] lacks author names and a full title. The authors should verify and complete all bibliographic entries.
[Author affiliations] The second author is listed with two different affiliation numbers on the same line, and the manuscript includes a repeated '2' affiliation. The affiliation list should be cleaned up.
[Throughout] There are numerous typographical and formatting issues, such as inconsistent spacing in 'UA V', 'SA VI', 'V ARI', and 'M GRVI'. A careful proofreading pass is needed.

Circularity Check

1 steps flagged · score 6.0 of 10

The headline 'comparable performance' claim rests entirely on threshold-defined class percentages, so the similar shares are built into the chosen index ranges rather than independently demonstrated.

self definitional [Section IV-B and Discussion IV-C, Tables III-IV]
"For V ARI, 61.08% of the vegetation was stressed ( 0.08 < V ARI ≤ 0.22), 38.70% moderate (0.22 < V ARI ≤ 0.59), and 0.23% dense ( V ARI > 0.59)... This indicates that RGB-based indices provide results comparable to multispectral indices."

The reported stress/moderate/dense percentages are, by construction, the empirical cumulative fractions of each index's pixel values falling inside the hand-set threshold brackets. The thresholds are not anchored to any ground-truth health labels or to a standard independent classifier, and they differ across indices (NDVI stress 0.1-0.3 vs VARI stress 0.08-0.22). Therefore the similarity between RGB and multispectral 'classifications' is a similarity between the chosen threshold brackets, not a measured agreement between two independently validated classifications. The conclusion that RGB indices are comparable follows directly from these threshold-defined percentages; it is a restatement of the threshold choices rather than an independent finding.

full rationale

The index formulas themselves (NDVI, SAVI, VARI, MGRVI) are standard and are not circular; the computation of each index from its bands is self-contained. The circularity is confined to the evidence used to support the abstract's claim that RGB indices offer performance comparable to multispectral indices. The only comparative evidence reported is the percentage of pixels falling into categories defined by manually chosen thresholds. For any index, those percentages are mathematically determined by the chosen thresholds, so comparing the percentages across indices is comparing the thresholds' locations in each index's empirical distribution, not comparing classification accuracy. Because the thresholds are unexplained, differ per index, and produce nearly identical shares, the similarity is effectively an input to the analysis rather than a discovered result. I am not flagging the absence of ground truth as circularity by itself; but here the absence matters because the threshold-defined shares are the sole basis for the 'comparable performance' conclusion, making that conclusion equivalent to the threshold choices by construction. There is no load-bearing self-citation: the cited references are external and do not supply the key comparison. The appropriate circularity score is 6, reflecting partial circularity of the central claim, while the underlying index computations remain independent.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central comparison rests on threshold selections and unstated assumptions about image alignment and ground truth. The paper contributes standard index formulas but no independent benchmark.

free parameters (4)

NDVI stress/moderate/dense thresholds = 0.1, 0.3, 0.6
Used in Table II to split vegetation classes; no derivation is given, and the thresholds appear chosen to make NDVI class shares match RGB indices.
SAVI stress/moderate/dense thresholds = 0.16, 0.33, 0.64
Used in Table II; no derivation, and the shifted thresholds produce nearly the same percentages as NDVI.
VARI stress/moderate/dense thresholds = 0.08, 0.22, 0.59
Used in Table IV; no agronomic basis is stated, and the thresholds yield class shares almost identical to NDVI and SAVI.
MGRVI stress/moderate/dense thresholds = 0.08, 0.2, 0.59
Used in Table IV; no derivation, again producing near-identical class percentages.

assumptions (4)

domain assumption Standard vegetation index formulas (NDVI, SAVI, VARI, MGRVI) are correctly applied to UAV imagery
Equations (1)-(4) are standard, but no radiometric calibration or processing details verify that the computed maps match ground reflectance.
domain assumption RGB and multispectral image sets observe the same plot and are spatially aligned
The comparison implies the two sensor streams cover the same trees; no co-registration or acquisition geometry is described in Section III.
ad hoc to paper The chosen class thresholds correspond to real vegetation health states
No ground truth or field survey is used; thresholds in Tables II and IV are stated without any agronomic justification.
domain assumption Multispectral indices are a valid reference for RGB comparison
Even the multispectral classification is not validated, yet it is treated as the benchmark for the RGB comparison.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Evaluation of UAV-Based RGB and Multispectral Vegetation Indices for Precision Agriculture in Palm Tree Cultivation." pith.science (2026). https://pith.science/paper/GJVG6UIB

