Assessing the feasibility of applying remote sensing technique in calculating water consumption

Irrigation water is limited and scarce in many regions of the world. Egypt has limited water resources with a gradual increase in water demand due to the rapid increase of the population. The gap between water supply and water demand is currently reaching 20.0 BCM/year. This gap is currently filled through the reuse of water losses and any more reduction in the resources will have a serious impact on the sustainability of agriculture.

For the previous reasons, the precise calculation of crop water consumption (WC)/requirement is essential for effective irrigation management and water conservation. In addition, the precise calculation of crop WC/requirement and in consequence, the irrigation water use could reduce the necessity for desalination and industrial wastewater treatment, which could be associated with a huge cost.

Crop evapotranspiration (ET_C) represents crop water requirements and is affected by weather and actual crop conditions. According to FAO irrigation and drainage, paper No.24 and No.56 ET_C can be obtained from reference crop evaporation (ET₀) and using a stage-dependent crop coefficient (K_C) from the relation (ET_C = ET₀*K_C). Calculating KC values is currently depending on collecting the cropping pattern manually. Such an approach is adversely affected by many factors such as the fragmentation of property and the difference in planting and harvesting times, which makes the accurate calculation of water requirements in this way practically difficult. Water Management Research Institute (WMRI) has previous experience in calculating the discharge required at the head of different canals (Matching Model), based on collecting cropping pattern data from the agricultural departments and using the Penman–Monteith equation and FAO K_C values. The model faced problems in its application, and one of the main problems was the inaccuracy of the collected cropping pattern data. The work with this approach could be achieved at the field level, and it is hard to use this approach for forecasting and water management. Therefore, remote sensing is the best technique to calculate all these parameters at the command area level (Bellvert 2018). This technology is attractive for modeling K_C since it provides a synoptic coverage at fixed time intervals and can therefore monitor changes over time (Rozenstein et al. 2018). Remote sensing data are useful for monitoring and mapping vegetation cover changes and have been used extensively for identifying, assessing, and mapping such changes in different regions (Almalki et al. 2022). The normalized difference vegetation index (NDVI) was one of the most successful attempts to simply and quickly identify vegetated areas and their condition. Studies have demonstrated that NDVI is effective to estimate various vegetation properties, including the leaf area index (LAI) (Tian et al. 2017), biomass (Zhu & Liu 2015), chlorophyll concentration in leaves (Guzman et al. 2015), plant productivity (Vicente-Serrano et al. 2016), fractional vegetation cover (Dutrieux et al. 2015), and plant stress (Chavez et al. 2016). Several studies have been established for local regression functions between NDVI and crop coefficient value (K_C). Using NDVI to calculate different agricultural elements including crop coefficients is a simple and considerably reliable way to calculate water requirements and in consequence, it could improve water management (Tan et al. 2021). Numerous models have been proposed for estimating evapotranspiration by using remotely sensed data, e.g., the Surface Energy Balance Algorithm for Land Model (SEBAL), Mapping Evapotranspiration at High Resolution with Internalized Calibration (METRIC), the Simplified Surface Energy Balance Index (SSEBI), and the Surface Energy Balance System Model (SEBSM). Garcia-Santos et al. (2022) reviewed different models, the validation of use for each model, and the advantages and disadvantages of each model. They also performed modifications and improvements for these models. The authors concluded that the new trend of using hybrid ET models, which raised over the recent 5 years is of key importance for best ET retrievals. Results showed that three-source Surface Energy Balance (SEB) models can be of great utility for irrigation management of patchy surfaces as a vineyard (Levine et al. 2022). With the progress of developing new techniques for using remote sensing techniques, work is being carried out on the development of increasingly higher resolution multispectral thermal sensors which will allow further progress in the applications of this parameter in the estimation of crop water requirements, among other applications. Different international models were developed for the calculation of water requirements elements including reference evapotranspiration and crop coefficients. The famous model might be the CROPWAT model. CROPWAT 8.0 for Windows software is developed by FAO with the help of the National Water Research Center Egypt and the Institute of Irrigation Development Studies of Southampton of UK. CROPWAT windows calculate crop evapotranspiration using the FAO-56 Penman–Monteith equation.

The main objective of this study is to investigate the feasibility of applying remote sensing techniques in the field of water management. WC will be calculated from satellite image information. The study will first compare a few equations for the relation between NDVI and K_C. The results were validated and the best equation was used to define the average crop coefficient value for the irrigation district. The calculated WC values will be checked through field measurements. Most of the previous remote sensing studies were applied at the field level. The current study aims to use the techniques to operate an irrigation system by calculating water requirements for each branch canal.

To perform the previous steps, different programs were used including Arc GIS 10.8 for processing the data and developing the shape files, CROPWAT 8.0 for calculation reference evapotranspiration, and MATLAB-based program to calculate crop coefficient and WC values. The SPSS program was used to develop stage-flow relationships to calculate water supply and drainage runoff to calculate the actual WC.

In this study, a series of sentinel-2 image data was downloaded for the summer season of 2019 as field measurements were done during the period from May to September 2019 (Figure 3). These images were downloaded from the USGS website https://earthexplorer.usgs.gov/sentinel data and are available at the website for free download. The images were selected to be almost cloud-free.

