Europe PMC
Nothing Special   »   [go: up one dir, main page]

Europe PMC requires Javascript to function effectively.

Either your web browser doesn't support Javascript or it is currently turned off. In the latter case, please turn on Javascript support in your web browser and reload this page.

This website requires cookies, and the limited processing of your personal data in order to function. By using the site you are agreeing to this as outlined in our privacy notice and cookie policy.

Precision Estimation of Crop Coefficient for Maize Cultivation Using High-Resolution Satellite Imagery to Enhance Evapotranspiration Assessment in Agriculture.

Author information

Affiliations

  • 1. Faculty of Agricultural and Food Sciences and Environmental Management, Institute of Water and Environmental Management, University of Debrecen, H-4032 Debrecen, Hungary.
      Authors
    • Nagy A 1
    • Kiss NÉ 1
    • Buday-Bódi E 1
    • Magyar T 1
    • Tamás J 1
    • Fehér ZZ 1
    (6 authors)
  • 2. Consorzio di Bonifica Canale Emiliano Romagnolo, Via E. Masi 8, 40137 Bologna, Italy.
      Authors
    • Cavazza F 2
    • Gentile SL 2
    (2 authors)
  • 3. Faculty of Geo-Information Science and Earth Observation (ITC), University of Twente, Drienerlolaan 5, P.O. Box 217, 7500 AE Enschede, The Netherlands.
      Authors
    • Abdullah H 3
    (1 author)

ORCIDs linked to this article

Show all (6)

Plants (Basel, Switzerland), 27 Apr 2024, 13(9):1212
https://doi.org/10.3390/plants13091212 PMID: 38732427 PMCID: PMC11085199

This article is in the Europe PMC Open access subset. Refer to the copyright information in the article for licensing details.
Free full text in Europe PMC

Abstract 

The estimation of crop evapotranspiration (ETc) is crucial for irrigation water management, especially in arid regions. This can be particularly relevant in the Po Valley (Italy), where arable lands suffer from drought damages on an annual basis, causing drastic crop yield losses. This study presents a novel approach for vegetation-based estimation of crop evapotranspiration (ETc) for maize. Three years of high-resolution multispectral satellite (Sentinel-2)-based Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Normalized Difference Red Edge Index (NDRE), and Leaf Area Index (LAI) time series data were used to derive crop coefficients of maize in nine plots at the Acqua Campus experimental farm of Irrigation Consortium for the Emilia Romagna Canal (CER), Italy. Since certain vegetation indices (VIs) (such as NDVI) have an exponential nature compared to the other indices, both linear and power regression models were evaluated to estimate the crop coefficient (Kc). In the context of linear regression, the correlations between Food and Agriculture Organization (FAO)-based Kc and NDWI, NDRE, NDVI, and LAI-based Kc were 0.833, 0.870, 0.886, and 0.771, respectively. Strong correlation values in the case of power regression (NDWI: 0.876, NDRE: 0.872, NDVI: 0.888, LAI: 0.746) indicated an alternative approach to provide crop coefficients for the vegetation period. The VI-based ETc values were calculated using reference evapotranspiration (ET0) and VI-based Kc. The weather station data of CER were used to calculate ET0 based on Penman-Monteith estimation. Out of the Vis, NDWI and NDVI-based ETc performed the best both in the cases of linear (NDWI RMSE: 0.43 ± 0.12; NDVI RMSE: 0.43 ± 0.095) and power (NDWI RMSE: 0.44 ± 0.116; NDVI RMSE: 0.44 ± 0.103) approaches. The findings affirm the efficacy of the developed methodology in accurately assessing the evapotranspiration rate. Consequently, it offers a more refined temporal estimation of water requirements for maize cultivation in the region.

Free full text 

Logo of plantsLink to Publisher's site
Plants (Basel). 2024 May; 13(9): 1212.
Published online 2024 Apr 27. https://doi.org/10.3390/plants13091212
PMCID: PMC11085199
PMID: 38732427

Precision Estimation of Crop Coefficient for Maize Cultivation Using High-Resolution Satellite Imagery to Enhance Evapotranspiration Assessment in Agriculture

Attila Nagy, Conceptualization, Methodology, Writing – review & editing, Funding acquisition,1 Nikolett Éva Kiss, Writing – original draft,1,* Erika Buday-Bódi, Data curation,1 Tamás Magyar, Data curation,1 Francesco Cavazza, Data curation,2 Salvatore Luca Gentile, Data curation,2 Haidi Abdullah, Data curation,3 János Tamás, Writing – review & editing, Funding acquisition,1 and Zsolt Zoltán Fehér, Conceptualization, Methodology1
Fulai Liu, Academic Editor
Author information Article notes Copyright and License information Disclaimer

Associated Data

Data Availability Statement

Abstract

The estimation of crop evapotranspiration (ETc) is crucial for irrigation water management, especially in arid regions. This can be particularly relevant in the Po Valley (Italy), where arable lands suffer from drought damages on an annual basis, causing drastic crop yield losses. This study presents a novel approach for vegetation-based estimation of crop evapotranspiration (ETc) for maize. Three years of high-resolution multispectral satellite (Sentinel-2)-based Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Normalized Difference Red Edge Index (NDRE), and Leaf Area Index (LAI) time series data were used to derive crop coefficients of maize in nine plots at the Acqua Campus experimental farm of Irrigation Consortium for the Emilia Romagna Canal (CER), Italy. Since certain vegetation indices (VIs) (such as NDVI) have an exponential nature compared to the other indices, both linear and power regression models were evaluated to estimate the crop coefficient (Kc). In the context of linear regression, the correlations between Food and Agriculture Organization (FAO)-based Kc and NDWI, NDRE, NDVI, and LAI-based Kc were 0.833, 0.870, 0.886, and 0.771, respectively. Strong correlation values in the case of power regression (NDWI: 0.876, NDRE: 0.872, NDVI: 0.888, LAI: 0.746) indicated an alternative approach to provide crop coefficients for the vegetation period. The VI-based ETc values were calculated using reference evapotranspiration (ET0) and VI-based Kc. The weather station data of CER were used to calculate ET0 based on Penman-Monteith estimation. Out of the Vis, NDWI and NDVI-based ETc performed the best both in the cases of linear (NDWI RMSE: 0.43 ± 0.12; NDVI RMSE: 0.43 ± 0.095) and power (NDWI RMSE: 0.44 ± 0.116; NDVI RMSE: 0.44 ± 0.103) approaches. The findings affirm the efficacy of the developed methodology in accurately assessing the evapotranspiration rate. Consequently, it offers a more refined temporal estimation of water requirements for maize cultivation in the region.

Keywords: Sentinel-2, vegetation index-based Kc, vegetation index-based crop evapotranspiration, maize water demand

1. Introduction

The ongoing trends in climate change are exerting profound implications on various facets of human society, particularly impacting agriculture, and the global food supply [1]. Improving agricultural water use efficiency and crop water productivity are the pillars of attaining higher agricultural production. Consequently, the optimal consumption of limited water resources is becoming of primary importance for the public sector, especially in South Europe [2]. This is particularly true in the Mediterranean region, where multiple stakeholders (agriculture, household, industry) compete for increasingly scarce water resources [3]. This is especially true in agricultural regions (e.g., northern Italy, Emilia Romagna), with water-intensive (maize, tomato) agricultural production. Changes in temperature, rainfall patterns, atmospheric carbon dioxide (CO2), and the frequency and intensity of extreme weather events all have a substantial influence on agricultural yields, leading to rising food insecurity [4]. Proper water management in agriculture is a key factor in agricultural production.

Due to the high spatiotemporal and temporal variability of meteorological parameters and vegetation as well as other environmental factors, such as soil conditions or the depth of the water table, planning optimal distribution of irrigation water is a very complex task [5]. Such highly heterogeneous environments can be surveyed as detailed as possible resulting in a large variety of data. Meanwhile, data-intensive modern artificial intelligence, pattern recognition, and neural networks may contribute to more efficient planning and assessment of irrigation water distribution [6]. On the other hand, the proper estimation of evapotranspiration (ETc) is a key parameter in the estimation of crop water demand for irrigation planning, design, and efficient irrigation water management of irrigation schemes [7,8]. One of the most commonly used methods for ETc estimation is the crop coefficient approach, which can be calculated based on reference evapotranspiration (ET0) and crop coefficient (Kc) [9]. The Kc changes throughout the growing season and reflects the proportion of the ground cover of the vegetation [10]. Several approaches can be used to measure and/or estimate crop evapotranspiration (ETc), such as eddy covariance, soil water balance approach, Bowen ratio energy balance system, pan evaporation, and weighting lysimeters [11]. However, these methods as well as ETc-based soil water balance models are expensive, hard to automate, and only provide point-based measurements. Hence, the application of these models encounters challenges when extended to expansive regions due to the spatial and temporal heterogeneity of input data. This is particularly notable as local conditions, including soil hydraulic parameters, interactions with groundwater, and variations in plant growth, pose intricate considerations that are not easily accounted for [12].