@misc{pith2026250507840,
  author       = {Pith},
  title        = {Pith review of: Evaluation of UAV-Based RGB and Multispectral Vegetation Indices for Precision Agriculture in Palm Tree Cultivation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GJVG6UIB}},
  note         = {Machine review of arXiv:2505.07840}
}

read the original abstract

Precision farming relies on accurate vegetation monitoring to enhance crop productivity and promote sustainable agricultural practices. This study presents a comprehensive evaluation of UAV-based imaging for vegetation health assessment in a palm tree cultivation region in Dubai. By comparing multispectral and RGB image data, we demonstrate that RGBbased vegetation indices offer performance comparable to more expensive multispectral indices, providing a cost-effective alternative for large-scale agricultural monitoring. Using UAVs equipped with multispectral sensors, indices such as NDVI and SAVI were computed to categorize vegetation into healthy, moderate, and stressed conditions. Simultaneously, RGB-based indices like VARI and MGRVI delivered similar results in vegetation classification and stress detection. Our findings highlight the practical benefits of integrating RGB imagery into precision farming, reducing operational costs while maintaining accuracy in plant health monitoring. This research underscores the potential of UAVbased RGB imaging as a powerful tool for precision agriculture, enabling broader adoption of data-driven decision-making in crop management. By leveraging the strengths of both multispectral and RGB imaging, this work advances the state of UAV applications in agriculture, paving the way for more efficient and scalable farming solutions.

Figures

Figures reproduced from arXiv: 2505.07840 by the authors.

**Figure 2.** Palm Tree Region This methodology follows a structured approach, starting with image acquisition, followed by data preprocessing, calculation of vegetation indices, and classification of vegetation health. The study emphasizes a comparative analysis of the cost-effectiveness and accuracy of RGB-based indices versus traditional multispectral indices, particularly in the context of arid regions like Dubai. The block … view at source ↗

**Figure 3.** NDVI and SAVI maps of the selected palm plot [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Visual Representation of Vegetation Index for RGB Image [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

**Figure 5.** Figure 5: Pie chart of vegetation indices (NDVI and SAVI) using multispectral images [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]

**Figure 6.** Figure 6: Pie chart of vegetation indices (VARI and MGRVI) of RGB images [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

12 extracted references · 12 canonical work pages

[1]

Application of precision agri- culture technologies for sustainable crop production and environmental sustainability: A systematic review,

S. Getahun, H. Kefale, and Y . Gelaye, “Application of precision agri- culture technologies for sustainable crop production and environmental sustainability: A systematic review,” The Scientific World Journal , vol. 2024, no. 1, p. 2126734, 2024

work page 2024
[2]

Viable smart sensors and their application in data driven agriculture,

K. Paul, S. S. Chatterjee, P. Pai, A. Varshney, S. Juikar, V . Prasad, B. Bhadra, and S. Dasgupta, “Viable smart sensors and their application in data driven agriculture,” Computers and Electronics in Agriculture , vol. 198, p. 107096, 2022

work page 2022
[3]

A review on the use of unmanned aerial vehicles and imaging sensors for monitoring and assessing plant stresses,

J. G. A. Barbedo, “A review on the use of unmanned aerial vehicles and imaging sensors for monitoring and assessing plant stresses,” Drones, vol. 3, no. 2, p. 40, 2019

work page 2019
[4]

A modified soil adjusted vegetation index,

J. Qi, A. Chehbouni, A. R. Huete, Y . H. Kerr, and S. Sorooshian, “A modified soil adjusted vegetation index,”Remote sensing of environment, vol. 48, no. 2, pp. 119–126, 1994

work page 1994
[5]

A soil-adjusted vegetation index (savi),

A. Huete, “A soil-adjusted vegetation index (savi),” Remote Sensing of Environment, vol. 25, no. 3, pp. 295–309, 1988

work page 1988
[6]

Evaluating multispectral images and vegetation indices for precision farming applications from uav images,

M. Dubbini, M. Gattelli, and R. Scopigno, “Evaluating multispectral images and vegetation indices for precision farming applications from uav images,” Remote Sensing, vol. 7, no. 4, pp. 4026–4047, 2015

work page 2015
[7]

Combining uav-based plant height from crop surface models, visible, and near-infrared vegetation indices for biomass monitoring in barley,

J. Bendig, A. Bolten, S. Bennertz, J. Broscheit, S. Eichfuss, and G. Bareth, “Combining uav-based plant height from crop surface models, visible, and near-infrared vegetation indices for biomass monitoring in barley,” International Journal of Applied Earth Observation and Geoinformation, vol. 39, pp. 79–87, 2015

work page 2015
[8]

