Next Article in Journal
Regional Differences and Dynamic Evolution of Agricultural Product Market Integration in China
Previous Article in Journal
Analysis of the Impact of Information Behavior on the Marketing and Fertilization Strategies of Small Cocoa Producers in the Provinces of Guayas and Los Ríos in Ecuador
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Agriculture
Volume 15
Issue 8
10.3390/agriculture15080859
agriculture-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

A Phenology-Based Evaluation of the Optimal Proxy for Cropland Suitability Based on Crop Yield Correlations from Sentinel-2 Image Time-Series

by
Dorijan Radočaj
* and
Mladen Jurišić
Faculty of Agrobiotechnical Sciences Osijek, Josip Juraj Strossmayer University of Osijek, Vladimira Preloga 1, 31000 Osijek, Croatia
*
Author to whom correspondence should be addressed.
Agriculture 2025, 15(8), 859; https://doi.org/10.3390/agriculture15080859
Submission received: 5 March 2025 / Revised: 9 April 2025 / Accepted: 14 April 2025 / Published: 15 April 2025
(This article belongs to the Section Digital Agriculture)
Browse Figures
Versions Notes

Abstract

:
Cropland suitability calculations quantify natural suitability according to abiotic conditions, thus making them crucial for sustainable land management. However, since ground-truth yield data are extremely scarce, there is a need to improve knowledge on the optimal proxy metric from satellite imagery, which represents cropland suitability and enables global applicability. This study evaluated four frequently used vegetation indices from Sentinel-2 image time-series (normalized difference vegetation index, enhanced vegetation index, enhanced vegetation index 2, and wide dynamic range vegetation index) with three phenology metrics for correlation analysis with maize and soybean yield. Four years (2019–2022) in two study areas (Iowa and Illinois) were utilized in this research, and 1000 ground-truth crop yield samples were created for each combination of study year and area. The combination of wide dynamic range vegetation index (WDRVI) and maximum vegetation index phenology metric (MAX) was an optimal proxy for maize yield prediction, while enhanced vegetation index 2 (EVI2) and MAX produced the highest correlation for soybean, producing Pearson’s correlation coefficient means of 0.506 and 0.519, respectively. This study improved our knowledge of the optimal proxy metric for cropland suitability by combining multiple large ground-truth crop yield datasets with 30 m spatial resolution satellite imagery, which can be further improved with the use of novel vegetation indices with improved resistance to a saturation effect.

1. Introduction

The projected growth of the global population to 9.7 billion by 2050 will create a massive food demand, which will require agricultural systems to produce at unprecedented levels [1]. The constraints imposed by urbanization together with soil degradation and climate change limit the expansion of cropland, so it is vital to enhance existing cropland utilization [2]. The determination of suitable croplands has become essential to determine areas that offer maximum agricultural productivity alongside minimal environmental degradation [3]. However, the prediction of suitable cropland areas represents an ambiguous concept because of the lack of a standardized measurement method to represent its suitability levels. The definition of a single quantitative proxy is a prerequisite for advanced predictive methods based on machine learning, which were proven to be superior to conventional geographic information system (GIS)-based multicriteria analysis by providing an objective and robust prediction with a straightforward accuracy assessment [4]. The most complete measure for assessing cropland suitability is generally crop yield data, which directly demonstrate the productivity level of specific areas [5]. However, there is a lack of dependable extensive crop yield databases worldwide, especially in areas where ground-based monitoring systems are underdeveloped [6]. This suggests that the knowledge of a proxy for cropland suitability should be based on globally accessible and reliable data sources, especially satellite-based vegetation indices.
Sentinel-2 satellite missions, alongside other satellite missions, have made vegetation indices more accessible and accurate, thus enabling their use in agricultural monitoring with a spatial resolution up to 10 m [7]. The modern agricultural monitoring sector relies heavily on vegetation indices which provide dependable scalable solutions to evaluate plant health and monitor growth and productivity levels [8]. Numerous vegetation indices, such as normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI), were reliably used for decades for vegetation health, biomass assessment, and crop stress indicators [9], all of which are directly related to cropland suitability [10]. The NDVI utilizes the contrast between near-infrared and red light reflectance to assess vegetation density while evaluating its vigor and is currently the most frequently used vegetation index in agricultural studies [11]. While widely used, it is particularly sensitive to saturation in high-biomass conditions, limiting its ability to differentiate between dense vegetation stages, which are especially important for crop yield assessment [12]. The EVI and the wide dynamic range vegetation index (WDRVI) contain modifications that boost their sensitivity across diverse vegetation conditions, especially when applied to high-biomass or heterogeneous landscapes [13]. So far, previous research has thoroughly demonstrated the correlation between various vegetation indices and crop yields of major crops under various agroecological conditions. However, these studies were dominantly based on either coarse-resolution satellite imagery [14,15,16,17] or generalized ground-truth yield data based on administrative units [15,18,19], meaning that spectral mixing with other land cover classes and disregarding in-field variability were frequently unavoidable. Newer studies resolved these issues by using yield monitor data from a combine harvester as ground-truth data [20,21,22,23]. While this approach enabled the reliable assessment of crop yield correlation with vegetation indices, these data did not include all crop varieties and cropland management approaches present in the study area, which is critical for cropland suitability studies. Therefore, present research lacks knowledge on the correlation between crop yield and vegetation indices, which cover a large quantity of ground-truth data while maintaining high spatial resolution, determining a single optimal proxy for cropland suitability levels.
Crop growth dynamics alongside yield correlate strongly with phenology metrics, which describe periodic biological events measured within seasonal climate contexts [13]. Phenology metrics enable an understanding of the key developmental stages through their measurements of the start of the growing season (SOS), peak growing season (POS), and end of the growing season (EOS) [24]. Unlike yield prediction studies, which generally benefit from accurately predicting crop yield based on early phenology stages, the existing research on cropland suitability prediction primarily focuses on either maximum vegetation index or vegetation index values at POS as a proxy for suitability [4,25,26]. Crop conditions evaluated at selected moments using these metrics may lack the ability to capture vital changes in crop growth patterns that are necessary for accurate yield forecasting and suitability evaluations [27]. Peak vegetation index measurements show crop status in a single timeframe without considering seasonal growth accumulation, which can be overcome by area under the curve (AUC) phenology metric, demonstrating season-long vegetation growth. The level of AUC measurement indicates long-term plant advancement, while a minimal AUC shows extended periods of crop stress or unfavorable growing environments [28]. One of the main constraints in the present state of research is the already mentioned usage of localized small-scale datasets because this limits how well the obtained results can scale up to diverse agricultural systems and broader geographic regions. The optimal proxy for cropland suitability should use crop yield data, but existing reliable large-scale yield databases present significant limitations; thus, the need for substitute proxies emerges from this gap, giving rise to vegetation indices combined with phenology metrics.
To address the lack of research on the optimal combination of vegetation indices from satellite imagery and phenology metrics to determine the optimal proxy metric for cropland suitability, the objective of this study was to evaluate four frequently used vegetation indices from Sentinel-2 image time-series with three phenology metrics for correlation analysis with maize and soybean yield. Moreover, to mitigate the annual effects of weather effects and land management, such as crop rotation systems, two study areas under four consecutive study years were utilized to meet this study’s objective.