To tackle such challenges, remote sensing (RS) offers low-cost labor-saving, and time-saving approaches for assessing the spatial and temporal patterns of vegetation at different scales [13,14]. Satellite-based global estimates of evapotranspiration are available like MOD16 (MODIS Global Evapotranspiration) with 1 km spatial resolution including both ET0, ETc estimates [15] or ECOsystem Spaceborne Thermal Radiometer Experiment on the Space Station (ECOSTRESS) with 70 m spatial resolution [16]. The combination of different satellite data sources enables state-of-the-art and very sophisticated assessment methods. Traditionally, optical and thermal cameras (i.e., Surface Energy Balance Algorithm for Land (SEBAL) [17], Two Source Energy Balance (TSEB) [18], Surface Energy Balance System (SEBS) [19], Mapping Evapotranspiration at High Resolution with Internalized Calibration (METRIC) [20], and the Simplified Surface Energy Balance (SSEB) [21] and others) that are coupled with synthetic aperture radars can provide a variety of near real-time snapshots of the environment, which can contribute to modern agricultural methods [22,23,24]. Earth observation (EO) approaches offer accurate data for precision agriculture applications and end-users (i.e., farmers, landowners, and decision-makers). High-resolution multispectral images may be used to schedule irrigation depending on crop demands in near real-time [25,26]. Despite substantial research on ET crop estimates for water management using EO data, the poor geographical and temporal resolution of the sensors has hindered their usefulness and technical transfer [27]. However, precision irrigation requires high-resolution estimates of crop evapotranspiration and irrigation water requirements at the field level. There are several other hydrological models which use ETc as an input parameter as well [28]. Using biomass-based daily estimation of Kc over the vegetation period fosters a more reliable estimation of evapotranspiration contributing to the calculation of a more accurate water balance [29,30].

In the crop coefficient approach, Kc is typically taken from the Food and Agricultural Organization, FAO [31], which does not capture the spatial and temporal variability in estimating crop evapotranspiration. However, proper detection of crop growth affects the estimates of evapotranspiration fluxes and irrigation needs. Nowadays, high-resolution satellite datasets (10 m of spatial resolution) such as Sentinel-2 are easy and free to access online, therefore the spatial and temporal variability in the Kc can be easily derived from VIs considering actual biomass, crop growth conditions, and stages [32]. The Sentinel-2 A/B was used in this study, with five days of returning time. Free satellite imagery, high-resolution, innovative spectrum capabilities, a 290 km swath width, and frequent return durations are boosting operational and commercial uses of EO data for precision agriculture applications and research initiatives [33,34]. The recent Sentinel-2 mission from the European Space Agency (ESA), as part of the Copernicus program [35], has significantly enhanced regular monitoring of agricultural parameters, like Normalized Difference Vegetation Index (NDVI), Normalized Difference Red Edge Index (NDRE), Normalized Difference Water Index (NDWI) and Leaf Area Index (LAI) [36,37,38]. NDVI is a widely used VI and could provide important information for crop modeling research [39]. NDVI time series are applied to estimate crop growth under different climatic conditions [40]. The NDVI data has been used to monitor crop conditions in irrigation management [41] and estimate crop yield under different agroecological conditions [42,43,44]. The spatial and temporal variability in Kc can also be easily derived from NDVI because actual crop growth conditions and stages influence the temporal changes in NDVI values [32]. The NDRE uses the reflectance from red edge bands to estimate plant health [45]. The NDRE responds to plant health, biomass, and photosynthetic activity. It is, therefore, appropriate for plant health and productivity monitoring [46,47]. The NDWI is a commonly used VI to determine the difference between water and vegetation cover, using the differences in the visible and near-infrared wavelength ranges [48]. It is used to delineate, measure, and track the size and the spatial extent of change in water bodies [49] as well as to assess the effect of soil moisture or drought stress on plants [50]. The monitoring of LAI distribution and variation is essential for tracking and detecting vegetation vitality and growth [51]. LAI value may also be determined from RS data by empirical relationships between canopy reflectance and VI regardless of time or costs [52].

This research evaluates the use of RS-based indices in ETc calculation for the region of Emilia Romagna, Italy. There is an IT Irrigation Advisory Service for Farm Water Management services (IRRIFRAME) which utilizes a water balance model that aims at supporting crop irrigation management on a field scale and results in seasonal water savings of about 120 million m3 in Italy [53]. The IRRIFRAME provides evapotranspiration data under standard conditions [54], calculating reference evapotranspiration data (ET0) by the Hargreaves formula with a 6.25 km2 geographical grid. This study aimed to develop an advanced RS-based crop coefficient (Kc) and crop evapotranspiration (ETc) estimation method to support the preciseness of the calculation of the potential maize water consumption with 10 m spatial resolution. Sentinel-2-derived remotely sensed indices (i.e., NDVI, NDWI, NDRE, and LAI) were integrated with climatic data to define (i) the lengths of the phenological phases, (ii) biomass-based crop coefficient and (iii) to calculate and evaluate more accurate RS-based crop evapotranspiration. Due to the 5-day revisit time of Sentinel-2 satellites, the results contribute to the timely calculation of crop evapotranspiration. Results could also contribute to further improvement in spatial resolution (10 m) of the water balance calculations at maize fields.

2. Materials and Methods

2.1. Study Site

Field trials were carried out by the Irrigation Consortium for the Emilia Romagna Canal (CER) at the experimental farm Acqua Campus, covering 12.5 hectares of area that can be divided into approximately 25 fields. The fields are located in the plain of the Po River, in the province of Bologna, near the village of Mezzolara di Budrio (44°34′ N, 11°32′ E). Various crops are cultivated on the farm (both perennial and annual). Crop rotation in the experimental fields is characterized by annual crops, such as winter wheat, soya, maize, onion, and processing tomato.

The farm’s soil is typical of the plain of the Po River and has a high content of silt, clay, and fine sand. The soil in the area can be described as clay/clayey loam. It belongs to the Italian soil group “Suoli SECCHIA franco argillosi”, which can be classified as “Oxyaquic Haplustepts” (fine loamy, mixed, superactive, and mesic according to Soil Taxonomy). These soils are calcareous and moderately alkaline; they have a texture of clay loam in the superficial layers and loam deeper in the profile. Soil layers are affected by multiple floods of the nearby Idice River. The soil within the farm is heterogeneous, and its hydraulic characteristics vary from field to field and layer to layer, with a slight gradient from east to west. The mean values for the soil parameters are specified in Table 1. An extended, shallow groundwater table usually is present at a depth ranging from −0.6 to −1.8 m. During winter and at the beginning of the growing season, the capillary rise could be significant in terms of replenishing the evapotranspiration of crops.

Table 1

Summary of mean values for the soil parameters.

ParameterUnitMean
Sand%32
Silt%50
Clay%18
pH-8.27
CaCO3 total%13.5
CaCO3 active%3.1
N total%0.06
K exchangeablemeq/100 g0.34
P (Olsen)meq/100 g5.49
CECmeq/100 g21.6

For the experimental farm, the water is supplied from the CER canal system, and its quality is regularly checked during the irrigation season [55]. CER’s irrigation water characteristics (long-term mean) are freely available online [56].

The farm has a fully operational weather station equipped with a rain gauge, anemometer, phreatimeter, and pan evaporimeter.

The climate of the site can be defined as sub-humid, with a mean annual temperature of 13.7 °C and a mean annual rainfall of 771 mm. The year 2018 of the field trials was characterized by severe drought. According to the estimates based on CER analyses, 2018 was the second harshest year in the study area, with a cumulative precipitation mostly below 200 mm in all areas which represents a deviation of 50% from the reference climate (period 1961–2018). These conditions contributed significantly to maize production and yield values. In 2019 and 2020, the temperature was above the regional average with occasional heatwaves; however, the precipitation exceeded 200 mm in the vegetation period of maize. For this study, three experimental fields were cultivated with maize on the Acqua Campus farm as shown in Figure 1. According to the CER data, the crop yield varied between 16.5 and 16.7 t/ha in 2018, 13.8 and 16.2 t/ha in 2019, and 18.8 and 19.9 t/ha in 2020, respectively.

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g001.jpg

Experimental fields on the Acqua Campus farm.

2.2. Study Framework and Data

In this study, data from three maize fields in three years (2018, 2019, 2020) were used to set an advanced RS-based crop evapotranspiration estimation method to support a more accurate calculation of the water consumption of maize. The daily meteorological data (minimum and maximum air temperature, relative humidity, radiation, and wind speed) were used to estimate ET0 using the Penman-Monteith method. The crop coefficient was defined by developing linear and power regression equations between the NDVI, NDWI, NDRE LAI, and FAO Kc for the maize crop. Then VI-based Kc was used to calculate ETc, and the performances of the index-based crop evapotranspiration were compared to ETc derived by FAO Kc.

The analysis was conducted using a diverse set of datasets, incorporating Sentinel-2 satellite-derived NDVI, NDWI, NDRE, and LAI datasets, each possessing a high spatial resolution of 10 m. The LAI was calculated from the Sentinel-2 data using the methodology developed in the S2ToolBox within SNAP 5.0. S2ToolBox is an essential component of the SNAP 5.0 software and was specifically developed for LAI retrieval. The methodology uses the approach of an Artificial Neural Network (ANN) to derive these parameters from instantaneous observations of Sentinel-2. By utilizing a pre-trained neural network, the necessary biophysical variables are rapidly retrieved for each pixel within the selected Sentinel-2 image. The training database is generated using a Radiative Transfer Model (RTM) [57].