Vegetation indices derived from remote sensing to assess crop growth and yield,

A. Gitelson, D. Rundquist, and Y . Kaufman, “Vegetation indices derived from remote sensing to assess crop growth and yield,” Agricultural and Forest Meteorology, vol. 112, no. 1, pp. 15–20, 2022

work page 2022

Show all 12 references

[9]

Evaluation of rgb vegetation indices derived from uav images for rice crop growth monitoring,

“Evaluation of rgb vegetation indices derived from uav images for rice crop growth monitoring,” ISPRS Annals of Photogrammetry, Remote Sensing and Spatial Information Sciences , vol. X-4, pp. 385–395, 2022

2022
[10]

Comparison of vegetation indices acquired from rgb and multispectral uav data,

C. Zhang and J. M. Kovacs, “Comparison of vegetation indices acquired from rgb and multispectral uav data,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–10, 2022

2022
[11]

A commentary review on the use of normalized difference vegetation index (ndvi) in the era of popular remote sensing,

S. Huang, L. Tang, J. P. Hupy, Y . Wang, and G. Shao, “A commentary review on the use of normalized difference vegetation index (ndvi) in the era of popular remote sensing,” Journal of Forestry Research , vol. 32, no. 1, pp. 1–6, 2021

2021
[12]

Uav application to estimate oil palm trees health using visible atmospherically resistant index (vari)(case study of cikabayan research farm, bogor city),

M. N. Anisa, R. Hernina et al. , “Uav application to estimate oil palm trees health using visible atmospherically resistant index (vari)(case study of cikabayan research farm, bogor city),” in E3S Web of Con- ferences, vol. 211. EDP Sciences, 2020, p. 05001. 8

2020

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Application of precision agri- culture technologies for sustainable crop production and environmental sustainability: A systematic review,

S. Getahun, H. Kefale, and Y . Gelaye, “Application of precision agri- culture technologies for sustainable crop production and environmental sustainability: A systematic review,” The Scientific World Journal , vol. 2024, no. 1, p. 2126734, 2024

work page 2024

[2] [2]

Viable smart sensors and their application in data driven agriculture,

K. Paul, S. S. Chatterjee, P. Pai, A. Varshney, S. Juikar, V . Prasad, B. Bhadra, and S. Dasgupta, “Viable smart sensors and their application in data driven agriculture,” Computers and Electronics in Agriculture , vol. 198, p. 107096, 2022

work page 2022

[3] [3]

A review on the use of unmanned aerial vehicles and imaging sensors for monitoring and assessing plant stresses,

J. G. A. Barbedo, “A review on the use of unmanned aerial vehicles and imaging sensors for monitoring and assessing plant stresses,” Drones, vol. 3, no. 2, p. 40, 2019

work page 2019

[4] [4]

A modified soil adjusted vegetation index,

J. Qi, A. Chehbouni, A. R. Huete, Y . H. Kerr, and S. Sorooshian, “A modified soil adjusted vegetation index,”Remote sensing of environment, vol. 48, no. 2, pp. 119–126, 1994

work page 1994

[5] [5]

A soil-adjusted vegetation index (savi),

A. Huete, “A soil-adjusted vegetation index (savi),” Remote Sensing of Environment, vol. 25, no. 3, pp. 295–309, 1988

work page 1988

[6] [6]

Evaluating multispectral images and vegetation indices for precision farming applications from uav images,

M. Dubbini, M. Gattelli, and R. Scopigno, “Evaluating multispectral images and vegetation indices for precision farming applications from uav images,” Remote Sensing, vol. 7, no. 4, pp. 4026–4047, 2015

work page 2015

[7] [7]

Combining uav-based plant height from crop surface models, visible, and near-infrared vegetation indices for biomass monitoring in barley,

J. Bendig, A. Bolten, S. Bennertz, J. Broscheit, S. Eichfuss, and G. Bareth, “Combining uav-based plant height from crop surface models, visible, and near-infrared vegetation indices for biomass monitoring in barley,” International Journal of Applied Earth Observation and Geoinformation, vol. 39, pp. 79–87, 2015

work page 2015

[8] [8]

Vegetation indices derived from remote sensing to assess crop growth and yield,

A. Gitelson, D. Rundquist, and Y . Kaufman, “Vegetation indices derived from remote sensing to assess crop growth and yield,” Agricultural and Forest Meteorology, vol. 112, no. 1, pp. 15–20, 2022

work page 2022