Modeling monthly crop coefficients of maize based on limited meteorological data: A case study in Nile Delta, Egypt
Introduction
The increase in population and water demand makes water resource management increasingly important, especially in North Africa and the Middle East, which are categorized as semi-arid and arid regions. These regions are expected to suffer water shortages in the coming decades. Egypt is one in the arid regions that is facing a high risk of water deficit because of the synergy of persistent drought and the increased water demand, especially in agricultural irrigation sector (Farg et al., 2012). Thence, the estimation of crop water needs plays an important role for improving crop water use efficiency and saving water amounts (Pôças et al., 2015). Crop evapotranspiration which means crop water needs is controlled by climatic and crop conditions. Most of the existing crop models are non-spatial; these models depend on point data of the reference evapotranspiration and crop coefficients from the standard database (Doorenbos and Pruitt, 1977).
The crop coefficient (kc) is an important factor for the development of irrigation water schedules (Hong et al., 2017), is primarily subject to the dynamics of canopy cover, leaf area index and greenness degree (Farg et al., 2012). Crop coefficients represent the unique differences for the reference and studied crops in terms of crop evapotranspiration (ETc = ETo × kc) according to Allen et al. (1998). In this context, Guerra et al. (2015) declared a linear relationship between ETo and values of midseason kc with determination coefficient (R2 = 0.87). Much less known that, the FAO CROPWAT Model can calculate the monthly value of crop coefficient indispensably using comprehensive and numerous data, and adopts standard values: 0.30, 1.20, and 0.35 for initial, mid, and late stages at humidity of 45% and wind speed of 2 ms−1. Actually, kc needs an adjustment according to climatic conditions (Allen et al., 1998). The values of crop coefficient that used from previous studies could provide feasible guidelines for water scheduling, but a considerable error in determining crop water needs may occur due to climate change and soil conditions (Jagtap and Jones, 1989). Therefore, it becomes necessary to make the corrections in values of crop coefficient under changing climatic conditions (Corbari et al., 2017, Gontia and Tiwari, 2010, Guerra et al., 2015, Irmak and Specht, 2013, Tyagi et al., 2000).
Many approaches have been used to calculate crop coefficient through different growth stages. Kamble et al. (2013) applied remote sensing method and found linear regression between normalized difference vegetation index (NDVI) with estimated kc and measured kc with R2 = 0.91 and 0.90, and RMSE were 0.16 and 0.19, respectively. Also, there are a number of methods for calculating crop coefficients such as a hydraulic weighing lysimeter (Abedinpour, 2015, Adeogun and Ogunlela, 2013, Shahrokhnia and Sepaskhah, 2013), FAO dual kc methodologies (Üzen et al., 2018), water balance (Trout and DeJonge, 2018), mathematical equations based on reference evapotranspiration (ETo), wind speed, humidity, crop height and soil moistening events (Silva et al., 2017), and watermarks and an atmometer (Anwer et al., 2016). Furthermore, Tyagi et al. (2000) found the estimated kc values were 11.6–74.2% higher than the values suggested by FAO Penman-Monteith data. These methods produced acceptable outcomes for predicting kc but it consumes much time, cost and efforts for the executing.
In this study, an artificial intelligence method was used to modeling crop coefficients in the selected Egyptian governorates by using limited number of variables, two or three only of climate data and to evaluate the performance of model outputs with estimated CROPWAT kc values. Artificial intelligence models are characterized by the initiation function, which uses interrelated information processing units to transform input into output by detecting relationships and pattern in data. These neural networks are considered to be the best procedures for extracting information from imprecise and non-linear data (Adisa et al., 2019). ANNs have been applied for many agricultural purposes, e.g., yield prediction (Akhand et al., 2018), who declared that the predicted yields are very close to the government-led statistical yields with an error of prediction less than 8%, irrigation scheduling based on soil moisture content (Adeyemi et al., 2018), who showed that the predictive system achieves a water saving ranging between 20 and 46% with high yield and water use efficiency, estimation of crop reference evapotranspiration (Nema et al., 2017), who found the optimal ANNs had correlation coefficients of 0.99 and 0.98 for the calibration and validation periods, respectively, crop cost forecasting (Mohan and Patil, 2017), who stated that the ANN achieved around 0.2–0.5% improvement than the previous approaches, and crop water requirement prediction (Khan et al., 2011), who found that the ANN achieved an accuracy of 95% between actual and predicted water usage. Therefore, modeling kc is an attractive research point for the accurate estimation of crop water-use with reference evapotranspiration.
Maize crop was selected in this study because it is one of the most important grain crops grown principally during the summer season in Egypt. Maize has a wealth of value because it can be used as a source for fructose, corn oil, and starch. Egypt imported 4,821,200 ton per year of maize as average over the period from 2000 to 2013, which represents around 28.1% of the total value of the imports of plant products (Yassin et al., 2015). With consideration with data availability, cost, and complexity for feasibly calculating kc, the objective of this study is to: (1) present an artificial neural network (ANN) model for predicting the monthly values of kc through growing season using data of two or three climate variables only in the major maize producing governorates in Egypt, and (2) compare the outputs of superior ANNs model with actual kc from FAO CROPWAT Model.