These datasets were acquired through the VULTUS API (https://api.vultus.se/graphql (accessed on 15 February 2021)). The temporal resolution of the indices was set at five days, aligning with the satellite’s revisit time. A comprehensive dataset, comprising 28 Vegetation Indices (VI) images and associated data, was annually collected, spanning from sowing to harvest. Only data free from cloud cover were considered for this study. This resulted in a final dataset of 996 observations. To ensure temporal coherence in the data, a Piecewise Cubic Hermite Interpolating Polynomial (PCHIP) approach was employed, resulting in a smoothed daily curve for indices throughout the vegetation period. While Acqua Campus of CER provided the weather data (daily minimum and maximum temperature, wind speed, relative humidity, and solar radiation) using an in situ weather station, as well as the Biologische Bundesanstalt, Bundessortenamt and CHemical industry (BBCH) [58] and LAI datasets (local observations). The weather data were used for calculating ET0, whilst BBCH and LAI were used to define the lengths of the phenological stages. Crop phenophases were assessed qualitatively by biweekly survey and sampling of maize plants in the field. The BBCH scale is used to identify the lengths of phenological development stages of plants [59]. LAI measurement was carried out to assess the length of phenological stages through direct methods, which involves the removal of plants in the field on certain test areas and measuring their area in the laboratory. More details for LAI measurements are described by Nagy et al. [60].

2.3. VI-Based ETc Calculations

The crop evapotranspiration ETc was calculated using the following equation (ETc = ET0 × Kc). ET0 was defined by temperature, relative air humidity, solar radiation, wind speed, and air pressure according to the FAO Penman-Monteith equation [10] (Equation (1)).

ET0=0.408∆Rn−G+γ900T+273u2es−ea∆+γ1+0.34u2
(1)

where:

  • ET0: reference crop evapotranspiration (mm day−1),

  • Rn: net radiation at the crop surface (MJ m−2 day−1),

  • G: soil heat flux density (MJ m−2 day−1),

  • T: mean daily air temperature at 2 m height (°C),

  • u2: wind speed at 2 m height (m s−1),

  • es: saturation vapor pressure (kPa),

  • ea: actual vapor pressure (kPa),

  • es − ea: saturation vapor pressure deficit (kPa),

  • Δ: slope vapor pressure curve (kPa °C−1),

  • γ: psychrometric constant (kPa °C−1).

Based on the BBCH and LAI, the lengths of the phenological stages were calculated for the three assessed years (Table 2). In general, the sowing time was in late March or early April, the harvesting was in September. The initial stage starts on the day of sowing and continues until about 10% of the soil cover is reached. The early root growth of maize makes it particularly sensitive to drought stress. In the early stages, evaporation is mainly in the form of soil evaporation and the leaf area is small. The mid-season is characterized by rapid growth of maize, from 10% soil cover to actual full soil cover. As the plant grows and covers a larger area of soil, transpiration gradually takes over as the primary mechanism, while evaporation becomes more limited, increasing water use. Water demand is highest in the mid-season stage due to high transpiration, increased biomass production, and grain formation, and Kc reaches its maximum value. In the late season stage, transpiration-related water demand may decrease, but at this stage, drought stress can still have a significant negative impact on productivity [61,62,63].

Table 2

The lengths of the phenological stages.

YearStagePeriod, T [Days]Kc Range, [–]
2018Initial3 April 2018–28 May 20180.4–0.8
Crop development29 May 2018–29 June 20180.8–1.2
Mid-season30 June 2018–4 August 20181.2–0.9
Late-season21 July 2018–21 September 20180.9–0.6
2019Initial20 March 2019–20 May 20190.4–0.8
Crop development21 May 2019–8 June 20190.8–1.2
Mid-season9 June 2019–25 July 20191.2–0.9
Late-season14 July 2019–14 September 20190.9–0.6
2020Initial25 March 2020–11 May 20200.4–0.8
Crop development12 May 2020–1 June 20200.8–1.2
Mid-season2 June 2020–26 July 20201.2–0.9
Late-season13 July 2020–7 September 20200.9–0.6

Since the determination method of Kc presented in the FAO-56 document shows a trapezoid development, and the crop coefficient is defined as a constant throughout the initial and mid-stage, it does not reflect the natural continuous process of canopy development [32]. This approach is referred to as “original” in this study. Therefore, in this study, the trapezoid character of the FAO Kc was smoothed to have continuously varying characteristics by the application of two technical approaches curve with continuously varying values to be closer to natural crop development. One of the approaches was to cut 10 days around the breakpoints and the spline algorithm was implemented to estimate missing values. The solution is called “curved” in this study. The other approach was the so-called “MidPoint”, where only the middle stage plateau was determined by its middle point and the length of the plateau phase.

To analyze the relationship between the FAO-56-based (“original”, “MidPoint”, “curved”) Kc values and the four Sentinel-2 derived indices (NDVI, NDRE, NDWI, and LAI), Pearson Product-Moment correlation was applied. Based on the results, FAO-56-based Kc was selected, which had the highest correlation with the Vis. Then linear and power regression models were developed based on VI data. Since certain Vis (such as NDVI) have proved to be an exponential nature compared to the other indices, both linear and power regression models were evaluated [61,64,65,66]. The use of standard linear or power regression models with standard estimation techniques is subject to several conditions regarding the explanatory (x) (which were the Vis) and output (y) variables (which were the Kc values) and their relationship.

Using the Vis-based estimated Kc and FAO-56 Kc, standard ETc was calculated for all parcels for each year, resulting in biomass-derived evapotranspiration for the vegetation period of the maize fields. To evaluate the VI-based estimation ETc models, the following statistical indicators were selected: adjusted R2 (R2adj) (Equation (2)), Root Mean Square Error (RMSE) (Equation (3)), Normalized Root Mean Square Error (NRMSE) (Equation (4)), Mean Bias Errors (MBE) (Equation (5)) and Mean Absolute Error (MAE) (Equation (6)). RMSE is the sample standard deviation of the differences between predicted and actual values and NRMSE is useful to reflect the relative error between the modeled and measured soil moisture. The MBE was used to assess over or underestimation of VI-based ETc. The MAE evaluates the mean magnitude of the errors in predictions without considering their sign. The reason behind employing RMSE is its ability to measure accuracy to compare predicting errors of varying models for a particular dataset and not between datasets. The study used adjusted R2 because it provides a measure of how well the predicted values are replicated by the model based on the variability from the actual values. All the parameters are defined as follows: ETcAi is the evapotranspiration based on theoretical Kc values, ETcPi is the simulated evapotranspiration based on the spectral index derived and simulated Kc values  ETc− is the mean value, n is the total number of data points and k is the number of variables in the model.

Adjusted R2:

Radj2=1−(1−∑(ETcpi−ETcAi)2∑(ETcpi−ETc−)2)(n−1)n−k−1
(2)

Root Mean Square Error:

RMSE=1N∑i=1N(ETcpi−ETcAi)2
(3)

Normalized Root Mean Square Error:

NRMSE=100×1N∑i=1N(ETcpi−ETcAi)2ETc−
(4)

Mean Bias Error:

MBE=1n∑i=1nETcpi−ETcAi
(5)

Mean Absolute Error:

MAE=1N∑i=1N(ETcpi−ETcAi)
(6)

3. Results and Discussion

3.1. Trends of NDWI, NDRE, NDVI, and LAI Values for the Different Growth Stages of Maize

The mean NDWI, NDRE, NDVI, and LAI values over the three years (2018–2020) for the different growth stages are presented in Figure 2. The NDWI is a commonly used index for monitoring vegetation water content. In the initial stage (days 80–140), the NDWI started with negative values, with a mean value of −0.05. Other authors shown that NDWI values for maize were low in the initial stage because the vegetation is still in its early stages of development and has not yet fully developed its leaf canopy [48,67]. As the plants grew and their leaf canopy expanded, the NDWI values began to increase steadily from 0.012 (day 120) in the crop development stage, reaching its maximum (0.374) on day 174 at the mid-season stage (days 141–224). These results are consistent with a study by Zhou et al. [68], which underscored the importance of vegetation monitoring by showing how effectively NDWI could capture the variability in canopy leaf abundance during specific phases. The amount of water stored in the plants decreases by a mean value of 0.0071 in the late season (days 225–249) due to the decreasing values.

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g002.jpg

The mean NDWI, NDRE, NDVI, and LAI values over the three years (2018–2020).

The trends and distribution of NDRE values are similar to NDWI values. The mean NDRE values during the initial phase (days 80–140) were relatively low, at 0.2. This suggests that the leaves had little moisture content and that vegetation had not yet reached its full potential. In the crop development stage (days 141–179), the NDRE increased steadily as the vegetation grew, reaching a maximum on day 174 (0.82). During the mid-season (days 180–224), the mean NDRE values were similar to those around 0.76 at the crop development stage, similar to Liu et al. [69], who found, among other things, that the NDRE index has the highest correlation with canopy nitrogen concentration in summer maize. In the late-season stage (days 225–240), NDRE values showed a decreasing trend due to the gradual depletion of crop moisture.

