3.1 Water distribution uniformity
These experiments were conducted in the sprinkler irrigation system under three operating conditions and climate parameters to reach the highest possible distribution uniformity as much energy saving as possible. For the three types of sprinklers, three different working pressures, three different discharge volumes and three different rise heights are applied (KA-4, FOX, 2520) which are described in (Table 1). The major factors affecting uniformity of water distribution are type of sprinkler nozzle, operating pressure, diameter for nozzle sprinkler, material of nozzle sprinkler, temperature and moisture 43, also, include wind speed, wind direction and sprinkler nozzle specifics are affecting factors 44. For KA-4 sprinkler, the highest CU value was 79% in the square system under operating pressure of 200kPa, riser height of 0.5m and the discharge was 2.5 m3/h (Nozzle 4.2mm) (Fig. 1a). However, it was 81.9% in the triangular system under the same condition but on the 1m height. Maybe the reason of the low CU at 1m back to the maximum and minimum temperature reached to 89.7 and 78.5˚F, respectively and the wind speed was 2m/s (Fig. 3a and Table 3). furthermore, when the operating pressure 150kPa was applied, the CU decreased by 10.97% in comparison with using a pressure of 200kPa in the same height in the square system. Moreover, when applied 150kPa in comparison with applying 200kPa in the triangle system, the CU decreased by 27.3%, although, in most cases the CU increased under 250kPa, except for one experiment in which the value of CU decreased, when using a riser of 1m height. The reason for this is due to the high maximum temperature which reached 90.1˚F and wind speed was 2m/s, as the high temperature lead up to the evaporation of the water collected in the cans used in the experiments, accordingly, there was a difference in the values of the water collected in the catch cans, and the high wind speed lead up to move some water from cans to others, and thus it affects the uniformity of water distribution. Also, the square system agreed with the triangle system in all cases except when using a height 0.5m, this could be because of strong winds was 3m/s, and temperature was 85.2˚F. It is noted in the square system, when operating at a pressure of 250kPa, the CU increased by an average rate was 10.65%, while, it was 3.4% in the triangular system. However, the system consumes more energy at 250kPa, in comparison with operating at 200kPa, that agree with, the findings with 45 who mentioned that CU increased with the increased operating pressure until it reached up to 250kPa. Further, the CU values at 200kPa in the triangular system, are larger than the square system, so it is preferable to use the triangle system.
On the other hand, the highest CU was 72.7% and 73.4% of the FOX sprinkler for the square and the triangular system, respectively under 200kPa, 0.5m height and 1.47m3/h discharge (Fig. 1b), also, the climate variable was closed in the two system during the experiment period (Fig. 3b). In the square system, the temperature was range between (89.1–87˚F), the relative humidity was 60% and wind speed was 3m/s, and for the triangular system, the temperature was range between (90.8–87.5˚F), the relative humidity and wind speed was 53% and 3m/s, respectively (Fig. 3b and Table 3). The values of CU when using operating pressures of 200 and 250kPa were ranged from 70.17–86.78% and 72.84–86%, respectively, The results are correspond with those obtained at 18, which were 73.27 and 77.86%.
On the other hand, the system consumes more energy when applied the high operating pressure of 250kPa, where it gives the highest CU values. When applied 250kPa in comparison with applying 200kPa, the CU increased in the square system by 2.28%. Moreover, in the triangular system the CU increased in most cases, except for one experiment in which the value of CU decreased by 4%, which may be due to the high wind speed (2m/s) when applied 250kPa in comparison with applying 200kPa. When the operating pressure was 150kPa, the CU values decreased, while when applying an operating pressure 200kPa in all experiments, the CU values increased and the system does not take high energy. There is an opposite association between the nozzle height and wind affect and so with drop diameter, the drop diameter under spinner type is smaller than of those for rotating type so we can see that the maximum CU and DU were agreed under 200 kPa operating pressure for both types 46.