2. Materials and Methods

Determination of the optimal proxy for cropland suitability based on crop yield correlations for maize and soybean was performed in four fundamental steps: (1) creation of random ground-truth crop yield samples individually for four study years and two study areas; (2) calculation of time-series for each of the ground-truth samples according to four vegetation indices, which were selected with regard to their frequency in recent studies indexed in the Web of Science Core Collection; (3) phenology analysis according to four vegetation indices and their combination with three phenology metrics, which represented proxy metrics for cropland suitability; and (4) correlation analysis of proxy metrics with ground-truth crop yield datasets, as well as correlation analysis of calculated phenology metrics to explore the presence of multicollinearity.

2.1. Study Area and Ground-Truth Crop Yield Data

The study area included Iowa and Illinois, which represent two of the top agricultural regions of the United States from 2019 to 2022 (Figure 1). These states are contained in the United States’ Corn Belt central territory, characterized by soil with high organic matter content and nitrogen levels [29,30]. The hot summer humid continental climate dominates the study area, classified as “Dfa” according to the Köppen climate classification. The Quantile Loss Domain Adversarial Neural Networks (QDANN) 30 m yield map was used as a data source for maize and soybean yield, providing subfield-level yield data with a 30 m spatial resolution for high-resolution yield estimation [31]. The QDANN framework draws county-level dataset information to generate precise yield maps at high spatial resolutions to overcome limited ground-truth data availability for training and evaluation [32]. A total of 1000 random points of both maize and soybean per state every year were generated to include diverse abiotic conditions across the study area. Their locations slightly differed on a year-to-year basis as the distribution of maize and soybean parcels changed due to crop rotation practices (Table 1).

2.2. Calculation of Vegetation Iindices

Harmonized Sentinel-2 Level-2A (L2A) image time-series was analyzed using the Google Earth Engine for vegetation index calculations [33], which were used as the input for phenology modeling separately for each of the four study years. These images provided bottom-of-atmosphere (BOA) reflectance, and they were preprocessed based on the Sen2Cor atmospheric correction algorithm for the conversion of top-of-atmosphere reflectance to BOA reflectance. To match the spatial resolution of QDANN data, all bands were resampled to 30 m spatial resolution prior to the vegetation index calculation. The images were filtered according to a cloud probability of less than 5% to reduce atmospheric interference. Four vegetation indices were selected in this study, with NDVI, EVI, and EVI2 being selected according to the highest frequency in recent studies indexed in the Web of Science Core Collection (Figure 2, Table 2), and their values were added to randomly generated points using a reducer operation from the Google Earth Engine. The NDVI is widely adopted for its simplicity and strong correlation with green biomass in early growth stages, though it is prone to saturation in dense canopies, such as mid-to-late-season maize [34]. EVI and EVI2 were included to mitigate this limitation, as their formulations reduce atmospheric and soil background effects, improving sensitivity in high-biomass conditions typical of soybean and maize fields [35]. Moreover, the WDRVI was used to evaluate the claim by Gitelson [36] that it allows for a more robust characterization of crop phenological properties than the NDVI and provides a linear response to high LAI, making it advantageous for yield prediction in intensive cropping systems, despite being underrepresented in studies which analyzed correlation with crop yield data.

2.3. Phenology Analysis and Calculation of Vegetation Indices at Key Phenology Stages

According to these four vegetation index values of all the available Sentinel-2 images per year during 2019–2022, three phenology metrics were calculated: maximum vegetation index (MAX), vegetation index at peak of season (POS), and area under the curve from the start of the season to the end of the season (AUC) (Figure 3). Previous studies indicated that MAX reflects critical periods of photosynthetic capacity, where higher values often correlate with optimal growing conditions and management practices, while POS aligns with peak crop maturity, a stage where vegetation indices stabilize and are least affected by short-term stressors [40,41]. Unlike metrics like green-up or maturity, which focus on transitions, POS provides a robust, temporally stable signal for yield prediction, particularly for maize and soybean. Meanwhile, AUC represents an extended timeframe by integrating seasonal productivity, accounting for both growth duration and intensity.
The input dataset for phenology analysis included a time-series of dates and corresponding vegetation index values, which consisted of filtered valid observations. Seasonal transitions were identified using a weighted Whittaker moving smoothing algorithm to reduce noise and a weighting function to account for data reliability, as proposed by Jönsson and Eklundh [42]. A curve-fitting approach utilizing the Beck method was used to characterize vegetation growth patterns based on phenological cycles. The fitting process involved two iterations to optimize the model parameters, with constraints on the extension of the growing season to align with the observed phenological phases. Key phenological parameters, including SOS, EOS, and POS, were calculated from the fitted curves. According to these values, vegetation indices at POS were extracted, as well as the maximum value of vegetation indices from all Sentinel-2 images per year. AUC was calculated based on the fitted curve parameters from SOS to EOS, in which an increment in the X axis was set to 1 day.

2.4. Correlation Analysis of Calculated Proxy Metrics According to Ground-Truth Crop Yield Data

The approach used resulted in a total of twelve proxy metrics of cropland suitability based on the combination of four vegetation indices and three phenology metrics individually for maize and soybean. These metrics were evaluated for each year during the 2019–2022 period and two states withing the study area, with a total of 96 repetitions for maize and 96 repetitions for soybean, which were evaluated according to ground-truth crop yield data. The outlier removal of the calculated proxy metrics and ground-truth crop yield data was performed using the interquartile range approach, with all values lower than 1.5 interquartile range below Q1 and higher than 1.5 interquartile range above Q3 considered outliers. The correlation analysis between the evaluated proxy metrics and ground-truth crop yield data was performed based on the linear regression model, using Pearson’s correlation coefficient to quantify their relationship. For each proxy metric, the mean value across two states and four study years was calculated and used to determine the optimal proxy metric for maize and soybean yield according to ground-truth crop yield data. Additionally, Pearson’s correlation coefficient was calculated for all three phenology metrics per study year and state to evaluate the presence of potential multicollinearity among the used phenology metrics.