In the initial phase (days 80–140), the mean NDVI values were low, with a mean value of 0.21. According to Lima et al. [70], NDVI values were low in the early part of the phenological cycle because of the higher radiation absorption in the near-infrared band during this period. In the crop development (days 141–179), NDVI values gradually increased and reached a maximum on day 174 (0.34). In the mid-season stage, values ranged from 0.56 to 0.67 until day 224 (late season). Reyes-González et al. [32] and Mebrie et al. [71] suggest that the crop coefficient varies constantly during the growing season and cannot remain constant throughout the mid-stage. In the late stage, the NDVI values showed a decreasing trend. Several studies have been conducted to investigate NDVI at different growth stages. Wang et al. [72] found similar values at different stages. The initial NDVI value of 0–0.2 increased with plant development up to 0.3–0.6, reaching a maximum of 0.7–0.8. Hatfield and Prueger [64] and Ji et al. [73] also divided the maize growing season into four growth phases and calculated phenological information metrics for each growth phase based on NDVI time series, while other authors have divided the growth phase into three, six, or even ten stages [74,75,76].

The trends and distribution of LAI values were similar to the previous indices. At the initial stage (days 80–140), the mean LAI values were low, around 0.5, as the vegetation was not fully developed, and the number of leaves was low. This statement aligns with the findings of Towers et al. [77], which supports the observation that at the initial stage, the mean LAI values were low due to underdeveloped vegetation and a low leaf count. In the crop development stage, the values increased as already described above and reached a maximum on day 174 (3.26). In the mid-season stage (days 180–224), the vegetation maintains a mean LAI value of around 2.7. The trend of data aligns with the findings of Zhang et al. [78] which supports the observation that LAI values follow an upward trend during the crop development stage and maintain a mean value during the mid-season stage in maize canopies. The LAI values also show a downward trend as leaf area decreases, like the other spectral indices. Similar trends in LAI values have been reported by several authors [79,80,81].

Overall, the trends and distributions of NDWI, NDRE, NDVI, and LAI values were similar, and the data followed well the vegetation development and environmental conditions.

3.2. Deriving VI-Based Crop Coefficient for the Vegetation Period of Maize

The lowest Kc values were observed during the initial stage (0.4) based on FAO-56. In the case of the “curved” dataset, it typically starts to gradually increase from this initial value about 10 days earlier compared to the “original” and “MidPoint” values. However, the “original” Kc values peak much earlier at the value of 1.2 in the crop development stage (days 166–170) than the other two datasets. Following the “original” Kc values, the “curved” dataset reaches its maximum (1.2) approximately 10 days later, followed by the “MidPoint” dataset after 7–10 days. As illustrated in Figure 3, this plateau phase represents approximately a 50-day period in the original Kc values, a 26-day period in the “curved” dataset for 2018 and 2019, and a 35-day period in 2020. In the case of the “MidPoint” dataset, this phase lasted for about only 8–9 days.

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g003.jpg

FAO-56-based (“original”, “MidPoint”, “curved”) Kc values for the years (2018–2020).

There was a significant correlation between FAO-56-based Kc values and VIs, thus proper regression models were capable of describing the real crop evapotranspiration based on the evaluated vegetation index in this study (Table 3). The correlation between Vis and FAO-56-based Kc smoothed by the “MidPoint” method was found to be the highest, therefore “MidPoint” Kc was selected to use for setting regression models between the VIs and FAO-56 Kc. The correlation matrix indicates that LAI was less effective in reproducing the temporal pattern of the theoretical curves, while NDWI and NDVI seem to be the most effective in predicting Etc.

Table 3

Correlation coefficients between estimated Kc curves.

FAO-56
n OriginalCurvedMidPoint
Spectral indicesNDWI1320.9410.9140.952
NDRE1320.9340.8880.939
NDVI1320.9420.9010.952
LAI1320.8940.8530.889

Both linear and power regression models were set based on “MidPoint” Kc and the mean NDWI, NDRE, NDVI, and LAI values (Figure 4). The linear correlation between mean NDWI values and FAO crop coefficient of maize (R2) was 0.83, indicating a strong positive correlation between the two variables. The R2 value of the power trend line was even higher (0.88). However, we could only fit this to the positive integers, which cover the period from day 129 to day 234 of the year, while the total growth stage was from day 84 to day 249. The relationship between NDRE and FAO Kc indicates a strong positive correlation for both linear and power fitting. The values were almost identical, R2 being 0.8858 in the former and 0.8856 in the latter case.

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g004.jpg

The linear and power regression models between the “MidPoint” Kc and the NDWI (n = 132), NDRE (n = 132), NDVI (n = 132), and LAI (n = 132) values of maize.

The correlation between NDVI and FAO data for maize has been investigated in several studies. In this study, for the NDVI, almost the same can be stated as for the NDRE index, as the R2 values were almost the same for both linear and power trend lines (R2 = 0.886 and 0.8875) and show a strong correlation between NDVI and FAO Kc. The relationship between the NDVI and the crop coefficient (Kc) for maize has been extensively studied. A strong linear association between the NDVI and FAO crop coefficient values was also discovered by Costa et al. [82]; the former’s value was 0.86, while the latter was 0.79. Reyes-González et al. [66] discovered that the NDVI and Kc FAO-56 had an even greater coefficient of determination of 0.97. Javed et al. [83] also reported that crop coefficients are highly correlated with satellite-derived NDVI values, suggesting a direct proportional relationship between the maize crop coefficient (FAO-56) and NDVI. Costa et al. [82] revealed a significant correlation between NDVI and the Kc values reported in the FAO-56 report with R2 = 0.794, an indication that there is a high correlation within these parameters. Singh and Irmak [31] formulated a linear regression model to determine the correlation between NDVI and ET0-based Kc. The results showed that the coefficient of variation for both NDVI and Kc was lower for maize, soybeans, sorghum, and alfalfa at midseason compared to the early growing season as well as the late growth stage.

The lowest R2 was detected in the case of LAI between the index and the crop coefficient. The value for linear regression was 0.7712, whereas R2 in power fitting was slightly lower at 0.7455. According to Kang et al. [84], the Kc exhibited a rapid increase as the LAI approached 3 (linear R2 = 0.9522), maintaining elevated values when LAI exceeded 2.5 (linear R2 = 0.7612). The study by Jia and Wang [85] revealed a significant exponential relationship between LAI and crop coefficient Kc (R2 = 0.7055) under water deficit conditions, indicating the influence of vegetation indices on Kc, and the impact of water stress on this relationship.

Figure 5 illustrates the comparison between the FAO crop coefficient and crop coefficients estimated by linear and power functions for the growth stages. In the case of NDWI, in the initial stage (days 80–140), the linear calibrated Kc values were higher than the FAO Kc data, except on day 84, when the linear Kc was 0.379 and the FAO Kc was 0.4. The power-calibrated Kc data, as can be seen in Figure 5, could only be examined from day 129 onwards. In this short initial period (days 129–140), the power-calibrated Kc values and the FAO Kc values were almost the same. In the crop development stage (days 141–179), the power and linear calibrated Kc values were very similar (with a mean value of 0.99 in this stage) and had values higher than the FAO Kc (mean value of 0.94). In the mid-season stage (days 180–224), both FAO Kc and VI calibrated Kc values were above 1. Linear and power-calibrated values were again similar. The maximum crop coefficient value for the FAO data was on the 194th day (1.2), while for the power and linear calibrated data it was on the 174th day, 1.18 in the former case and 1.22 in the latter case. In the late season (days 225–250), values were again below 1, and similar to the mid-season, FAO Kc values were higher for both power and linear calibrated values. For the NDRE index, the linear and power-calibrated Kc values were almost equal in all stages. The FAO Kc data showed higher values than the calibrated data between days 91 and 115 of the initial stage (days 80–140), but for the remaining days of the initial stage and the whole crop development stage (days 141–179), the calibrated data were generally higher than the FAO Kc data. While in the initial stage, the mean of the calibrated data was 0.46 (for both power and linear), the FAO Kc was 0.41. The maximum values of power and linear calibrated Kc’s were also the highest on the 174th day, 1.166 and 1.18, respectively. In the crop development stage, the mean of the calibrated Kc values for power and linear were 1.01 and 1.03, respectively, while the FAO Kc was 0.94. FAO Kc values were on average higher in the mid-season (days 180–224) and at the late season (days 225–250) stages, except for days 208–212 and 217–225, when linear calibrated Kc values were higher than both power calibrated and FAO Kc data. While the power and linear calibrated crop coefficient was 1.6–1.7 in mid-season and 0.7 in late season (for both power and linear). In the case of the crop coefficient values of the NDVI index, the linear and power-calibrated Kc values matched even better than in the case of the NDRE. The trend of the values was also similar to the trend already presented for the NDRE index, so the obtained Kc values from the power and linear fittings in the initial (days 80–140) and crop development (days 141–179) stages (0.44 and 0.45 in the initial stage, and 1.04 and 1.05 in the crop development stage) were higher than the FAO Kc mean values (0.4 in the initial stage and then 0.94 in the crop development stage). In the mid-season (days 180–224) and late-season stages (days 225–250), the FAO Kc values were higher (1.14 and then 0.82 with a mean value) than the power and linear calibrated Kc values. The maximum values of the power and linear calibrated Kc’s for the NDVI were also highest on day 174, 1.207, and 1.224, respectively. For LAI, the calibrated values, on average the linear calibrated Kc values (mean value of 0.55), were higher in the initial phase (days 80–140). However, the highest value at this stage was produced by power-calibrated Kc values with 0.774 on day 129. This trend continued in the crop development phase (days 141–179), the mean value of the FAO Kc’s was 0.94, while the calibrated Kc values exceeded 1. The maximum crop coefficient values for the power and linear calibrated data were 1.19 in the former case and 1.29 in the latter case. If the mean values were taken into account, the mean of the FAO Kc values were higher (1.13 and 0.82) in the mid-season (180–224th day) and late-season stages (225–250th day). Among the calibrated crop coefficients, in the mid-season, the linear calibrated values were higher (with a mean value of 1.12, while the power was 1.08), while in the mid-season the power calibrated Kc values were slightly higher (0.65—with a mean value of 0.64).