Sentinel-2 contains 12 bands and calculating NDVI depends on Bands 4 (red) and 8 (near-infrared). Wave lengths were 665 and 842 m for the two bands, respectively. The pixel size was 10 × 10 m for both bands.

NDVI is calculated for different pixels using Arc Map program as described before.

The current study verified different equations to select the one to be used for calculating K_C from NDVI values. The investigated equations were

• The first equation was developed by Roch et al. (2012), and it has the form:

  

  (2)
• The second equation was developed by Badr et al. (2017) and it has the form:

  

  (3)
• The third equation was developed by Kamble et al. (2013) and it has the form:

  

  (4)

Six satellite images were downloaded and KC values were calculated using the three different equations for each image. Water supply and runoff values were calculated as the average of 5 days around each satellite image date.

Actual WC values were calculated as the difference between water supply and drainage runoff and these values were used to verify the calculated WC values.

After testing different equations, the best equation (Equation (2)) was used to calculate WC and irrigation efficiency for an irrigation district. This was an example of using remote sensing techniques in the water management field. On the other hand, this was a second verification for the equation as calculated WC was compared with the difference between total water supply and total drainage runoff.

After testing the equation, and verifying the best equation at district irrigation level, the equation was used to define overall irrigation efficiency at the irrigation district level. The values were calculated as a ratio between water demand (calculated WC values) and water supply at the head of the district.

The previous results have two main parts: the first part is the selection of a suitable equation for the relationship between NDVI and K_C. Three different empirical equations were tested by using them to calculate WC for five different periods and compare the results with field measurements. RMSE and NOF were calculated for the three equations, and there was a significant difference between one of them and the other two. It should be noted that the winner equation was developed as a general equation for different crops, which was the case in this study as it calculates the average K_C for an irrigation district. The other two equations were developed for specific crops.

The second part of the study used the selected equation for calculating WC and irrigation efficiency for the El-Bostan irrigation district, which could be considered as a second verification for the equation as well as an example of using remote sensing techniques in water management. WC was calculated for the entire district and the results were compared with the difference between water supply and drainage runoff, which is considered as actual WC. At the same time, calculated WC values were compared with the total water supply to calculate the average irrigation efficiency for the district. The results for both items supported the accuracy of calculating WC based on the remote sensing technique. The results were compatible with the results of Kharrou et al. (2021) who conducted that irrigation managers would likely appreciate using operational tools based on the remote sensing information, which could allow for a reliable assessment of irrigation delivery performance at the scale of irrigation schemes and provide an enhanced irrigation advisory service by delivering customized and near real-time information on crop development status and irrigation needs to farmers or Water Users Associations. These tools enable the improvement of irrigation scheduling for efficient and profitable agricultural water management, particularly in water-limited areas.

The current study aimed to investigate the feasibility of applying remote sensing techniques in the water management field by using satellite images to calculate WC values to use the calculated values to define the required water supply at the heads of different irrigation canals.

The current approach of calculating WC values based on collecting the cropping pattern manually and using fixed crop WC values is adversely affected by many factors that diminish its accuracy. These factors include land fragmentation, stopping the crop cycle, and the difficulty of collecting cultivation dates and crop life cycles. The suggested technique provides a better alternative for calculating WC, which could positively improve water management and, as a consequence water use efficiency. The previous results presented the calculated remote sensing-based WC. The study applied three different empirical equations that connect NDVI with crop coefficient value (K_C). Field measurements were performed to calculate the water supply at the head of the main canal and the outlets of the main drains, which gave the chance to calculate actual WC accurately and therefore use such WC data to test the three empirical equations. RMSE, NOF, and correlation efficiency were calculated for different equations and it was concluded that one of these equations (Equation (2)) is outperforming other equations. In this equation, RMSE was 0.68, NOF was 0.042, and the correlation coefficient value was 0.92. The selected equation, which was developed as a general equation for different crops was used to calculate average monthly irrigation efficiency values for the El-Bostan irrigation district. Average monthly irrigation efficiency values were also calculated based on sentinel 2 WC values and water supply at the head of the district. The values, which were between 86% in August 2019 and 94% in May 2019, were consistent with the expected situation in the district as a modern irrigation system is applied in the entire district and drainage runoff values were considerably low compared to water supply values.

The current study is important due to the following:

• The current situation in Egypt with limited water resources requires calculating water requirements precisely.

• There is difficulty in calculating WC based on the traditional technique due to land fragmentation and the diversity of the cropping pattern and cultivation dates.

• The remote sensing technology provides a suitable alternative that can overcome the difficulties of the previous approach by calculating crop coefficient values based on NDVI.

The study selected the proper empirical equation after testing different equations. The selected equation was verified based on actual measurements at the irrigation district level, and then it was used to define the irrigation efficiency for this district. The results showed a strong relation between calculated and measured values, which refer to the accuracy of calculated WC values and the reliability of this approach, and its ability to overcome the disadvantages of the current approach that depend on collecting cropping pattern manually.

It is recommended to use the current approach as the main tool to distribute water supply between different irrigation canals, by developing a model that incorporates remote sensing-based WC values with other information to define the required water supply at the head of different branch canals and different control structures.

The suggested model could be a remarkable step in improving water distribution in the Egyptian irrigation system, and it could be the main step to sustaining agriculture in Egypt with limited water resources.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.