Table 3
water distribution uniformity and wind speed (m/s) under the square and triangle sprinklers system
Observation | Sprinkler type |
KA4 | FOX | 2500 |
CU | DU | WS | CU | DU | WS | CU | DU | WS |
| | | | | Square | | | | |
1 | 70.68 | 62.32 | 3 | 72.81 | 63.73 | 2 | 83.92 | 82.12 | 1 |
2 | 79.35 | 72.96 | 2 | 72.74 | 68.45 | 3 | 86.78 | 84.67 | 0 |
3 | 76.60 | 73.24 | 3 | 73.09 | 61.82 | 1 | 82.90 | 77.24 | 0 |
4 | 69.10 | 68.32 | 2 | 63.59 | 54.21 | 2 | 84.43 | 81.73 | 3 |
5 | 71.42 | 62.16 | 2 | 70.18 | 59.03 | 2 | 85.24 | 83.91 | 3 |
6 | 79.88 | 72.59 | 3 | 74.59 | 63.73 | 2 | 85.58 | 82.06 | 3 |
7 | 69.58 | 66.41 | 1 | 67.73 | 61.76 | 2 | 83.86 | 81.17 | 4 |
8 | 70.09 | 65.00 | 1 | 72.57 | 62.41 | 3 | 86.03 | 82.49 | 4 |
9 | 78.59 | 72.50 | 2 | 72.84 | 61.30 | 3 | 86.00 | 82.24 | 4 |
Triangle |
1 | 63.24 | 54.12 | 3 | 68.67 | 55.05 | 2 | 87.64 | 85.29 | 1 |
2 | 73.47 | 63.35 | 3 | 73.49 | 80.25 | 3 | 86.60 | 84.03 | 1 |
3 | 77.04 | 68.28 | 2 | 77.36 | 80.15 | 3 | 83.81 | 79.69 | 0 |
4 | 71.88 | 62.85 | 2 | 67.79 | 57.89 | 2 | 87.26 | 85.43 | 1 |
5 | 81.98 | 77.08 | 3 | 70.31 | 58.80 | 3 | 87.38 | 85.45 | 1 |
6 | 76.54 | 67.24 | 2 | 77.26 | 66.46 | 3 | 87.81 | 84.29 | 1 |
7 | 69.79 | 59.68 | 2 | 61.18 | 47.20 | 2 | 89.22 | 90.93 | 3 |
8 | 78.63 | 71.56 | 3 | 70.11 | 59.57 | 2 | 85.60 | 83.01 | 1 |
9 | 80.46 | 74.36 | 2 | 66.03 | 55.25 | 2 | 85.14 | 82.75 | 2 |
Therefore, it is preferable to operate the system by 200kPa. The lowest value of the CU was 63.5% and 61% in the square and triangular system (Fig. 1b), respectively at 150kPa, a discharge of 1.27m3/h (Nozzle 3.8mm) and a riser height of 1 m in the square system but 1.5m in the triangular system. But, the CU values at 200kPa in the triangular system, are larger than the square system, so it is preferable to use the triangle system.
On the other hand, the highest CU of the 2520 sprinkler under 200kPa and 0.5m height was 86.7% by applying a discharge of 0.855m3/h (Nozzle 2.5mm) in the square system, and for the triangular system it was 87.3% at the same pressure and discharge, and 1 m height (Fig. 1c), also, the climate variable was in the square system, the maximum temperature was 78.8˚F, the minimum was 76.8˚F, the relative humidity was 64% and wind speed was 0m/s and for the triangular system, the temperature were range between (86.6–88.3˚F), the relative humidity was 55% and wind speed was 1m/s )Fig. 3c and Table 3). Although the lowest CU value in the two systems at 250kPa, a riser height 0.5m, and a discharge of 0.999m3/h, was 82.8% and 83.8% in the square system and the triangular system, respectively (Fig. 1c). The CU decreased in the square system by 2.2% and 1.86%, when applied 150kPa, in comparison with applying 200kPa or 250kPa, respectively but, it does not affect the CU value, therefore it is preferred applied by 200kPa in the square system, where, the system does not consume high energy. The CU was increased when the pressure increased for all the nozzles 2. In the triangle system, the CU increased by 1.86% and 4% was applied 150kPa, in compared to applying 200kPa with a riser height 0.5 or 1.5m, but it is small increase does not affect in the CU, while when using a riser height 1 m, the CU was equal at the three pressures, therefore it is preferable to use the triangle system applied by 200kPa.