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g005.jpg

Comparison of FAO-56 crop coefficient, and linear and power-calibrated crop coefficient of maize with NDWI, NDRE, NDVI, and LAI indices.

3.3. Estimated Crop Evapotranspiration of Maize along the Growth Stages

The estimation of crop evapotranspiration for maize at different growth stages is a critical aspect of agricultural water management (Table 4). The minimum NDWI-based crop evaporation was on day 80 for both the power (1.34 mm/day ± 0.102) and linear (1.37 mm/day ± 0.049) calibrated data, while the maximum value for power was on day 185 (6.45 mm/day ± 0.367) and for the linear calibrated ETc data on day 170 (6.51 mm/day ± 0.434).

Table 4

Vegetation index (NDWI, NDRE, NDVI, LAI)-based estimation of crop evapotranspiration (ETc) (mm/day).

Day of Year8095110125140155170185200215230245
NDWIpower1.34 ± 0.1021.54 ± 0.5342.01 ± 0.2761.87 ± 0.2422.46 ± 0.2844.1 ± 0.6465.88 ± 0.496.45 ± 0.3676.32 ± 0.5945.85 ± 0.5454.7 ± 0.5073.4 ± 0.704
linear1.37 ± 0.0491.48 ± 0.4281.97 ± 0.3342.09 ± 0.4082.6 ± 0.5814.19 ± 0.9586.51 ± 0.4346.22 ± 0.5726.12 ± 0.2945.9 ± 0.4534.54 ± 0.6913.23 ± 1.229
NDREpower1.37 ± 0.0411.36 ± 0.3551.86 ± 0.11.72 ± 0.1482.32 ± 0.2214.13 ± 0.9576.11 ± 0.5576.76 ± 0.3376.62 ± 0.346.11 ± 0.5724.78 ± 0.2233.31 ± 0.43
linear1.38 ± 0.071.29 ± 0.2931.9 ± 0.3021.89 ± 0.3532.53 ± 0.3654.54 ± 1.1876.4 ± 0.8385.92 ± 0.466.09 ± 0.2485.93 ± 0.5224.92 ± 0.4093.23 ± 0.802
NDVIpower1.32 ± 0.0241.31 ± 0.3661.78 ± 0.2111.65 ± 0.2492.23 ± 0.2233.98 ± 0.9415.89 ± 0.5026.53 ± 0.3316.4 ± 0.3755.92 ± 0.5384.57 ± 0.2083.14 ± 0.431
linear1.36 ± 0.0511.22 ± 0.2861.82 ± 0.2651.84 ± 0.3642.42 ± 0.2174.43 ± 1.1626.46 ± 0.6695.79 ± 0.725.92 ± 0.3095.8 ± 0.4384.64 ± 0.462.98 ± 0.735
LAIpower1.32 ± 0.0151.38 ± 0.3321.83 ± 0.2521.71 ± 0.3232.22 ± 0.1313.73 ± 0.9125.31 ± 0.3135.83 ± 0.4925.74 ± 0.5545.31 ± 0.454.17 ± 0.3572.99 ± 0.302
linear1.34 ± 0.0361.27 ± 0.2931.87 ± 0.232.14 ± 0.2832.26 ± 0.2813.77 ± 1.4275.93 ± 0.3225.7 ± 0.6185.92 ± 0.5365.2 ± 0.6863.66 ± 0.3922.57 ± 0.368

For NDRE, the minimum values were on day 95, which is still in the initial stage, with power at 1.36 mm/day ± 0.355 and linear calibrated ETc at 1.29 mm/day ± 0.293. The maximum ETc values were similar to the NDWI-based ETc, i.e., the power-calibrated ETc had its highest value at mid-season (6.76 mm/day ± 0.337 on day 185), while the linear calibrated ETc had its highest value towards the end of the crop development stage (6.4 ± 0.838 on day 170).

The evolution of NDVI-based Kc values was very similar to NDRE-based Kc values. The minimum value was on day 95, 1.31 mm/day ± 0.366 for the power calibration and 1.22 mm/day ± 0.286 for the linear calibration, and the maximum was on day 185 for the power calibrated Kc values (6.53 mm/day ± 0.331) and on day 170 for the linear calibrated values (6.46 mm/day ± 0.669).

The power-calibrated LAI-based Kc was lowest on day 80 (1.32 mm/day ± 0.015) and highest on day 185 (5.83 mm/day ± 0.492). In contrast, the linear calibrated Kc values had a minimum on day 95 (1.27 mm/day ± 0.293) and a maximum on day 170 (5.93 mm/day ± 0.322).

In general, the vegetation index-based crop coefficient values were lowest in the initial stage, while the highest values were towards the end of the crop development stage (for linear calibrated, etc.) and in the mid-season stage (for power-calibrated ETc).

3.4. Evaluation of Vegetation Indices (NDWI, NDRE, NDVI, and LAI) Based on ETc Compared to FAO-56 Kc

Based on the adjusted R-squared, the values closest to 1 were observed for the NDWI, NDRE, and NDVI indices, indicating that these models were able to explain a larger proportion of the variability in the data (Figure 6). For these three indices, the power regression model exceeded the linear regression model in general by a small margin. The LAI index had the lowest adjusted R-squared value, suggesting that the models could not fully account for the variability in the data. For LAI, the power regression model outperformed the linear regression model by a small margin; nonetheless, both models’ predictions displayed very significant variability, indicating that the models are less accurate in predicting LAI compared to NDWI, NDRE, or NDVI.

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g006.jpg

Adjusted R2 between ETc estimated with the vegetation indices (NDWI, NDRE, NDVI, and LAI) and that estimated with FAO-56 Kc.

According to the RMSE, the NDWI, NDRE, and NDVI indices exhibit similar RMSE values across both regression models (ranging between 0.43 and 0.45 mm/day), indicating comparable performance for predicting these values (Figure 7). The LAI index demonstrated the highest RMSE values for both regression models. However, the linear regression model (0.5 mm/day) appears to provide a slight advantage compared to the power regression model (0.53 mm/day).

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g007.jpg

RMSE and NRMSE between ETc estimated with the vegetation indices (NDWI, NDRE, NDVI, and LAI) and that estimated with FAO-56 Kc.

The results for NRMSE were similar to those for the statistical indices described so far (Figure 7). For the NDRE index, the NRMSE values were relatively close, with the linear regression model achieving slightly lower values compared to the power regression model (11.36%). This indicates that both models perform similarly in predicting NDRE values. The values for the NDVI index also showed a mixed pattern. The linear regression model achieves relatively lower mean NRMSE values (11.56%) compared to the power regression model (11.48%), suggesting a small advantage for the linear regression model in predicting NDVI values. The LAI index exhibited the highest NRMSE values for both regression models.

For MAE, the lowest values were observed for NDVI (0.32 and 0.33), but NDWI and NDRE values were also close to this (0.34 and 0.35) (Figure 8). The standard deviation of the values ranged from 0.08 to 0.11. Overall, the NDWI, NDRE, and NDVI indices exhibited similar MAE values across both regression models, indicating comparable performance for predicting these values. The LAI index demonstrates the highest MAE values, for both regression models, with the power regression model achieving a mean value of 0.40 mm/day and the linear regression model achieving a mean value of 0.37 mm/day.

An external file that holds a picture, illustration, etc. Object name is plants-13-01212-g008.jpg

MAE and MBE between ETc estimated with the vegetation indices (NDWI, NDRE, NDVI, and LAI) and that estimated with FAO-56 Kc.

In the case of MBE (Figure 8), the estimates of the linear regression model show mean smaller deviations from the actual values than the estimates of the power regression model. This was particularly the case for NDWI and LAI, where the mean NDWI for the power regression model was −0.04 mm and −0.005 mm for the linear model, while for LAI it was −0.042 mm and −0.003 mm. The NDRE and NDVI indices were closer to 0 for both the power and linear regression models.

4. Conclusions

In the Mediterranean, water competition among different stakeholders becomes more intense and affects agricultural production, especially in areas of high-intensive use like northern Italy. Changes in temperature patterns, rainfall, and extreme weather events affect agricultural outputs and make it difficult to plan optimal distribution of water and could lead to food insecurity. Effective water management becomes critical to addressing these challenges. For solving these issues, modern technological approaches such as RS provide effective solutions. In this study, researchers have created an innovative RS-based approach for calculating crop coefficients and evapotranspiration based on a case study site in Emilia Romagna, Italy. Through the use of remotely sensed indices coupled with climatic data, the accuracy in estimating crop evapotranspiration was substantially increased for maize fields.