3. Results and Discussion

3.1. Correlation Analysis Between Evaluated Phenology Metrics

The correlation analysis of the three evaluated phenology metrics for maize and soybean across Iowa and Illinois from 2019 to 2022 indicated a high or moderately high correlation among those variables, primarily between MAX and POS (Figure 4). Pearson’s correlation coefficients between them were stable across the study years and states, ranging from 0.93 to 0.95, which indicated a potential presence of multicollinearity. While the AUC did not produce such a high correlation with either MAX or POS, the AUC produced a slightly higher Pearson’s correlation coefficient with POS than MAX, ranging from 0.78 to 0.85 for maize and from 0.62 to 0.79 for soybean. Similarly, MAX produced a higher Pearson’s correlation coefficient with AUC for maize (0.75–0.80) than soybean (0.65–0.73). As MAX and POS typically occur at similar temporal stages, a slight discrepancy between vegetation index values at these stages may be caused by differences in crop phenology among maize and soybean varieties cultivated in the study area, management practices, or sensitivity to environmental stressors.

3.2. Correlation Analysis Between Evaluated Proxy Metrics and Ground-Truth Crop Yield Data

Pearson’s correlation coefficients between the evaluated proxy metrics for cropland suitability with ground-truth crop yield data are presented in Table 3, with all values being significant at a level of 0.05. The WDRVI consistently produced the strongest correlations with maize yield, particularly for the MAX phenology metric, with Pearson’s correlation coefficients ranging from 0.385 to 0.582. This suggests that the WDRVI is a robust indicator of maize yield, likely due to its resistance to saturation caused by high biomass and canopy structure, which is a known issue of many vegetation indices, especially the NDVI [43]. During 2021 and 2022, the WDRVI based on AUC notably outperformed MAX and POS in Iowa, while EVI2 based on MAX produced the highest correlation for 2021 in Iowa. While EVI2 produced only a slightly higher correlation than the WDRVI in a single case, a higher WDRVI based on AUC for two consecutive years in Iowa notably differs from all the other analyzed cases. However, there is no apparent reason for this, as air temperature and precipitation were higher than the long-term average in 2021 [44] and lower than the long-term average in 2022 [45]. The official agricultural statistics of Iowa for 2022 also noted only minor deviations from multi-year averages in terms of planting and harvesting dates [46]. Correlation analysis with soybean yield strongly indicated that EVI2 was the most suitable vegetation index, while both MAX and POS produced relatively high Pearson’s correlation coefficient means, with MAX being superior in five out of eight datasets. AUC produced weaker correlations than the other two phenology metrics for soybean, but unlike for maize, it was consistently inferior across all eight datasets.
Overall, the maximum Pearson’s correlation coefficient means per crop were slightly higher for soybean (0.519) than for maize (0.506), and their value ranges were also higher for soybean (0.420–0.681) compared to maize (0.388–0.600). These results are comparable to the previous research in the study area by Ji et al. [19], which achieved a maximum coefficient of determination (R2) of 0.44 and which was based on a 250 m Moderate Resolution Imaging Spectroradiometer (MODIS) and crop yield statistics on a county level. Another similar study, which utilized Sentinel-2 images with a yield mapping system integrated in a combine harvester across ten parcels, achieved correlation of vegetation indices with a crop yield higher than 0.6 [47]. In terms of comparable studies which could be used to derive observations relevant for determining an optimal proxy for cropland suitability, maximum Pearson’s correlation coefficients in the range of 0.6–0.7 can be expected. Several studies have produced a significantly higher correlation between vegetation indices and crop yield, frequently higher than 0.9, but these were performed on either a micro-scale area [48] or utilized unmanned aerial vehicles (UAVs) [49,50], which does not meet the requirement for cropland suitability studies in terms of macro-scale predictions [51].
The scatterplots and regression models indicate significant positive correlations between the evaluated cropland suitability proxy metrics and crop yield, although their strength varies on a year-to-year basis (Figure 5). While the results from this study highlight the benefit of selecting an optimal combination of vegetation indices and phenology metrics, there is still a notable level of saturation at higher crop yield values. These cases are frequently associated with higher biomass, which produces a well-known issue of saturation effect for the majority of vegetation indices [52]. The relationships between ground-truth crop yield and evaluated proxy metrics for cropland suitability suggest the presence of saturation effects, particularly in the WDRVI, where the relationship flattens at higher vegetation densities, indicating that the index becomes less responsive under dense canopy conditions. In contrast, EVI2 produced greater resistance to saturation, maintaining a stronger linear relationship with yield even at higher vegetation levels, likely due to its design, which minimizes atmospheric and soil background interference [53]. The observed year-to-year variability in correlation strength and regression slopes also suggest the effect of varying environmental factors and localized management practices on the relationship between vegetation indices and yield [54]. This was most pronounced for maize in Iowa, where the slope of the regression line decreases from 2019 to 2022, indicating a potential decline in the responsiveness of yield to vegetation health over time.