Further, the DU increased by 9.6% and 2.3% for the KA-4 sprinkler, when was applied 150kPa in the square system in comparison with applying 200kPa when using a 1m or 1.5m riser, which (Fig. 2a), may be due to low the maximum and minimum temperature at 150 kPa in compared to 200kPa where was (83.5, 85.7˚F) (87.5, 89.7˚F), respectively for a 1m riser and (79.1, 79.1˚F) (79.7, 81.1˚F), respectively for a 1.5m riser, but, the percentage increase is small it does not affect in the DU value, although, when using a riser 0.5m, the DU decreased was applied 150kPa in comparison with 200kPa, the reason for this is due to the high temperature maximum and minimum (85.5, 88˚F), (81.4, 83.2˚F), respectively. which is in agreement with the DU values were low at 0.5m riser at pressures of 62 and 82KPa and then increased rapidly with increasing pressure 4. On the other hand, the DU decreased when applying 150kPa in compered to apply 200, 250kPa in the tringle system, moreover, the difference when was applied 200kPa and 250kPa was don’t affect in the DU value. On the other hand, for the FOX sprinkler in the square and tringle system the DU increased when applying 200kPa in comparison with applying 250kPa in the most cases (Fig. 2b), which may be due to low the maximum and minimum temperature at 150kPa, therefore, it is preferred applied by 200kPa in both systems. In addition to, the DU for 2520 sprinkler in the square system increased when applying 200kPa in compared to applied 150kPa, which, may be due to low wind speed in some cases and the other cases may be because of low the maximum and minimum temperature at 200kPa. Further, when the operating pressure was 200kPa in comparison with using operating pressure 250kPa the DU increased by 5% at 0.5m and 1m (Fig. 2c). which, it is in agreement with the DUlq increased as the pressure increased, DUlq values were low with the riser 0.5m for the pressures 62 and 82kPa then it increased rapidly with the increase of the pressure 4. Moreover, in the tringle system, the DU increased by 2.42% in all the experiments when applying 200kPa in compared to applied 250kPa, which, may be due to low the maximum and minimum temperature at 200kPa. Therefore, it is preferred applied by 200kPa in the square and the tringle systems, it is also does not consume high energy. Which, it is in agreement with 47 was applied two types of Floppy sprinkler system (FSS) (imported and domestic) to evaluated three factors, distribution uniformity (DU), application efficiency of the low quarter (AELQ) and uniformity coefficient (CU), under various grades of height of riser and operating pressures. The system is running in a diverse pressure (150–400kpa) and three heights of riser 3, 3.5 and 4.5m for two sprinklers. As a result, both types of floppy sprinklers produced maximum DU, AELQ and CU at a height of riser 3m and an operating pressure of 250kpa, Values of DU decreased at the same operating pressure when the riser height increased for 3.5 and 4m.
The relationship was positive for height, discharge, pressure, relative humidity and wind speed with the CU value, the strongest positive relationship was discharge 0.77 and pressure 0.75 for KA-4 sprinkler (Fig. 4). However, the relationship was negative for maximum and minimum temperature with the CU value. Moreover, for FOX sprinkler the relationship was positive for discharge, pressure, relative humidity and wind speed with the CU value, discharge and pressure were strongest positive relationship where were 0.65 and 0.64, respectively (Fig. 4). While, height maximum and minimum temperature were in negative relationship with the CU value. On the other hand, the relationship was positive for height, wind speed, maximum and minimum temperature, although, the relationship was negative for discharge, pressure and relative humidity with the CU value for 2520 sprinkler. The strongest positive relationship was maximum temperature was 0.59 and minimum temperature was 0.57. Further, for KA-4 sprinkler was height, discharge, pressure, relative humidity and wind speed were in positive relationship with the DU value, the strongest positive relationship was discharge and pressure were 0.65 and 0.64, respectively. While, the relationship was negative for maximum and minimum temperature with the DU value. On the other hand, the relationship was positive for discharge, pressure, maximum temperature, relative humidity and wind speed with the DU value for FOX sprinkler, where, the strongest positive relationship was wind speed 0.53, discharge 0.44 and pressure 0.42. However, the relationship was negative for height and minimum temperature. Moreover, for 2520 sprinkler height, maximum temperature, minimum temperature and wind speed were in positive relationship with the DU value, the strongest positive relationship was maximum temperature was 0.65 and minimum temperature was 0.63. While, discharge, pressure and relative humidity were in negative relationship with the DU value (Fig. 4).