The comparative study of NDVI, NDWI, NDRE, and LAI values in different stages of maize growth helps understand the changes that occur during vegetation development under changing environmental conditions. In addition, the relationship between these indices and Kc reveals their suitability in predicting crop evapotranspiration. The observed strong positive correlations between NDVI, NDWI, NDRE, and LAI values with FAO Kc indicate the possibility of using RS data to obtain reasonable estimates for crop water requirements. Regression models of NDVI, NDWI, and NDRE exhibit slightly different predictive performances in determining Kc values with power regression generally being superior to linear regressions. The most significant correlations are found for NDVI and NDWI, indicating their strength in predicting Kc values in maize. Nevertheless, there are lower correlation values for LAI which indicates that the predictive accuracy of this index is less than other indices.

Generally, NDWI, NDRE, and NDVI-based ETc estimations outperform consistently compared to LAI with power regression providing better results. In the end, it can be concluded that VI (NDWI, NDRE, NDVI, and LAI)-based crop evapotranspiration ETc are indicating different performances in predicting water requirements for crops compared to FAO-56 Kc-based ETc. NDWI, NDRE, and NDVI have good explanatory power with relatively accurate predictions for both models of regression while the LAI index shows some limitations in its accuracy. The power regression model usually slightly exceeds the linear regression model, NDWI, NDRE, and NDVI have better predictive accuracy than LAI. These results emphasize the importance of vegetation indices, especially NDWI and NDVI as promising approaches to estimate crop water requirements with further improvements required for LAI-based predictions.

In conclusion, the analysis shows that NDWI and NDVI indices are effective in monitoring maize growth stages and estimating crop evapotranspiration. Using RS data and regression modeling techniques provides appropriate tools for the improvement of agricultural water management practices to achieve higher crop productivity. More research and improvement of these methodologies can lead to more accurate and consistent strategies for crop water management in agricultural systems.

Acknowledgments

The authors would like to thank VULTUS company (https://www.vultus.se/) for providing their API for remote sensing data acquisition.

Funding Statement

The research was funded by the European Union’s Horizon 2020 “WATERAGRI Water retention and nutrient recycling in soils and streams for improved agricultural production” research and innovation program under Grant Agreement No. 858375. This research was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences.

Author Contributions

F.C., S.L.G. and H.A. contributed to the data acquisition; E.B.-B. and T.M. contributed to the data analysis; Z.Z.F. contributed to the data analysis and conceptualization; A.N. contributed to the conceptualization and funding; A.N. and N.É.K. contributed to the data analysis, interpretation, and writing; J.T. revised the manuscript and funding. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding authors.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

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.