3.3. Study Limitations and Future Considerations

While there are alternatives available to the vegetation indices evaluated in this study for use as a proxy metric for cropland suitability, the present literature lacks a comprehensive study which evaluates multiple novel indices with the established indices, such as the NDVI, EVI, and EVI2. To mitigate the issue of a saturation effect in the results from this study, several previous studies focused on developing new vegetation indices with a higher resistance to the saturation effect. Among them, the plant phenology index (PPI) [55], three red-edge bands vegetation index [56], modified triangular vegetation index [57], and inverted difference vegetation index [58] were mostly compared to the NDVI, to which they showed a superior performance in terms of establishing a correlation with crop yield at a high leaf area index. Most notably, a service within Copernicus based on Sentinel-2 images was developed for the PPI calculation and is updated with daily data [59], as the PPI theoretically maintains a strong sensitivity to canopy changes, even at advanced growth stages, improving yield prediction accuracy in high-biomass areas. Also, red-edge-based indices outperform traditional broadband indices like the NDVI by capturing subtle variations in crop health, particularly in mid-to-late growing seasons when saturation typically occurs [56], but their use as proxy metric for crop yield data is still not studied. Moreover, previous studies have utilized multiple linear regression [19] and machine learning [47] for evaluating vegetation indices and phenology metrics for crop yield prediction, which provide a superior statistical foundation for researching their relationship between crop yield. However, cropland suitability studies commonly require the use of a single proxy metric for suitability [60], which justifies the use of a linear regression model for evaluating the correlation between combinations of vegetation indices and phenology metrics with crop yield in this study. Knowledge of the optimal proxy metric derived from satellite images further enables the use of machine learning for cropland suitability prediction, as well as modeling non-linear relationships between an optimal proxy metric and abiotic covariates using interpretable machine learning [61], including climate, soil, and topography data.
Despite the increased availability of crop yield data in recent years, there are still issues with the lack of ground-truth crop yield samples outside the United States on a large scale. This study succeeded in combining multiple large ground-truth crop yield datasets with 30 m spatial resolution satellite imagery, thus improving the limitations of previous studies in terms of spatial resolution and study coverage, but there is a potential bias in used crop yield samples. While the used QDANN data are the state of the art in crop yield modeling, their root mean square error (RMSE) according to the sampled field data was 2.29 t ha−1 for maize and 0.85 t ha−1 for soybean [31]. This approach provided immensely larger coverage than individual yield mapping systems in combine harvesters and included in-field variability, which is not present in frequently used county-level crop yield statistics. However, yield mapping systems using combine harvesters are presently the most accurate and reliable approach for obtaining ground-truth data for similar studies [62], which cannot be matched with statistical modeling. Also, its reliance on statistical modeling introduces uncertainties, as yield estimates may be influenced by regional biases, input data quality, or unaccounted agronomic factors, such as pest outbreaks or localized management practices. With the improved availability of ground-truth data, future studies should simultaneously evaluate multiple novel vegetation indices to explore their potential to serve as proxy metrics for cropland suitability. Additionally, these vegetation indices could be combined with other phenology metrics than MAX and POS, such as greenup and maturity, both of which proved important for crop monitoring in previous studies [63,64]. While this study focuses on only two specific regions in the United States, the methodology used in this study does not limit the applicability of the model to other geographic regions with different soil types, climate conditions, and crop management practices, as the only requirement is the presence of ground-truth yield data.

4. Conclusions

This study aimed to determine the optimal proxy metric for cropland suitability based on a combination of an extensive ground-truth crop yield dataset (1000 samples per four study years and two study areas) with multitemporal Sentinel-2 satellite images, which were resampled to 30 m spatial resolution to match crop yield data. The main conclusions based on the results of the correlation analysis based on four frequently used vegetation indices and three phenology metrics with maize and soybean ground-truth yield data are as follows:
  • WDRVI was an optimal vegetation index for maize yield prediction, while EVI2 produced the highest correlation for soybean, producing Pearson’s correlation coefficient means of 0.506 and 0.519 in combination with MAX, respectively.
  • The majority of proxy metrics for cropland suitability with the highest correlation per dataset were achieved based on the MAX phenology metric. AUC outperformed it in two out of eight datasets for maize, while POS outperformed it in three cases for soybean yield. The correlations between MAX and POS were very high overall, which reflected only minor differences between their correlations with crop yield and indicated potential multicollinearity. MAX represented a global maximum of vegetation index values per sample, which required a fast and straightforward calculation process, unlike POS.
  • This study combined multiple large ground-truth crop yield datasets with 30 m spatial resolution satellite imagery, but there is a potential bias in the used crop yield samples. The used ground-truth crop yield data were based on predictive modeling, providing larger coverage than individual yield mapping systems in combine harvesters, but they had limited accuracy.
  • The saturation effect at higher crop yield values was observed during correlation analysis between the evaluated proxy metrics for cropland suitability and ground-truth crop yield data. Since several novel vegetation indices were developed to overcome this issue but were not simultaneously evaluated in combination with phenology metrics, future studies should explore their correlation with ground-truth crop yield data.

Author Contributions

Conceptualization, D.R.; methodology, D.R.; software, D.R.; validation, D.R. and M.J.; formal analysis, D.R. and M.J.; investigation, D.R.; resources, D.R.; data curation, D.R.; writing—original draft preparation, D.R.; writing—review and editing, D.R.; visualization, D.R.; supervision, M.J.; project administration, D.R.; funding acquisition, D.R. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Acknowledgments

