Elsevier

Agricultural Water Management

Volume 255, 1 September 2021, 107052
Agricultural Water Management

Applications of Gaussian process regression for predicting blue water footprint: Case study in Ad Daqahliyah, Egypt

https://doi.org/10.1016/j.agwat.2021.107052Get rights and content

Highlights

  • Four GPR kernels models were developed for modeling blue WF.

  • Ten scenarios based on variable elimination technique were developed as inputs to GPR models.

  • GPRPUK Model is better than the other three models in predicting blue WF followed by Poly-kernel.

  • Statistical errors have decreased and correlation index has increased by increasing number of meteorological variables.

Abstract

Timely and reliable water footprint prediction is imperative and prerequisite to mitigate climate risk and ensure water and food security and enhance the water-use efficiency. This study aims to model the Water Footprint (WF) by using the four kernels of Gaussian processes models (Polynomial, Normalized Poly, Radial Basis Function RBF, and Pearson Universal Function PUK) and select the best kernel with best climate scenario. This study investigates the predicting WF of maize based on meteorological variables including maximum temperature (Tmax), minimum temperature (Tmin), mean temperature (Tmean), solar radiation (SR), wind speed (WS), and vapor pressure deficit (VPD), Extraterrestrial radiation (Ra) relative humidity (RH) and crop coefficient (Kc) by applying ten scenarios of climate variables in the Egyptian Nile Delta, Ad Daqahliyah Governorate for predicting blue WF of maize during 2000–2019. The main findings are following as, firstly; based on developing four kernels, the performance of the PUK kernel in predicting blue WF is far better than the other three kernels followed by the Poly kernel. Secondly; for PUK kernel, model 7 (Tmax, Tmin, Tmean, WS, Sunshine Hours (SH), VPD and SR) has good performance which is close to models 8 (model 7 + Ra), model 9 (model 7 + Ra and RH) and model 10 (all inputs). Thirdly; in all four kernels, the error rate in small blue WF values is higher than the other values, moreover, the error value decreases at the medium blue WF values, while, it increases again at large WF values. Therefore, the developed models in this study can help and promote the decision makers to manage and secure the water resources management under the extreme climate events.

Introduction

Environmental challenges have attracted more attention in most countries all over the world as a result of unsustainable and inappropriate exploitation of environmental resources (Engström et al., 2008, Lenschow et al., 2016, Fan et al., 2018). The accelerated resource consumption and the constant abuse of nature, associated with rapid climate change have negatively affected ecosystems recovery potentials (Chowdhary et al., 2018, Duckett et al., 2016). On the other hand, acceleration of global economic growth led to environmental problems and retrogression of ecosystem services and functions (Han et al., 2021). In this context, the United Nations (UN) launched the UN-2030 Agenda and its sustainable development goals (SDGs) to meet the eminent earth’s challenges (UN, 2015). The SDGs are an ambitious plan to achieve earths sustainability within various dimensions (i.e. 17 SDGs) with a special attention to land (SDG-15), climate change (SDG-13), and water (SDG-6) (UN, 2015; Di Baldassarre et al., 2019). In a changing climate, water availability and its efficient management (SDG-6) is one of the urgent challenges that need to be addressed, due to its direct or indirect role in achieving other SDGs for insuring sustainability (El-mageed et al., 2017; Di Baldassarre et al., 2019; Jerin et al., 2021; Elbeltagi et al., 2020e, 2021; Kumar et al., 2021; Mokhtar et al., 2021; Bajirao et al., 2021; Zerouali et al., 2021; Adnan et al., 2021; Suryakant et al., 2021). Hence, proper management of water resources by using new approaches is a vital issue for achieving SDGs goals especially in arid and semi-arid regions. In this sense, water footprint (WF) (Hoekstra, 2008) could be one of the tools for meeting water security challenges, especially in the agricultural sector (Elbeltagi et al., 2020a).