References

1. Malhi G.S., Kaur M., Kaushik P. Impact of climate change on agriculture and its mitigation strategies: A review. Sustainability. 2021;13:1318. 10.3390/su13031318. [CrossRef] [Google Scholar]
2. Maselli F., Battista P., Chiesi M., Rapi B., Angeli L., Fibbi L., Magno R., Gozzini B. Use of Sentinel-2 MSI data to monitor crop irrigation in Mediterranean areas. Int. J. Appl. Earth Obs. Geoinf. 2020;93:102216. 10.1016/j.jag.2020.102216. [CrossRef] [Google Scholar]
3. Hartmann J.A., West J., Renforth P., Köhler P., De La Rocha C.L., Wolf-Gladrow D.A., Dürr H.H., Scheffran J. Enhanced chemical weathering as a geoengineering strategy to reduce atmospheric carbon dioxide, supply nutrients, and mitigate ocean acidification. Rev. Geophys. 2013;51:113–149. 10.1002/rog.20004. [CrossRef] [Google Scholar]
4. Jalilvand E., Tajrishy M., Hashemi S.A.G.Z., Brocca L. Quantification of irrigation water using remote sensing of soil moisture in a semi-arid region. Remote Sens. Environ. 2019;231:111226. 10.1016/j.rse.2019.111226. [CrossRef] [Google Scholar]
5. Guzinski R., Nieto H. Evaluating the feasibility of using Sentinel-2 and Sentinel-3 satellites for high-resolution evapotranspiration estimations. Remote Sens. Environ. 2019;221:157–172. 10.1016/j.rse.2018.11.019. [CrossRef] [Google Scholar]
6. Giannini A., Bagnoni V. Schede di Tecnica Irrigua per L’agricoltura Toscana. ARSIA—Servizio Telematico Irrigazione, Regione Toscana, EFFEMME Lito; Florence, Italy: 2020. pp. 66–97. [Google Scholar]
7. Maina M.M., Amin M.S.M., Rowshon M.K., Aimrun W., Samsuzana A.A., Yazid M.A. Effects of crop evapotranspiration estimation techniques and weather parameters on rice crop water requirement. Aust. J. Crop Sci. 2014;8:495–501. [Google Scholar]
8. Djaman K., Mel V.C., Ametonou F.Y., Namaky R.E., Diallo M.D., Koudahe K. Effect of Nitrogen Fertilizer Dose and Application Timing on Yield and Nitrogen Use Efficiency of Irrigated Hybrid Rice under Semi-Arid Conditions. J. Agric. Sci. Food Res. 2018;9:223. [Google Scholar]
9. Yimam A.Y., Afa T.T., Sishu F.K., Tilahun S.A., Reyes M.R., Vara Prasad P.V. Estimating surface and groundwater irrigation potential under different conservation agricultural practices and irrigation systems in the Ethiopian highlands. Water. 2021;13:1645. 10.3390/w13121645. [CrossRef] [Google Scholar]
10. Allen R.G., Pereira L.S., Raes D., Smith M. FAO Irrigation and Drainage Paper 56. Food and Agriculture Organization of the United Nations; Rome, Italy: 1998. Crop evapotranspiration guidelines for computing crop requirements; pp. 19–40. [Google Scholar]
11. Zhao L., Xia J., Xu C., Wang Z., Sobkowiak L., Long C. Evapotranspiration estimation methods in hydrological models. J. Geogr. Sci. 2013;23:359–369. 10.1007/s11442-013-1015-9. [CrossRef] [Google Scholar]
12. Abdollahi K., Bazargan A., McKay G. Water Balance Models in Environmental Modeling. In: Hussain C., editor. Handbook of Environmental Materials Management. Springer; Cham, Switzerland: 2019. [CrossRef] [Google Scholar]
13. Nagy A., Tamás J. Non-invasive water stress assessment methods in orchards. Commun. Soil. Sci. Plant Anal. 2013;44:1–4. 10.1080/00103624.2013.742308. [CrossRef] [Google Scholar]
14. Romano G., Ricci G.F., Gentile F. Influence of Different Satellite Imagery on the Analysis of Riparian Leaf Density in a Mountain Stream. Remote Sens. 2020;12:3376. 10.3390/rs12203376. [CrossRef] [Google Scholar]
15. Mu Q., Zhao M., Running S.W. Improvements to a MODIS Global Terrestrial Evapotranspiration Algorithm. Remote Sens. Environ. 2011;115:1781–1800. 10.1016/j.rse.2011.02.019. [CrossRef] [Google Scholar]
16. Fisher J.B., Lee B., Purdy A.J., Halverson G., Cawse-Nicholson K., Wang A., Anderson R.G., Aragon B., Arain M.A., Baldocchi D.D., et al. First evapotranspiration results from NASA’s ECOSTRESS mission. AGU Fall Meet. Abstr. 2019;2019:H22B-07. [Google Scholar]
17. Bastiaanssen W.G.M., Meneti M., Feddes R.A., Holtslag A.A.M. A remote sensing surface energy balance algorithm for land (SEBAL) J. Hydrol. 1998;212:198–212. 10.1016/S0022-1694(98)00253-4. [CrossRef] [Google Scholar]
18. Norman J.M., Kustas W.P., Humes K.S. Source approach for estimating soil and vegetation energy fluxes in observations of directional radiometric surface temperature. Agric. For. Meteorol. 1995;77:263–293. 10.1016/0168-1923(95)02265-Y. [CrossRef] [Google Scholar]
19. Basit A., Khalil R.Z., Haque S. Application of simplified surface energy balance index (S-SEBI) for crop evapotranspiration using Landsat 8. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. ISPRS Arch. 2018;XLII-1:33–37. 10.5194/isprs-archives-XLII-1-33-2018. [CrossRef] [Google Scholar]
20. Allen R.G., Tasumi M., Trezza R. Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)—Model. J. Irrig. Drain. Eng. 2007;133:380–394. 10.1061/(asce)0733-9437(2007)133:4(380). [CrossRef] [Google Scholar]
21. Chen M., Senay G.B., Singh R.K., Verdin J.P. Uncertainty analysis of the operational simplified surface energy balance (SSEBop) model at multiple flux tower sites. J. Hydrol. 2016;536:384–399. 10.1016/j.jhydrol.2016.02.026. [CrossRef] [Google Scholar]
22. Ozdogan M., Yang Y., Allez G., Cervantes C. Remote sensing of irrigated agriculture: Opportunities and challenges. Remote Sens. 2010;2:2274–2304. 10.3390/rs2092274. [CrossRef] [Google Scholar]
23. Ambika A.K., Wardlow B., Mishra V. Remotely sensed high resolution irrigated area mapping in India for 2000 to 2015. Sci. Data. 2016;3:160118. 10.1038/sdata.2016.118. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
24. Zhang K., Kimball J.S., Running S.W. A review of remote sensing based actual evapotranspiration estimation. WIREs Water. 2016;3:834–853. 10.1002/wat2.1168. [CrossRef] [Google Scholar]
25. Calera A., Campos I., Osann A., D’Urso G., Menenti M. Remote sensing for crop water management: From ET modelling to services for the end users. Sensors. 2017;17:1104. 10.3390/s17051104. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
26. Magyar T., Fehér Z., Buday-Bódi E., Tamás J., Nagy A. Modeling of soil moisture and water fluxes in a maize field for the optimization of irrigation. Comput. Electron. Agric. 2023;213:108159. 10.1016/j.compag.2023.108159. [CrossRef] [Google Scholar]
27. Bisquert J., Garcia-Belmonte G., Pitarch A., Bolink H.J. Negative capacitance caused by electron injection through interfacial states in organic light-emitting diodes. Chem. Phys. Lett. 2016;422:184–191. 10.1016/j.cplett.2006.02.060. [CrossRef] [Google Scholar]
28. Tran H.Q., Fehér Z.Z. Water balance calculation capability of hydrological models. Acta Agrar. Kaposváriensis. 2022;26:37–53. 10.31914/aak.2877. [CrossRef] [Google Scholar]
29. Corbari C., Charfi I.B., Bitar A.A., Skokovic D., Sobrino J.A., Perelli C., Branca G., Mancini M. A fully coupled crop-water-energy balance model based on satellite data for maize and tomato crops yield estimates: The FEST-EWB-SAFY model. Agric. Water Manag. 2022;272:107850. 10.1016/j.agwat.2022.107850. [CrossRef] [Google Scholar]
30. Lu Y., Wei C., McCabe M.F., Sheffield J. Multi-variable assimilation into a modified AquaCrop model for improved maize simulation without management or crop phenology information. Agric. Water Manag. 2022;266:107576. 10.1016/j.agwat.2022.107576. [CrossRef] [Google Scholar]
31. Singh R.K., Irmak A. Estimation of crop coefficients using satellite remote sensing. J. Irrig. Drain. Eng. 2009;135:597–608. [Google Scholar]
32. Reyes-González A., Kjaersgaard J., Trooien T., Hay C., Ahiablame L. Estimation of crop evapotranspiration using satellite remote sensing-based vegetation index. Adv. Meteorol. 2018;2018:4525021. 10.1155/2018/4525021. [CrossRef] [Google Scholar]
33. Revel C., Lonjou V., Marcq S., Desjardins C., Fougnie B., Coppolani-Delle Luche C., Guilleminot N., Lacamp A.-S., Lourme E., Miquel C., et al. Sentinel-2A and 2B absolute calibration monitoring. Eur. J. Remote Sens. 2019;52:122–137. 10.1080/22797254.2018.1562311. [CrossRef] [Google Scholar]
34. Coluzzi R., Fascetti S., Imbrenda V., Italiano S.S.P., Ripullone F., Lanfredi M. Exploring the Use of Sentinel-2 Data to Monitor Heterogeneous Effects of Contextual Drought and Heatwaves on Mediterranean Forests. Land. 2020;9:325. 10.3390/land9090325. [CrossRef] [Google Scholar]
35. Drusch M., Del Bello U., Carlier S., Colin O., Fernandez V., Gascon F., Hoersch B., Isola C., Laberinti P., Martimort P., et al. Sentinel-2: ESA’s Optical High-Resolution Mission for GMES Operational Services. Remote Sens. Environ. 2012;120:25–36. 10.1016/j.rse.2011.11.026. [CrossRef] [Google Scholar]
36. Herrmann A., Pimstein A., Karnieli Y., Cohen A., Bonfil D.J. LAI assessment of wheat and potato crops by VENμS and Sentinel-2 bands. Remote Sens. Environ. 2011;115:2141–2151. 10.1016/j.rse.2011.04.018. [CrossRef] [Google Scholar]
37. Laurent V.C.E., Schaepman M.E., Verhoef W., Weyermann J., Chávez R.O. Bayesian object-based estimation of LAI and chlorophyll from a simulated sentinel-2 top-of-atmosphere radiance image. Remote Sens. Environ. 2014;140:318–329. 10.1016/j.rse.2013.09.005. [CrossRef] [Google Scholar]
38. Verrelst J.P., Rivera J., Veroustraete F., Muñoz-Marí J., Clevers J.G.P.W., Camps-Valls G., Moreno J. Experimental Sentinel-2 LAI estimation using parametric, non-parametric and physical retrieval methods—A comparison. ISPRS J. Photogramm. Remote Sens. 2015;108:260–272. 10.1016/j.isprsjprs.2015.04.013. [CrossRef] [Google Scholar]
39. Zapata F., Smith E., Kreinovich V., Phuong N.H. Why Normalized Difference Vegetation Index (NDVI)? In: Phuong N.H., Kreinovich V., editors. Biomedical and Other Applications of Soft Computing. Volume 1045. Springer; Cham, Switzerland: 2023. pp. 25–36. Studies in Computational Intelligence. [CrossRef] [Google Scholar]
40. Panek E., Gozdowski D. Relationship between MODIS Derived NDVI and Yield of Cereals for Selected European Countries. Agronomy. 2021;11:340. 10.3390/agronomy11020340. [CrossRef] [Google Scholar]
41. Inman D., Khosda R., Reich R., Westfall D.G. Normalized difference vegetation index and soil-color based management zones in irrigated maize. Agron. J. 2008;100:60–66. 10.2134/agronj2007.0020. [CrossRef] [Google Scholar]
42. Rembold F., Atzberger C., Savin I., Rojas O. Using Low Resolution Satellite Imagery for Yield Prediction and Yield Anomaly Detection. Remote Sens. 2013;5:1704–1733. 10.3390/rs5041704. [CrossRef] [Google Scholar]
43. Karki T.B. Yield prediction and nitrogen recommendation in maize using normalized difference vegetation index. Agron. J. Nepal. 2013;3:82–88. 10.3126/ajn.v3i0.9009. [CrossRef] [Google Scholar]