This research was supported by the project “Soybean cropland suitability prediction based on machine learning regression” from the research team “Technical and technological systems in agriculture, GIT, precision agriculture and environment protection” of the Faculty of Agrobiotechnical Sciences Osijek.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Gu, D.; Andreev, K.; Dupre, M.E. Major Trends in Population Growth Around the World. China CDC Wkly. 2021, 3, 604–613. [Google Scholar] [CrossRef] [PubMed]
  2. Aboye, B.H.; Gebre-Egziabher, T.; Kebede, B.; Minale, A.S. Changing Farming Practices as Integral to Sustenance and Cropland-Use Loss in the Context of Urban Expansion: The Case of Jimma City, Southwest Ethiopia. Clean. Environ. Syst. 2023, 9, 100117. [Google Scholar] [CrossRef]
  3. Akpoti, K.; Kabo-bah, A.T.; Zwart, S.J. Agricultural Land Suitability Analysis: State-of-the-Art and Outlooks for Integration of Climate Change Analysis. Agric. Syst. 2019, 173, 172–208. [Google Scholar] [CrossRef]
  4. Radočaj, D.; Jurišić, M. GIS-Based Cropland Suitability Prediction Using Machine Learning: A Novel Approach to Sustainable Agricultural Production. Agronomy 2022, 12, 2210. [Google Scholar] [CrossRef]
  5. Ozsahin, E.; Ozdes, M. Agricultural Land Suitability Assessment for Agricultural Productivity Based on GIS Modeling and Multi-Criteria Decision Analysis: The Case of Tekirdağ Province. Environ. Monit. Assess. 2021, 194, 41. [Google Scholar] [CrossRef]
  6. Cravero, A.; Pardo, S.; Sepúlveda, S.; Muñoz, L. Challenges to Use Machine Learning in Agricultural Big Data: A Systematic Literature Review. Agronomy 2022, 12, 748. [Google Scholar] [CrossRef]
  7. Zeng, T.; Wu, L.; Peduto, D.; Glade, T.; Hayakawa, Y.S.; Yin, K. Ensemble Learning Framework for Landslide Susceptibility Mapping: Different Basic Classifier and Ensemble Strategy. Geosci. Front. 2023, 14, 101645. [Google Scholar] [CrossRef]
  8. Omia, E.; Bae, H.; Park, E.; Kim, M.S.; Baek, I.; Kabenge, I.; Cho, B.-K. Remote Sensing in Field Crop Monitoring: A Comprehensive Review of Sensor Systems, Data Analyses and Recent Advances. Remote Sens. 2023, 15, 354. [Google Scholar] [CrossRef]
  9. Zeng, Y.; Hao, D.; Huete, A.; Dechant, B.; Berry, J.; Chen, J.M.; Joiner, J.; Frankenberg, C.; Bond-Lamberty, B.; Ryu, Y.; et al. Optical Vegetation Indices for Monitoring Terrestrial Ecosystems Globally. Nat. Rev. Earth Environ. 2022, 3, 477–493. [Google Scholar] [CrossRef]
  10. Vidican, R.; Mălinaș, A.; Ranta, O.; Moldovan, C.; Marian, O.; Ghețe, A.; Ghișe, C.R.; Popovici, F.; Cătunescu, G.M. Using Remote Sensing Vegetation Indices for the Discrimination and Monitoring of Agricultural Crops: A Critical Review. Agronomy 2023, 13, 3040. [Google Scholar] [CrossRef]
  11. Radočaj, D.; Šiljeg, A.; Marinović, R.; Jurišić, M. State of Major Vegetation Indices in Precision Agriculture Studies Indexed in Web of Science: A Review. Agriculture 2023, 13, 707. [Google Scholar] [CrossRef]
  12. Rehman, T.H.; Lundy, M.E.; Linquist, B.A. Comparative Sensitivity of Vegetation Indices Measured via Proximal and Aerial Sensors for Assessing N Status and Predicting Grain Yield in Rice Cropping Systems. Remote Sens. 2022, 14, 2770. [Google Scholar] [CrossRef]
  13. Gong, Z.; Ge, W.; Guo, J.; Liu, J. Satellite Remote Sensing of Vegetation Phenology: Progress, Challenges, and Opportunities. ISPRS J. Photogramm. Remote Sens. 2024, 217, 149–164. [Google Scholar] [CrossRef]
  14. Garcia-Perez, M.A.; Rodriguez-Galiano, V.; Sanchez-Rodriguez, E.; Egea-Cobrero, V. Yield Estimation of Wheat Using Cropland Masks from European Common Agrarian Policy: Comparing the Performance of EVI2, NDVI, and MTCI in Spanish NUTS-2 Regions. Remote Sens. 2023, 15, 5423. [Google Scholar] [CrossRef]
  15. Ji, Z.; Pan, Y.; Zhu, X.; Zhang, D.; Wang, J. A Generalized Model to Predict Large-Scale Crop Yields Integrating Satellite-Based Vegetation Index Time Series and Phenology Metrics. Ecol. Indic. 2022, 137, 108759. [Google Scholar] [CrossRef]
  16. Nolasco, M.; Ovando, G.; Sayago, S.; Magario, I.; Bocco, M. Estimating Soybean Yield Using Time Series of Anomalies in Vegetation Indices from MODIS. Int. J. Remote Sens. 2021, 42, 405–421. [Google Scholar] [CrossRef]
  17. Prasad, N.R.; Patel, N.R.; Danodia, A. Cotton Yield Estimation Using Phenological Metrics Derived from Long-Term MODIS Data. J. Indian Soc. Remote Sens. 2021, 49, 2597–2610. [Google Scholar] [CrossRef]
  18. Bognár, P.; Kern, A.; Pásztor, S.; Steinbach, P.; Lichtenberger, J. Testing the Robust Yield Estimation Method for Winter Wheat, Corn, Rapeseed, and Sunflower with Different Vegetation Indices and Meteorological Data. Remote Sens. 2022, 14, 2860. [Google Scholar] [CrossRef]
  19. Ji, Z.; Pan, Y.; Zhu, X.; Wang, J.; Li, Q. Prediction of Crop Yield Using Phenological Information Extracted from Remote Sensing Vegetation Index. Sensors 2021, 21, 1406. [Google Scholar] [CrossRef]
  20. Al-Shammari, D.; Whelan, B.M.; Wang, C.; Bramley, R.G.V.; Bishop, T.F.A. Assessment of Red-Edge Based Vegetation Indices for Crop Yield Prediction at the Field Scale across Large Regions in Australia. Eur. J. Agron. 2025, 164, 127479. [Google Scholar] [CrossRef]
  21. Amaral, L.R.; Oldoni, H.; Baptista, G.M.M.; Ferreira, G.H.S.; Freitas, R.G.; Martins, C.L.; Cunha, I.A.; Santos, A.F. Remote Sensing Imagery to Predict Soybean Yield: A Case Study of Vegetation Indices Contribution. Precis. Agric. 2024, 25, 2375–2393. [Google Scholar] [CrossRef]