3.2 Machine learning models
Four scenarios were applied to predict the water distribution uniformity based on machine learning algorithms (Table 2). In CU, the highest value of R2 is 0.80, 0.83 and 0.93 in RF, XGB and XGB-RF, respectively in the first scenario, however, the lowest value of R2 is 0.13, 0.41 and 0.42 in RF, XGB and XGB-RF, respectively in the second scenario. On the other hand, the lowest value of RMSE is 2.92, 2.75 and 2.01 in RF, XGB and XGB-RF, respectively in the first scenario, therefore, the highest value of RMSE is 6.12, 5.12 and 5.26 in RF, XGB and XGB-RF, respectively in the second scenario (Fig. 5a). Moreover, the lowest value of MAE is 2.34, 2.22 and 1.48 in RF, XGB and XGB-RF, respectively in the first scenario, also, the highest value of MAE is 4.85, 4.28 and 3.60 in RF, XGB and XGB-RF, respectively in the second scenario (Fig. 7a). In addition, the lowest value of SI is 0.04, 0.036 and 0.026 in RF, XGB and XGB-RF, respectively in the first scenario, however, the highest value of SI is 0.079, 0.066 and 0.067 in RF, XGB and XGB-RF, respectively in the second scenario (Fig. 6a), which, the results are consistent with the results of 1, where, were used five models ( ANN, NF-GP, LS-SVM, NF-SC and GEP) to predict the distribution uniformity of water in the fixed sprinkler irrigation system, and, the best results for the statistical coefficients were value of R2 is 0.970, 0.867 and 0.891 in ANN, NF-GP and LS-SVM, respectively. Moreover, value of SI is 0.038, 0.080 and 0.072 in ANN, NF-GP and LS-SVM, respectively. In addition, value of MAE is 1.842, 3.662 and 3.572 in ANN, NF-GP and LS-SVM, respectively. 48 not agreement with that research, where, his study demonstrated the effectiveness of using XGBoost as a sophisticated machine learning (ML) model to predict measurement phase levels over 5-min timescales with variety observation time windows. A Flood Alert System (FAS) based on XGBoost models is assessed by two historical flood events in the flood-prone watershed of Houston, Texas. The prospective phase values of the FAS are compared to the monitored values showing good performance by statistical measures (RMSE and KGE). The results of RMSE were 0.14 at Gauge 520m and 0.043 at Gauge 540m.
The study of 49, a Random Forest model was developed using sample data derived from meteorological measurements including air temperature (Ta), relative humidity (RH), wind speed (WS) and photosynthetic active radiation (Par) to predict the lower baseline (T wet) and upper baseline (T dry) canopy temperatures for Chinese Brassica from 27 November to 31 December 2020 (E1) and from 25 May to 20 June 2021 (E2). The study showed the feasibility of using random forest model for prediction T wet and T dry. The value of R2 is 0.90 and 0.91 in E1 and E2, respectively for T wet, also, the value of R2 is 0.88 and 0.89 in E1 and E2, respectively for T dry. Moreover, the value of RMSE is 0.96 and 0.97 in E1 and E2, respectively for T wet, therefore, the value of RMSE is 1.33 and 1.65 in E1 and E2, respectively for T dry. On the other hand, the value of MAE is 0.76 and 0.83 in E1 and E2, respectively for T wet, also, the value of MAE is 1.03 and 1.45 in E1 and E2, respectively for T dry, which, it is in agreement with results this study when using RF to predict the distribution uniformity.