The term WF, firstly introduced, is a method for quantifying the amount water that is used for producing one unit of a target product through the whole supply chain. Scientifically, the WF is a combination between green-WF (rainfall (R) (m3); blue-WF (evapotranspiration (ET) – R, when ET ≥ R) (m3); and gray-WF (water pollution/ quality – neglected-) (m3)) (Lovarelli et al., 2016, Jamshidi et al., 2020). Since, 85% of the global freshwater is consumed by agricultural sector (Mekonnen and Hoekstra, 2011, Bhat et al., 2017a, Bhat et al., 2017b, Elbeltagi et al., 2020a, Elbeltagi et al., 2020b, Elbeltagi et al., 2020c, Elbeltagi et al., 2020d, Elbeltagi et al., 2020e); the WF was broadly employed in the agriculture sector to calculate water consumption, environmental impact, and investigate the better water management (Novoa et al., 2019, Nezamoleslami and Hosseinian, 2020). In this sense, water footprint assessment (WFA) was previously applied in a global scale for maize (Liu et al., 2017), wheat (Mekonnen and Hoekstra, 2010), crop production (Babaee et al., 2021, Mekonnen and Hoekstra, 2011), hydropower, animal products (Mekonnen and Hoekstra, 2012), and humanity (Hoekstra and Mekonnen, 2012). Notably, WFA research was carried out in many parts of the world. For instance, in Morocco; China (Zhai et al., 2019); Romania (Ene et al., 2013); Italy (De Girolamo et al., 2019); South Africa (Pahlow et al., 2015); Iran (Ababaei and Etedali, 2017, Karandish and Hoekstra, 2017); Turkey (Muratoglu, 2019), and Spain (Pellicer-Martínez and Martínez-Paz, 2016).

So far, many models were developed for WFA such as Crop Water Use (CWU) Model (De Miguel et al., 2015); WFA approach (Hoekstra et al., 2011), WF model (Nezamoleslami and Hosseinian, 2020); and WF- HYDRUS (Karandish and Šimůnek, 2018). Recently, WFA was integrated with machine learning (ML) technique, however, holistic approach between WFA and machine learning is still emerging. For example; Dai et al. (2021) proposed a fuzzy modeling approach, which enhanced WF under different levels of credibility. Mokarram et al. (2021) used artificial neural networks (ANNs) for calculating WF, and reported that radial basis function (RBF) has better estimation than other models. Interestingly, Elbeltagi et al. (2020a) developed an ANN model for calculating WF in Egyptian Delta for maize, where the result showed a high trustworthy outcome with low deviation, even if the climate data were limited. Their findings observed that the combination of solar radiation, vapor pressure deficit, and humidity (H) was the best variables for predicting blue WF. Moreover, Elbeltagi et al. (2020b) studied the impact of climate changes on the future water footprint of two agro-nomically important crops by applying deep neural networks (DNN) under three scenarios of climate changes. In the proposed work, we have presented new techniques in modeling WF easily without conducting field experiments, no cost, simple and achieve satisfactory result. Gaussian processes under four kernels as a rapid decision tools were developed to model blue WF in the study area as a new tool for water users and developers. Furthermore, to our knowledge, there are no studies available in the literature that model and simulate the WF in the semi-arid regions using the processes of Gaussian kernels.

Egypt, is located in the northern Africa, suffers from water shortage, which reaches 13.5–24 Bm3 yr−1, due to arid climate, rapid population growth (94 million), and limited water resources (90% from Nile river (55.5 Bm3 yr−1)) (Allam et al., 2015, Mohie El Din and Moussa, 2016, El-Essawy et al., 2019, Assar et al., 2020). However, water scarcity is the main challenges facing the agricultural sector and limiting crop production and national food security (Allam and Allam, 2007, El-Essawy et al., 2019). Nevertheless, 50% of the labor forces working in agricultural sector contributed to about 23% of national GDP (El-Essawy et al., 2019). Thus, defining water consumption from main produced crops (i.e. maize, wheat, and cotton) is critical for efficient water management and sustainability of agricultural system (Elbeltagi et al., 2020a, Elbeltagi et al., 2020b, Elbeltagi et al., 2020c, Elbeltagi et al., 2020d, Elbeltagi et al., 2020e).