  22. Mezera, J.; Lukas, V.; Horniacek, I.; Smutny, V.; Elbl, J. Comparison of Proximal and Remote Sensing for the Diagnosis of Crop Status in Site-Specific Crop Management. Sensors 2022, 22, 19. [Google Scholar] [CrossRef] [PubMed]
  23. Tunca, E.; Koksal, E.S.; Taner, S.C. Silage Maize Yield Estimation by Using Planetscope, Sentinel-2A and Landsat 8 OLI Satellite Images. Smart Agric. Technol. 2023, 4, 100165. [Google Scholar] [CrossRef]
  24. Bellini, E.; Moriondo, M.; Dibari, C.; Leolini, L.; Staglianò, N.; Stendardi, L.; Filippa, G.; Galvagno, M.; Argenti, G. Impacts of Climate Change on European Grassland Phenology: A 20-Year Analysis of MODIS Satellite Data. Remote Sens. 2023, 15, 218. [Google Scholar] [CrossRef]
  25. Yang, S.; Wang, H.; Tong, J.; Bai, Y.; Alatalo, J.M.; Liu, G.; Fang, Z.; Zhang, F. Impacts of Environment and Human Activity on Grid-Scale Land Cropping Suitability and Optimization of Planting Structure, Measured Based on the MaxEnt Model. Sci. Total Environ. 2022, 836, 155356. [Google Scholar] [CrossRef]
  26. Zhang, L.; Zhang, K.; Zhu, X.; Chen, H.; Wang, W. Integrating Remote Sensing, Irrigation Suitability and Statistical Data for Irrigated Cropland Mapping over Mainland China. J. Hydrol. 2022, 613, 128413. [Google Scholar] [CrossRef]
  27. Manafifard, M.; Huang, J. A Comprehensive Review on Wheat Yield Prediction Based on Remote Sensing. Multimed. Tools Appl. 2024, 83, 1–74. [Google Scholar] [CrossRef]
  28. Che, Y.; Gu, Y.; Bai, D.; Li, D.; Li, J.; Zhao, C.; Wang, Q.; Qiu, H.; Huang, W.; Yang, C.; et al. Accurately Estimate Soybean Growth Stages from UAV Imagery by Accounting for Spatial Heterogeneity and Climate Factors across Multiple Environments. Comput. Electron. Agric. 2024, 225, 109313. [Google Scholar] [CrossRef]
  29. Xu, H.; Twine, T.E.; Girvetz, E. Climate Change and Maize Yield in Iowa. PLoS ONE 2016, 11, e0156083. [Google Scholar] [CrossRef]
  30. Zhang, J.; Lu, C.; Feng, H.; Hennessy, D.; Guan, Y.; Wright, M.M. Extreme Climate Increased Crop Nitrogen Surplus in the United States. Agric. For. Meteorol. 2021, 310, 108632. [Google Scholar] [CrossRef]
  31. Ma, Y.; Liang, S.-Z.; Myers, D.B.; Swatantran, A.; Lobell, D.B. Subfield-Level Crop Yield Mapping Without Ground Truth Data: A Scale Transfer Framework. Remote Sens. Environ. 2024, 315, 114427. [Google Scholar] [CrossRef]
  32. Awesome-Gee-Community-Catalog. QDANN 30 m Yield Map. Available online: https://gee-community-catalog.org/projects/qdann/ (accessed on 21 February 2025).
  33. Earth Engine Data Catalog. Harmonized Sentinel-2 MSI: MultiSpectral Instrument, Level-1C. Available online: https://developers.google.com/earth-engine/datasets/catalog/COPERNICUS_S2_HARMONIZED (accessed on 22 February 2025).
  34. Wu, H.; Liu, B.; Zhu, B.; Zhen, Z.; Song, K.; Ren, J. Combining Vegetation Indices to Identify the Maize Phenological Information Based on the Shape Model. Agriculture 2024, 14, 1608. [Google Scholar] [CrossRef]
  35. Swain, S.; Wardlow, B.D.; Narumalani, S.; Rundquist, D.C.; Hayes, M.J. Relationships Between Vegetation Indices and Root Zone Soil Moisture under Maize and Soybean Canopies in the US Corn Belt: A Comparative Study Using a Close-Range Sensing Approach. Int. J. Remote Sens. 2013, 34, 2814–2828. [Google Scholar] [CrossRef]
  36. Gitelson, A.A. Wide Dynamic Range Vegetation Index for Remote Quantification of Biophysical Characteristics of Vegetation. J. Plant Physiol. 2004, 161, 165–173. [Google Scholar] [CrossRef]
  37. Rouse, J.; Haas, R.H.; Schell, J.A.; Deering, D.W. Monitoring Vegetation Systems in the Great Plains with ERTS. NASA Spec. Publ. 1974, 351, 309. [Google Scholar]
  38. Huete, A.; Didan, K.; Miura, T.; Rodriguez, E.P.; Gao, X.; Ferreira, L.G. Overview of the Radiometric and Biophysical Performance of the MODIS Vegetation Indices. Remote Sens. Environ. 2002, 83, 195–213. [Google Scholar] [CrossRef]
  39. Jiang, Z.; Huete, A.R.; Didan, K.; Miura, T. Development of a Two-Band Enhanced Vegetation Index without a Blue Band. Remote Sens. Environ. 2008, 112, 3833–3845. [Google Scholar] [CrossRef]
  40. Bajocco, S.; Di Giulio, M.; Sallah, A.H.M.; Bregaglio, S. Exploring the Relationships Between Ground Observations and Remotely Sensed Hazelnut Spring Phenology. Int. J. Biometeorol. 2025, 69, 281–294. [Google Scholar] [CrossRef]
  41. Park, T.; Chen, C.; Macias-Fauria, M.; Tømmervik, H.; Choi, S.; Winkler, A.; Bhatt, U.S.; Walker, D.A.; Piao, S.; Brovkin, V.; et al. Changes in Timing of Seasonal Peak Photosynthetic Activity in Northern Ecosystems. Glob. Change Biol. 2019, 25, 2382–2395. [Google Scholar] [CrossRef]
  42. Jönsson, P.; Eklundh, L. TIMESAT—A Program for Analyzing Time-Series of Satellite Sensor Data. Comput. Geosci. 2004, 30, 833–845. [Google Scholar] [CrossRef]
  43. Zhen, Z.; Chen, S.; Yin, T.; Chavanon, E.; Lauret, N.; Guilleux, J.; Henke, M.; Qin, W.; Cao, L.; Li, J.; et al. Using the Negative Soil Adjustment Factor of Soil Adjusted Vegetation Index (SAVI) to Resist Saturation Effects and Estimate Leaf Area Index (LAI) in Dense Vegetation Areas. Sensors 2021, 21, 2115. [Google Scholar] [CrossRef] [PubMed]
  44. Iowa Annual Weather Summary. 2021. Available online: https://iowaagriculture.gov/sites/default/files/climatology/2021/Iowa-Annual-Weather-Summary-2021.pdf (accessed on 24 February 2025).
  45. Iowa Annual Weather Summary. 2022. Available online: https://iowaagriculture.gov/sites/default/files/climatology/2022/Iowa-Annual-Weather-Summary-2022.pdf (accessed on 24 February 2025).