44. Sharma L.K., Bu H., Denton A., Franzen D.W. Active-Optical Sensors Using Red NDVI Compared to Red Edge NDVI for Prediction of Corn Grain Yield in North Dakota, U.S.A. Sensors. 2015;15:27832–27853. 10.3390/s151127832. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
45. Barboza T.O.C., Ardigueri M., Souza G.F.C., Ferraz M.A.J., Gaudencio J.R.F., Santos A.F.d. Performance of Vegetation Indices to Estimate Green Biomass Accumulation in Common Bean. AgriEngineering. 2023;5:840–854. 10.3390/agriengineering5020052. [CrossRef] [Google Scholar]
46. Boiarskii B., Hasegawa H. Comparison of NDVI and NDRE indices to detect differences in vegetation and chlorophyll content. J. Mech. Contin. Math. Sci. 2019;4:20–29. 10.26782/jmcms.spl.4/2019.11.00003. [CrossRef] [Google Scholar]
47. Colovic M., Yu K., Todorovic M., Cantore V., Hamze M., Albrizio R., Stellacci A.M. Hyperspectral Vegetation Indices to Assess Water and Nitrogen Status of Sweet Maize Crop. Agronomy. 2022;12:2181. 10.3390/agronomy12092181. [CrossRef] [Google Scholar]
48. Gao B.-C. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space. Remote Sens. Environ. 1996;58:257–266. 10.1016/S0034-4257(96)00067-3. [CrossRef] [Google Scholar]
49. Ashok A., Rani H.P., Jayakumar K.V. Monitoring of dynamic wetland changes using NDVI and NDWI based landsat imagery. Remote Sens. Appl. Soc. Environ. 2021;23:100547. 10.1016/j.rsase.2021.100547. [CrossRef] [Google Scholar]
50. Serrano J., Shahidian S., Marques da Silva J. Evaluation of Normalized Difference Water Index as a Tool for Monitoring Pasture Seasonal and Inter-Annual Variability in a Mediterranean Agro-Silvo-Pastoral System. Water. 2019;11:62. 10.3390/w11010062. [CrossRef] [Google Scholar]
51. Fang H., Zhang Y., Wei S., Li W., Ye Y., Sun T., Liu W. Validation of global moderate resolution leaf area index (LAI) products over croplands in northeastern China. Remote Sens. Environ. 2019;233:111377. 10.1016/j.rse.2019.111377. [CrossRef] [Google Scholar]
52. Broge N., Leblanc E. Comparing prediction power and stability of broadband and hyperspectral vegetation indices for estimation of green leaf area index and canopy chlorophyll density. Remote Sens. Environ. 2001;76:156–172. 10.1016/S0034-4257(00)00197-8. [CrossRef] [Google Scholar]
53. Giannerini G., Genovesi R. The water saving with Irriframe platform for thousands of Italian farms. J. Agr. Inf. 2015;6:49–55. 10.17700/jai.2015.6.4.237. [CrossRef] [Google Scholar]
54. Mannini P., Genovesi R., Letterio T. IRRINET: Large scale DSS application for on-farm irrigation scheduling. Procedia Environ. Sci. 2013;19:823–829. 10.1016/j.proenv.2013.06.091. [CrossRef] [Google Scholar]
55. Cavazza F., Galioto F., Raggi M., Viaggi D. The Role of ICT in Improving Sequential Decisions for Water Management in Agriculture. Water. 2018;10:1141. 10.3390/w10091141. [CrossRef] [Google Scholar]
56. [(accessed on 15 February 2021)]. Available online: https://consorziocer.it/it/strumenti/monitoraggio-risorse/
57. Abdullah H., Omar D.k., Polat N., BilgiliSharef S.H. A comparison between day and night land surface temperatures using acquired satellite thermal infrared data in a winter wheat field. Remote Sens. Appl. Soc. Environ. 2020;19:100368. 10.1016/j.rsase.2020.100368. [CrossRef] [Google Scholar]
58. Hack H., Bleiholder H., Buhr L., Meier U., Schnock-Fricke U., Weber E., Witzenberger A. Einheitliche Codierung der phänologischen Entwicklungsstadien mono- und dikotyler Pflanzen—Erweiterte BBCH-Skala, Allgemein. Nachrichtenbl. Deut. Pflanzenschutzd. 1992;44:265–270. [Google Scholar]
59. Meier U. BBCH Monograph. Federal Biological Research Centre for Agriculture and Forestry; Bonn, Germany: 2001. Growth Stages of Mono and Dicotyledonous Plants. [Google Scholar]
60. Nagy A., Szabó A., Adeniyi O.D., Tamás J. Wheat Yield Forecasting for the Tisza River Catchment Using Landsat 8 NDVI and SAVI Time Series and Reported Crop Statistics. Agronomy. 2021;11:652. 10.3390/agronomy11040652. [CrossRef] [Google Scholar]
61. [(accessed on 16 March 2021)]. Available online: https://www.fao.org/3/X0490E/x0490e0a.htm.
62. Rai R.K., Singh V.P., Upadhyay A. Planning and Evaluation of Irrigation Projects, Methods and Implementation. Academic Press; Cambridge, MA, USA: 2017. Estimating Irrigation Design Parameters; pp. 243–282. [CrossRef] [Google Scholar]
63. Irmak S. Crop evapotranspiration. Encycl. Ecol. 2008:1432–1438. [Google Scholar]
64. Hatfield J.L., Prueger J.H. Value of using different vegetative indices to quantify agricultural crop characteristics at different growth stages under varying management practices. Remote Sens. 2010;2:562–578. 10.3390/rs2020562. [CrossRef] [Google Scholar]
65. Chu D. Aboveground biomass estimates of grassland in the North Tibet using MODIS remote sensing approaches. Appl. Ecol. Environ. Res. 2020;18:7655–7672. 10.15666/aeer/1806_76557672. [CrossRef] [Google Scholar]
66. Reyes-Gonzalez A., Hay C., Kjaersgaard J., Neale C. Use of Remote Sensing to Generate Crop Coefficient and Estimate Actual Crop Evapotranspiration; Proceedings of the 2015 ASABE Annual International Meeting Sponsored by ASABE; New Orleans, LA, USA. 26–29 July 2015. [Google Scholar]
67. Alvino F.C.G., Aleman C.C., Filguerias R., Althoff D., da Cunha F.F. Vegetation indices for irrigated corn monitoring. Eng. Agrícola. 2020;40:322–333. 10.1590/1809-4430-Eng.Agric.v40n3p322-333/2020. [CrossRef] [Google Scholar]
68. Zhou H., Zhou G., Song X., He Q. Dynamic characteristics of canopy and vegetation water content during an entire maize growing season in relation to spectral-based indices. Remote Sens. 2022;14:584. 10.3390/rs14030584. [CrossRef] [Google Scholar]
69. Liu L., Peng Z., Zhang B., Wei Z., Han N., Lin S., Chen H., Cai J. Canopy Nitrogen Concentration Monitoring Techniques of Summer Corn Based on Canopy Spectral Information. Sensors. 2019;19:4123. 10.3390/s19194123. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
70. Lima G.C., Silva M.L.N., Curi N., da Silva M.A., Oliveira A.H., Avanzi J.C., Ummus M.E. Avaliação da cobertura vegetal pelo índice de vegetação por diferença normalizada (IVDN) Rev. Ambiente Água. 2013;8:204–214. [Google Scholar]
71. Mebrie D.W., Assefa T.T., Yimam A.Y., Belay S.A. A remote sensing approach to estimate variable crop coefficient and evapotranspiration for improved water productivity in the Ethiopian highlands. Appl. Water Sci. 2023;13:168. 10.1007/s13201-023-01968-5. [CrossRef] [Google Scholar]
72. Wang R., Cherkauer K., Bowling L. Corn response to climate stress detected with satellite-based NDVI time series. Remote Sens. 2016;8:269. 10.3390/rs8040269. [CrossRef] [Google Scholar]
73. 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. 10.3390/s21041406. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
74. Martin K.L., Girma K., Freeman K.W., Teal R.K., Tubaña B., Arnall D.B., Raun W.R. Expression of variability in corn as influenced by growth stage using optical sensor measurements. Agron. J. 2007;99:384–389. 10.2134/agronj2005.0268. [CrossRef] [Google Scholar]
75. Yin X., McClure M.A. Relationship of corn yield, biomass, and leaf nitrogen with normalized difference vegetation index and plant height. Agron. J. 2013;105:1005–1016. 10.2134/agronj2012.0206. [CrossRef] [Google Scholar]
76. Vian A.L., Bredemeier C., Drum M.A., Pires J.L.F., Fochesatto E. Vegetation sensors as a tool for plant population identification and corn grain yield estimation. Pesqui. Agropecuária Trop. 2021;51:e66926. 10.1590/1983-40632021v5166926. [CrossRef] [Google Scholar]
77. Towers P.C., Strever A., Poblete-Echeverría C. Comparison of vegetation indices for leaf area index estimation in vertical shoot positioned vine canopies with and without grenbiule hail-protection netting. Remote Sens. 2019;11:1073. 10.3390/rs11091073. [CrossRef] [Google Scholar]
78. Zhang F., Zhou G. Estimation of vegetation water content using hyperspectral vegetation indices: A comparison of crop water indicators in response to water stress treatments for summer maize. BMC Ecol. 2019;19:18. 10.1186/s12898-019-0233-0. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
79. Bi W., Wang M., Weng B., Yan D., Yang Y., Jinjie W. Effects of drought-flood abrupt alternation on the growth of summer maize. Atmosphere. 2019;11:21. 10.3390/atmos11010021. [CrossRef] [Google Scholar]
80. Huang M., Zhang Z., Sheng Z., Zhu C., Zhai Y., Lu P., Brinkman D. Soil salinity and maize growth under cycle irrigation in coastal soils. Agron. J. 2019;111:2276–2286. 10.2134/agronj2018.10.0684. [CrossRef] [Google Scholar]
81. Johansen K., Ziliani M.G., Houborg R., Franz T.E., McCabe M.F. Cubesat constellations provide enhanced crop phenology and digital agricultural insights using daily leaf area index retrievals. Sci. Rep. 2022;12:5244. 10.21203/rs.3.rs-1134595/v1. [Europe PMC free article] [Abstract] [CrossRef] [Google Scholar]
82. Costa T.S., Santos R.A.d., Santos R.L., Filgueiras R., Cunha F.F.d., Pereira A.J., Salles R.A.d. Spatio-temporal variability of biophysical parameters of irrigated maize using orbital remote sensing. Semin. Ciências Agrárias. 2021;42:2181–2202. 10.5433/1679-0359.2021v42n4p2181. [CrossRef] [Google Scholar]
83. Javed M.A., Ahmad S.R., Awan W.K., Munir B. Estimation of crop water deficit in lower bari doab, pakistan using reflection-based crop coefficient. ISPRS Int. J. Geo-Inf. 2020;9:173. 10.3390/ijgi9030173. [CrossRef] [Google Scholar]
84. Kang S., Gu B., Du T., Zhang J. Crop coefficient and ratio of transpiration to evapotranspiration of winter wheat and maize in a semi-humid region. Agric. Water Manag. 2003;59:239–254. 10.1016/s0378-3774(02)00150-6. [CrossRef] [Google Scholar]
85. Jia Q., Wang Y. Relationships between leaf area index and evapotranspiration and crop coefficient of hilly apple orchard in the loess plateau. Water. 2021;13:1957. 10.3390/w13141957. [CrossRef] [Google Scholar]
Articles from Plants are provided here courtesy of Multidisciplinary Digital Publishing Institute (MDPI)

Full text links 

Read article at publisher's site: https://doi.org/10.3390/plants13091212

Similar Articles 

To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.

    Funding 

    Funders who supported this work.

    European Union's Horizon 2020 "WATERAGRI Water retention and nutrient recycling in soils and streams for improved agricultural production" (1)

    • Grant ID: Grant Agreement No. 858375

      0 publications

    European Union’s Horizon 2020 “WATERAGRI Water retention and nutrient recycling in soils and streams for improved agricultural production” (1)

    • Grant ID: 858375

      0 publications

    János Bolyai Research Scholarship of the Hungarian Academy of Sciences