Section snippets
Study area
The study area lay on Nile Delta in Egypt. The Egyptian Nile delta is situated in northern Egypt, where the river Nile reaches the Mediterranean Sea. The Nile River originates northward from the equator and is the longest river in the world with 7000 km long. The Nile Delta covers only about 2% of Egypt's area but comprises nearly 63% of its agricultural land. The Delta begins approximately 20 km north of Cairo and extends north for about 150 km. At the coast, the delta is about 250 km wide,
Optimization combinations of independent variables selection
There are two methods for selecting the best climate variables in ANNs for predicting kc: dimensionality reduction algorithm such as a principle component analysis (PCA) that identify patterns in data based on the correlation between features, select the best features of variables as inputs to ANNs, and the second approach depends on training and testing various variable combinations in the ANNs as followed in the study, to arrive at the optimal or best combination with high accuracy,
Conclusions
Crop evapotranspiration (ETc) plays a vital role in evaluating the performance of water management strategies for improving water use, especially in arid and semi-arid regions such as Egypt. To make a precise irrigation schedules, crop coefficients (kc) consider an important factor for calculation ETc. This work aimed to evaluate and model the values of monthly crop coefficients of maize by using artificial neural networks (ANNs) method. The climatic data variables from 2006 to 2016 were
Author contributions
Ahmed Elbeltagi and Jinsong Deng had the original idea for the study and all coauthors conceived and designed the study. Ahmed Elbeltagi analyzed data and wrote the paper, which was revised by Jinsong Deng; and all authors read and approved the final manuscript.
Acknowledgement
This work was supported by Zhejiang Provincial Natural Science Foundation of China [grant number LY18G030006].
Cordial thanks are extended to the editor and two anonymous reviewers for their valuable comments.
Declaration of Competing Interest
The authors declared that there is no conflict of interest.
References (39)
- et al.
Estimation of Evapotranspiration ETcand Crop Coefficient Kcof Wheat, in south Nile Delta of Egypt Using integrated FAO-56 approach and remote sensing data
Egypt. J. Remote Sens. Sp. Sci.
(2012) - et al.
Digital soil mapping of soil carbon at the farm scale: A spatial downscaling approach in consideration of measured and uncertain data
Geoderma
(2017) - et al.
Reference evapotranspiration estimation and modeling of the Punjab Northern India using deep learning
Comput. Electron. Agric.
(2019) - et al.
Remote sensing and GIS for mapping and monitoring land cover and land-use changes in the Northwestern coastal zone of Egypt
Appl. Geogr.
(2007) - et al.
TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958–2015
Sci. Data
(2018) Evaluation of growth-stage-specific crop coefficients of maize using weighing lysimeter
Soil Water Res.
(2015)- et al.
Determination of crop coefficients of maize for the estimation of crop water use
Cont. J. Eng. Sci.
(2013) - et al.
Dynamic neural network modelling of soil moisture content for predictive irrigation scheduling
Sensors (Switzerland)
(2018) - et al.
Application of artificial neural network for predicting maize production in South Africa
Sustain.
(2019) - et al.
Wheat yield prediction in Bangladesh using artificial neural network and satellite remote sensing data
Glob. J. Sci. Front. Res. D Agric. Vet.
(2018)
Crop evapotranspiration - Guidelines for computing crop water requirements. Irrig. Drain. Pap. No. 56
Estimating evaporation from bare soil and the crop coefficient for the initial period using common soils information
J. Irrig. Drain. Eng.
Predicting the crop coefficient values for maize in Iraq
Eng. & Tech. J.
Assessing crop coefficients for natural vegetated areas using satellite data and eddy covariance stations
Sensors (Switzerland)
Guidelines for predicting crop water requirements
Estimation of crop coefficient and evapotranspiration of wheat (Triticum aestivum) in an irrigation command using remote sensing and GIS
Water Resour. Manage.
Correcting midseason crop coefficients for climate
J. Irrig. Drain. Eng.
Determination of growth stage-specific crop coefficients (Kc) of sunflowers (Helianthus annuus L.) under salt stress
Water (Switzerland)
Hourly and daily single and basal evapotranspiration crop coefficients as a function of growing degree days, days after emergence, leaf area index, fractional green canopy cover, and plant phenology for Soybean
Trans. ASABE
Cited by (36)
A crop variety yield prediction system based on variety yield data compensation
2022, Computers and Electronics in AgricultureCitation Excerpt :The period from 80 to 95 corresponds to the late stage of grain-filling; this stage involves the accumulation of maize dry materials, and meteorological factors directly determine the plumpness of maize seeds. Water, temperature, and sunshine substantially influence maize yield (Ahmed et al., 2020; Hadi and Mojgan, 2021). For example, in general, when the temperature difference between day and night is large, ventilation is good, humidity is moderate, and sunshine is sufficient, the maize yield is relatively high.
Hybrid deep learning techniques for estimation of daily crop evapotranspiration using limited climate data
2022, Computers and Electronics in AgricultureAn atmospheric and soil thermal-based wheat crop coefficient method using additive crop growth models
2022, Agricultural Water ManagementCitation Excerpt :To further verify the performance of the proposed method, the prediction results of other five methods (RDT-, ANN-, GDDE- and GDDA-based Kc method and the stage constant model) were also given in Fig. 11. Details of these five methods were shown in Supplementary Table 2, referencing to Meng et al. (2021), Dingre and Gorantiwar (2020), Elbeltagi et al. (2020), Martel et al. (2018), and Detar (2004). As shown in Fig. 11, there were certain differences in the prediction accuracy between different methods.
Predicting individuals' car accident risk by trajectory, driving events, and geographical context
2022, Computers, Environment and Urban Systems