  46. Iowa Agricultural Statistics. 2022. Available online: https://www.nass.usda.gov/Statistics_by_State/Iowa/Publications/Annual_Statistical_Bulletin/2022-Iowa-Annual-Bulletin.pdf (accessed on 24 February 2025).
  47. Amankulova, K.; Farmonov, N.; Mukhtorov, U.; Mucsi, L. Sunflower Crop Yield Prediction by Advanced Statistical Modeling Using Satellite-Derived Vegetation Indices and Crop Phenology. Geocarto Int. 2023, 38, 2197509. [Google Scholar] [CrossRef]
  48. Ali, A.; Martelli, R.; Lupia, F.; Barbanti, L. Assessing Multiple Years’ Spatial Variability of Crop Yields Using Satellite Vegetation Indices. Remote Sens. 2019, 11, 2384. [Google Scholar] [CrossRef]
  49. Parida, P.K.; Somasundaram, E.; Krishnan, R.; Radhamani, S.; Sivakumar, U.; Parameswari, E.; Raja, R.; Shri Rangasami, S.R.; Sangeetha, S.P.; Gangai Selvi, R. Unmanned Aerial Vehicle-Measured Multispectral Vegetation Indices for Predicting LAI, SPAD Chlorophyll, and Yield of Maize. Agriculture 2024, 14, 1110. [Google Scholar] [CrossRef]
  50. Bao, L.; Li, X.; Yu, J.; Li, G.; Chang, X.; Yu, L.; Li, Y. Forecasting Spring Maize Yield Using Vegetation Indices and Crop Phenology Metrics from UAV Observations. Food Energy Secur. 2024, 13, e505. [Google Scholar] [CrossRef]
  51. Radočaj, D.; Jurišić, M.; Gašparović, M.; Plaščak, I.; Antonić, O. Cropland Suitability Assessment Using Satellite-Based Biophysical Vegetation Properties and Machine Learning. Agronomy 2021, 11, 1620. [Google Scholar] [CrossRef]
  52. Rinaldi, M.; Ruggieri, S.; Garofalo, P.; Vonella, A.V.; Satalino, G.; Soldo, P. Leaf Area Index Retrieval Using High Resolution Remote Sensing Data. Ital. J. Agron. 2010, 5, 155–166. [Google Scholar] [CrossRef]
  53. Zhen, Z.; Chen, S.; Yin, T.; Gastellu-Etchegorry, J.-P. Globally Quantitative Analysis of the Impact of Atmosphere and Spectral Response Function on 2-Band Enhanced Vegetation Index (EVI2) over Sentinel-2 and Landsat-8. ISPRS J. Photogramm. Remote Sens. 2023, 205, 206–226. [Google Scholar] [CrossRef]
  54. Vélez, S.; Martínez-Peña, R.; Castrillo, D. Beyond Vegetation: A Review Unveiling Additional Insights into Agriculture and Forestry Through the Application of Vegetation Indices. J 2023, 6, 421–436. [Google Scholar] [CrossRef]
  55. Jin, H.; Eklundh, L. A Physically Based Vegetation Index for Improved Monitoring of Plant Phenology. Remote Sens. Environ. 2014, 152, 512–525. [Google Scholar] [CrossRef]
  56. Qiao, K.; Zhu, W.; Xie, Z.; Wu, S.; Li, S. New Three Red-Edge Vegetation Index (VI3RE) for Crop Seasonal LAI Prediction Using Sentinel-2 Data. Int. J. Appl. Earth Obs. Geoinf. 2024, 130, 103894. [Google Scholar] [CrossRef]
  57. Yao, X.; Wang, N.; Liu, Y.; Cheng, T.; Tian, Y.; Chen, Q.; Zhu, Y. Estimation of Wheat LAI at Middle to High Levels Using Unmanned Aerial Vehicle Narrowband Multispectral Imagery. Remote Sens. 2017, 9, 1304. [Google Scholar] [CrossRef]
  58. Sun, Y.; Ren, H.; Zhang, T.; Zhang, C.; Qin, Q. Crop Leaf Area Index Retrieval Based on Inverted Difference Vegetation Index and NDVI. IEEE Geosci. Remote Sens. Lett. 2018, 15, 1662–1666. [Google Scholar] [CrossRef]
  59. Europe, Daily. Plant Phenology Index 2016-Present (Raster 10 m). Available online: https://land.copernicus.eu/api/en/products/vegetation/high-resolution-plant-phenology-index (accessed on 27 February 2025).
  60. Radočaj, D.; Gašparović, M.; Jurišić, M. Cropland Suitability Prediction Method Based on Biophysical Variables from Copernicus Data and Machine Learning. Appl. Sci. 2025, 15, 372. [Google Scholar] [CrossRef]
  61. Rudin, C. Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead. Nat. Mach. Intell. 2019, 1, 206–215. [Google Scholar] [CrossRef]
  62. Perich, G.; Turkoglu, M.O.; Graf, L.V.; Wegner, J.D.; Aasen, H.; Walter, A.; Liebisch, F. Pixel-Based Yield Mapping and Prediction from Sentinel-2 Using Spectral Indices and Neural Networks. Field Crops Res. 2023, 292, 108824. [Google Scholar] [CrossRef]
  63. Shen, Y.; Zhang, X.; Yang, Z.; Ye, Y.; Wang, J.; Gao, S.; Liu, Y.; Wang, W.; Tran, K.H.; Ju, J. Developing an Operational Algorithm for Near-Real-Time Monitoring of Crop Progress at Field Scales by Fusing Harmonized Landsat and Sentinel-2 Time Series with Geostationary Satellite Observations. Remote Sens. Environ. 2023, 296, 113729. [Google Scholar] [CrossRef]
  64. Diao, C.; Li, G. Near-Surface and High-Resolution Satellite Time Series for Detecting Crop Phenology. Remote Sens. 2022, 14, 1957. [Google Scholar] [CrossRef]
Agriculture 15 00859 g001
Figure 1. Study area with random yield samples for maize and soybean during the 2019–2022 period.
Figure 1. Study area with random yield samples for maize and soybean during the 2019–2022 period.
Agriculture 15 00859 g001
Agriculture 15 00859 g002
Figure 2. A cumulative number of studies indexed in the Web of Science Core Collection with the topic of “yield prediction” AND “satellite images” AND vegetation index name since 2010 for normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), normalized difference water index (NDWI), soil-adjusted vegetation index (SAVI), and wide dynamic range vegetation index (WDRVI).
Figure 2. A cumulative number of studies indexed in the Web of Science Core Collection with the topic of “yield prediction” AND “satellite images” AND vegetation index name since 2010 for normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), normalized difference water index (NDWI), soil-adjusted vegetation index (SAVI), and wide dynamic range vegetation index (WDRVI).