50 proposed a novel ET0i estimation model named PSO-XGBoost, as, the main regression model and the parameters of XGBoost were optimized using the Particle Swarm Optimization (PSO) algorithm. Weather and soil moisture data during the cultivation process of the two crops were used as experimental data with the calculation of ET0i based on the improved Penman-Monteith equation. The results indicated that the PSO algorithm could stably improve the parameters of the XGBoost model. The PSO-XGBoost model was able to precisely appreciation ET0i in a variety of data modes. The values of performance evaluation R2, RMSE and MAE are 0.938, 0.258 and 0.199, respectively by XGB model. That agrees with results this study in results R2 but not agreement with results this study in results RMSE and MAE. On the other hand, the results of PSO-XGBoost model of R2, RMSE and MAE are 0.987, 0.177 and 0.125, respectively.
The R2 decreased in the third scenario by 1.325%, 3.032% and 8.706% in comparison with the first scenario in RF, XGB and XGB-RF, respectively. However, the RMSE increased in the third scenario by 3.879%, 6.004% and 41.643% in comparison with the first scenario in RF, XGB and XGB-RF, respectively. The R2 increased by 16.66% in the first scenario when applied XGB-RF in comparison with RF. Moreover, The RMSE decreased by 31.25% in the first scenario when applied XGB-RF in comparison with RF. So, it is preferable to apply XGB-RF to predict the water distribution uniformity. Therefore, the first scenario is the best one, with the highest results followed by the third scenario. On the other hand, the results were not significantly changed in scenario three and four, that means, the sprinkler height did not significantly impact on the results, because, the climate impact is not significantly during the experiment period, for example, the wind speed was not high during the experiment period, thus, it did not impact on the distribution uniformity. Also, for temperature, the experiments were conducted in the morning to avoid the negative effect of temperature.
On contrast, for the DU, the lowest value of R2 is 0.275, 0.481 and 0.522 in RF, XGB and XGB-RF, respectively in the second scenario, on the other hand, the highest value of R2 is 0.701, 0.479 and 0.827 in RF, XGB and XGB-RF, respectively in the first scenario (Fig. 5b). Similar conclusions were studied by 51 used XGBoost regression (XGBR) model to calculate ET, his study was for three years (2019–2021) to model the effects on ET of eight meteorological factors (net solar radiation (Rn), mean temperature (Ta), minimum temperature (Tamin), maximum temperature (Tmax), relative humidity (RH), minimum relative humidity (RHmin), maximum relative humidity (RH max) and wind speed (V)) using a greenhouse drip irrigated tomato crop ET prediction model (XGBR-ET) that was based on XGBoost regression (XGBR). The model was compared with seven other common regression models. Were lined up, the eight models in terms of forecast precision, XGBR-ET > GBR-ET > SVR-ET > ABR-ET > BR-ET > LR-ET > KNR-ET > RFR-ET. The parameters of the XGBR-ET model were ablated to show that the order of importance of meteorological factors on XGBR-ET was Rn > RH > RHmin > Tmax > RH max > Tamin > Ta > V. The values of performance evaluation R2, RMSE and MAE are 0.981, 0.163 and 0.132, respectively. however, for the DU, the highest value of SI is 0.12 and 0.11 in RF and XGB, respectively in the second scenario (Fig. 6b), but for XGB-RF is 0.083 in the fourth scenario. In addition, the lowest value of SI is 0.07 and 0.05 in RF and XGB-RF, respectively in the first scenario, while is 0.08 in the third scenario in XGB model (Fig. 6b). Moreover, the lowest value of MAE is 4.6 and 5.7 in RF and XGB, respectively in the first scenario, although, in XGB-RF is 2.9 in the second scenario (Fig. 7b), also, the highest value of MAE is 8.11 and 6.77 3.60 in RF and XGB, respectively in the second scenario, but in XGB-RF is 4.41in the fourth scenario (Fig. 7b) for the DU.