Maize (Zea mays L. spp.) is one of the important crops in Egypt (Abdrabbo et al., 2013). According to Food and Agriculture Organization (FAO) the cultivated area by maize is nearly about 1.1 M ha, which is slightly higher than year 2019 by 2.72%. Regardless, the efficiency of maize and other C4 crops in water use comparing to C3 crops (Begg and Turner, 1976); maize cultivation in Egypt and especially in Nile Delta required a careful assessment of water balance and water use efficiency for sustainable production. Nonetheless, few studies were carried out in the Mediterranean region (Novoa et al., 2019), and in Egypt -in particular- for estimation blue WF of agricultural crops. Thus, to bridge the gap in literature, this work aims to: (1) compare between the developed GPR models under different kernels in Egyptian Nile Delta Governorate, and (2) select the best model and algorithm, which achieve a high degree of accuracy and low error for predicting blue WF of maize.

Section snippets

Study area

The study field is on Egypt's Nile Delta. It covers the northern portion (Lower Egypt) of the Mediterranean Sea. The Nile River is the world's longest river, with a total length of about 7000 kilometers. The Nile Delta accounted for approximately 2% of Egypt's total territory, with up to 63% of agricultural property. The Delta started about 20 km north of Cairo and up to 150 km north. The width of the delta is about 250 km i.e. Alexandria governorate in the west and Port Said governorate in the

Results and discussion

In this study, different meteorological data were combined in 10 scenarios as inputs into the GPR to model the blue WF of maize. Performance of four different kernels (including Poly kernel, RBF kernel, PUK kernel and Normalized poly kernel) in GPR model are presented from Table 4, Table 5, Table 6, Table 7, respectively, for training and testing periods. As Table 4 shows, by increasing the number of meteorological variables in GPR in the Poly kernel, the amount of error (in all 4 indexes of

Conclusion

In this study, we have developed four kernels in GPR models based on ten scenarios from meteorological variables including (Tmax, Tmin, Tmean, WS, SH, VPD, SR, Ra,H and Kc). In order to select the best kernel with scenario for predicting blue WF of maize in Egyptian Nile Delta Governorate during the period from 2000 to 2019. Overall, the performance of the PUK kernel in predicting blue WF is the best kernel followed by the Poly kernel, the Normalized poly kernel, and finally the RBF kernel. For

CRediT authorship contribution statement

Ahmed Elbeltagi: Conceptualization, Formal analysis, Writing - original draft, Writing - review & editing. Nasrin Azad: Writing - review & editing. Arfan Arshad: Writing - review & editing. Safwan Mohammed: Writing - review & editing. Ali Mokhtar: Writing - review & editing. Chaitanya Pande: Writing - review & editing. Hadi Ramezani Etedali: Writing - review & editing. Shakeel Ahmad Bhat: Writing - review & editing. Abu Reza Md. Towfiqul Islam: Writing - review & editing. Deng Jinsong:

Funding

This work was supported by the Ministry of Science and Technology of the People's Republic of China (2016YFC0503404) and the Natural Science Foundation of Zhejiang Province (LY18G030006).

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (79)

  • A. Elbeltagi et al.

    Modeling long-term dynamics of crop evapotranspiration using deep learning in a semi-arid environment

    Agric. Water Manag.

    (2020)
  • A. Elbeltagi et al.

    The impact of climate changes on the water footprint of wheat and maize production in the Nile Delta, Egypt

    Sci. Total Environ.

    (2020)
  • A. Elbeltagi et al.

    Modeling monthly crop coefficients of maize based on limited meteorological data: a case study in Nile Delta, Egypt

    Comput. Electron. Agric.

    (2020)
  • S.A. Ene et al.

    Water footprint assessment in the winemaking industry: a case study for a Romanian medium size production plant

    J. Clean. Prod.

    (2013)
  • R. Engström et al.

    Which environmental problems get policy attention? Examining energy and agricultural sector policies in Sweden

    Environ. Impact Assess. Rev.

    (2008)
  • B. Han et al.

    An integrated evaluation framework for Land-Space ecological restoration planning strategy making in rapidly developing area

    Ecol. Indic.

    (2021)
  • F. Karandish et al.

    An application of the water footprint assessment to optimize production of crops irrigated with saline water: a scenario assessment with HYDRUS

    Agric. Water Manag.

    (2018)
  • W. Liu et al.

    Towards improvement of grey water footprint assessment: with an illustration for global maize cultivation

    J. Clean. Prod.

    (2017)
  • D. Lovarelli et al.

    Water Footprint of crop productions: a review

    Sci. Total. Environ.

    (2016)
  • B.P. Malone 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)
  • S. Mehdizadeh et al.

    Implementing novel hybrid models to improve indirect measurement of the daily soil temperature: Elman neural network coupled with gravitational search algorithm and ant colony optimization

    Measurement

    (2020)
  • A. Muratoglu

    Water footprint assessment within a catchment: a case study for Upper Tigris River Basin

    Ecol. Indic.

    (2019)
  • G.S. Nearing et al.

    Information loss in approximately Bayesian estimation techniques: a comparison of generative and discriminative approaches to estimating agricultural productivity

    J. Hydrol.

    (2013)
  • R. Nezamoleslami et al.

    An improved water footprint model of steel production concerning virtual water of personnel: the case of Iran

    J. Environ. Manag.

    (2020)
  • V. Novoa et al.

    Understanding agricultural water footprint variability to improve water management in Chile

    Sci. Total. Environ.

    (2019)
  • F. Pellicer-Martínez et al.

    Grey water footprint assessment at the river basin level: accounting method and case study in the Segura River Basin, Spain

    Ecol. Indic.

    (2016)
  • A. Shalaby 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)
  • R. Taormina et al.

    ANN-based interval forecasting of streamflow discharges using the LUBE method and MOFIPS

    Eng. Appl. Artif. Intell.

    (2015)
  • C.L. Wu et al.

    Prediction of rainfall time series using modular soft computing methods

    Eng. Appl. Artif. Intell.

    (2013)
  • C. Xinchun et al.

    Water footprint assessment for crop production based on field measurements: a case study of irrigated paddy rice in East China

    Sci. Total Environ.

    (2018)
  • Y. Zhai et al.

    Impact-oriented water footprint assessment of wheat production in China

    Sci. Total Environ.

    (2019)
  • M.A.A. Abdrabbo et al.

    Evaluation of CSM-Ceres-Maize Model for simulating maize production in Northern Delta of Egypt

    Life Sci.

    (2013)
  • J.T. Abatzoglou et al.

    TerraClimate, a high-resolution global dataset of monthly climate and climatic water balance from 1958-2015

    Sci. Data

    (2018)
  • R.M. Adnan et al.

    Novel ensemble forecasting of streamflow using locally weighted learning algorithm

    Sustainability

    (2021)
  • M.N. Allam et al.

    Water resources in Egypt: future challeges and opportunities

    Water Int.

    (2007)
  • W. Assar et al.

    Effect of water shortage and pollution of irrigation water on water reuse for irrigation in the Nile delta

    J. Irrig. Drain. Eng.

    (2020)
  • M. Babaee et al.

    Artificial intelligence approach to estimating rice yield*

    Irrig. Drain.

    (2021)
  • T.S. Bajirao et al.

    Superiority of hybrid soft computing models in daily suspended sediment estimation in highly dynamic rivers

    Sustainability

    (2021)
  • S.A. Bhat et al.

    Water requirements and irrigation scheduling of maize crop using CROPWAT model

    Int. J. Curr. Microbiol. Appl. Sci.

    (2017)
  • Cited by (0)

    View full text