Agriculture 15 00859 g002
Agriculture 15 00859 g003
Figure 3. A display of three phenology metrics used for evaluation of correlation between vegetation indices and ground-truth crop yield data.
Figure 3. A display of three phenology metrics used for evaluation of correlation between vegetation indices and ground-truth crop yield data.
Agriculture 15 00859 g003
Agriculture 15 00859 g004
Figure 4. The correlation plots based on Pearson’s correlation coefficients between used phenology metrics per study year and state. MAX designates the maximum vegetation index value, POS indicates the vegetation index value at the peak of the season, while AUC designates the area under the curve between the start and end of the season.
Figure 4. The correlation plots based on Pearson’s correlation coefficients between used phenology metrics per study year and state. MAX designates the maximum vegetation index value, POS indicates the vegetation index value at the peak of the season, while AUC designates the area under the curve between the start and end of the season.
Agriculture 15 00859 g004
Agriculture 15 00859 g005
Figure 5. Scatterplots between used phenology metrics per study year and state with fitted linear regression model.
Figure 5. Scatterplots between used phenology metrics per study year and state with fitted linear regression model.
Agriculture 15 00859 g005
Table 1. Mean yield and coefficient of variation (CV) per input ground-truth dataset.
Table 1. Mean yield and coefficient of variation (CV) per input ground-truth dataset.
CropYearIowaIllinois
Mean (kg ha−1)CVMean (kg ha−1)CV
Maize201912,882.60.12612,310.50.106
202012,561.00.12213,049.20.115
202113,535.80.10713,206.40.109
202213,143.80.09913,854.40.087
Soybean20193446.40.1023471.60.116
20203662.10.1183909.80.107
20213959.80.0944024.70.105
20223731.30.0773910.40.089
Table 2. Vegetation indices used in the study based on Sentinel-2 images.
Table 2. Vegetation indices used in the study based on Sentinel-2 images.
Vegetation IndexAbbreviationFormulaReference
Normalized Difference Vegetation IndexNDVI NDVI = NIR − R NIR + R [37]
Enhanced Vegetation IndexEVI EVI = 2.5 × NIR − R NIR + 6 R − 7.5 B + 1 [38]
Enhanced Vegetation Index 2EVI2 EVI 2 = 2.4 × NIR − R NIR + R + 1 [39]
Wide Dynamic Range Vegetation IndexWDRVI WDRVI = 0.2 NIR − R 0.2 NIR + R [36]
NIR: bottom of atmosphere surface reflectance in near-infrared band, R: bottom of atmosphere surface reflectance in red band, B: bottom of atmosphere surface reflectance in blue band.
Table 3. Pearson’s correlation coefficients between evaluated proxy metrics for cropland suitability with ground-truth crop yield data.
Table 3. Pearson’s correlation coefficients between evaluated proxy metrics for cropland suitability with ground-truth crop yield data.
CropVegetation IndexPhenology MetricIowaIllinoisMean
20192020202120222019202020212022
MaizeNDVIMAX0.5330.4870.4090.4500.5450.3950.3710.4020.449
POS0.4890.4640.4340.4340.4770.3670.3400.3680.422
AUC0.1710.1960.2540.3550.2800.1340.0500.3390.222
EVIMAX0.0370.0000.0030.1130.1320.0270.0270.0480.048
POS0.4900.4650.3900.4820.5450.3420.3630.3870.433
AUC0.2980.3600.4160.4740.4160.2060.2280.4280.353
EVI2MAX0.5700.4710.3820.4930.5810.3420.3880.3960.453
POS0.5570.5100.4210.5030.5750.3600.3670.4270.465
AUC0.3000.3580.3580.4890.4010.2130.2040.4260.344
WDRVIMAX0.5780.5210.4100.5580.5820.4840.3850.5310.506
POS0.5610.4620.4370.5100.5580.4560.3520.5200.482
AUC0.4000.4940.5040.6000.4650.3220.3110.4930.449
SoybeanNDVIMAX0.3720.6340.4860.3610.5130.4660.3850.4570.459
POS0.3300.6200.4210.3190.4520.4390.3070.4570.418
AUC0.1270.1980.1400.0970.1550.0400.0110.1930.120
EVIMAX0.0050.1230.1430.0750.1430.0170.0560.3330.112
POS0.4400.6430.5280.3660.5330.4840.4180.5330.493
AUC0.2750.4200.4130.2520.2430.2140.2220.4510.311
EVI2MAX0.4720.6640.5520.3870.5710.5320.4200.5570.519
POS0.4430.6810.5410.4210.5430.5350.4000.5520.515
AUC0.2890.4020.4150.2560.2530.2240.1820.4170.305
WDRVIMAX0.4360.6380.4730.3650.4980.5050.3570.4590.466
POS0.4010.6340.4520.3750.4810.4760.3340.4280.448
AUC0.3400.4030.3360.2750.3270.2740.1540.4220.316
The highest Pearson’s correlation coefficients per study year and state are bolded.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Radočaj, D.; Jurišić, M. A Phenology-Based Evaluation of the Optimal Proxy for Cropland Suitability Based on Crop Yield Correlations from Sentinel-2 Image Time-Series. Agriculture 2025, 15, 859. https://doi.org/10.3390/agriculture15080859

AMA Style

Radočaj D, Jurišić M. A Phenology-Based Evaluation of the Optimal Proxy for Cropland Suitability Based on Crop Yield Correlations from Sentinel-2 Image Time-Series. Agriculture. 2025; 15(8):859. https://doi.org/10.3390/agriculture15080859

Chicago/Turabian Style

Radočaj, Dorijan, and Mladen Jurišić. 2025. "A Phenology-Based Evaluation of the Optimal Proxy for Cropland Suitability Based on Crop Yield Correlations from Sentinel-2 Image Time-Series" Agriculture 15, no. 8: 859. https://doi.org/10.3390/agriculture15080859

APA Style

Radočaj, D., & Jurišić, M. (2025). A Phenology-Based Evaluation of the Optimal Proxy for Cropland Suitability Based on Crop Yield Correlations from Sentinel-2 Image Time-Series. Agriculture, 15(8), 859. https://doi.org/10.3390/agriculture15080859

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop