*4.1. Performance of the Smart ET-Based Controller*

It is crucial to evaluate the manufacturer programmed and user-defined settings of the available ET-based commercial controllers and investigate the reliability of their algorithms for efficient autonomous irrigation scheduling [11,21]. The actual water applied by the irrigation controller in this study in both years was substantially higher than the initial target irrigation levels across the treatments. The main reason for overirrigation was the initial underestimation of the irrigation precipitation rate using the catch-cans method. We [7] also reported the same issue about underestimating the irrigation precipitation rate using the catch-cans test. The inadequacy of the catch-cans method might be attributed to spatial variation in water applied within each plot and limited sampling areas represented by the catch devices. Moreover, a substantial fraction of applied water may not get into catch devices because of water hitting the cans at an angle and thus splashing out. Accurate estimation of irrigation precipitation rate and proper selection of minimum plant factors are the most critical factors for reliable autonomous irrigation scheduling using a Weathermatic ET-based controller [7].

The second reason for the overirrigation was the overestimation of ETo by the controller compared with CIMIS ETo. We [7] obtained on average 5.7% overestimation compared with CIMIS ETo for the same Weathermatic controller equipped with an on-site temperature sensor and latitude-based solar radiation estimations in inland Southern California. This error range is very close to the 4–5% ETo differences observed in this study, indicating an acceptable estimation of ETo by the controller. Hargreaves and Samani [17] performed even better when long-term weather data were used. Further studies are needed to determine the performance of temperature-based ETo models in different climate regions across the state. Both this study and our recently published study in Southern California [7] focused on the summer months with the highest irrigation demand when no rainfall is typically received, and therefore, water conservation is essential in Central and Southern California. The higher reported overirrigation values of 10% in North Carolina [22] and 32% in Florida [23] suggest that considerable precipitation may negatively impact the efficiency of ET-based controllers as the incorporation of rain into irrigation scheduling by these controllers is often very simplistic.

#### *4.2. Turfgrass Irrigation Management*

In their review paper, Colmer and Barton [24] gathered 29 bermudagrass crop coefficient values ranging from 0.40 to 1.27 in well-watered conditions and from 0.52 to 0.94 under deficit irrigation. Variation in water requirements reported for warm-season grass species shows that local crop coefficient and irrigation recommendations information should be developed, and utilizing a nominal value for all cultivars and locations is not optimal. The TWRF estimated VR values based on long-term mean ETo demand suggested 70–80% ETo as the minimum irrigation application to maintain the acceptable hybrid bermudagrass quality in central California during the high water demand months (i.e., May to August). Reducing the irrigation level to 60% ETo only maintains the turf quality for approximately 75 days before it falls below the minimum acceptable value of 6. We [7] conducted a similar analysis using TWRF developed based on three years of experimental data in inland southern California. In that study, we reported 75% ETo as the minimum requirement to maintain the 'Tifgreen' hybrid bermudagrass quality in peak summer months, which agrees with the findings of this study. Wherley et al. [25] also reported that a commonly used crop coefficient of 0.6 for warm-season grass underestimated water requirement for 'Tifway' hybrid bermudagrass in Florida.

Richie et al. [26] conducted a two-year tall fescue field irrigation trial in Riverside, California. They reported visual qualities between 4 and 6 for most of the experimental periods (~mid-June to mid-November) for irrigation application of 79–85% ETo. Brown et al. [27] conducted a research project to evaluate the response of 'Monarch' tall fescue to different irrigation regimes in Las Vegas, NV, USA. They reported 80% ETo as the optimum irrigation

level to maintain color and coverage of tall fescue turfgrass at an acceptable condition. Ervin and Koski [28] conducted a study near Fort Collins, CO, USA and reported 75% ETo as the safe irrigation level to maintain an acceptable quality for tall fescue turfgrass (a blend of 'Rebel Jr.', 'Crewcut', and 'Monarch'). In our study, the 83% ETo irrigation level only maintained the visual rating above the minimum threshold for two months in 2018. The 108% ETo level with 3 d wk−<sup>1</sup> frequency kept the VR values in the acceptable range in 2018, except for a short period in mid-trial. Both Brown et al. [27] and Ervin and Koski [28] estimated the irrigation volumes applied by measuring the volume of water collected in catch-can devices. If we use the precipitation rate estimated by the catch-cans test, the 108% ETo treatment (which maintained the turf quality for much of the trial in 2018) will be reduced to 89% ETo. However, the catch-cans-based precipitation rate estimations were proved to be inaccurate in our study. The same issue regarding the inadequacy of the catch-cans method to estimate the precipitation rate of landscape irrigation systems was also reported in our recently published study conducted in Southern California [7]. Therefore, we recommend the water savings and crop coefficients reported in the literature based on the catch-cans-estimated precipitation rates be re-evaluated and used cautiously.

In 2019, the VR values of tall fescue plots were relatively low across all irrigation treatments. This might be related to the minimal fertilizer application rates, diminishing growth and greenness of tall fescue in 2019. The positive impact of higher nitrogen fertilization treatment on color ratings was also reported by [27,29]. The TWRF estimations suggest that 60% ETo would be sufficient to maintain the tall fescue quality within the acceptable range for approximately 40 days. Hong et al. [30] conducted a two-year dry down study in Kansas and reported 50% ETo irrigation held 'Seed Research 8650' tall fescue quality above minimum acceptable rating (VR = 6) for 45 and 82 days in two years of their study. The TWRF estimations suggest that 100% ETo would be only sufficient to maintain the tall fescue quality for approximately 55 days. This might be an underestimation impacted by the low VR data in 2019 and, therefore, should be further investigated in the future.

Limiting irrigation watering days to specific days per week is a common strategy that is particularly popular and imposed by water agencies and municipalities in California during droughts to help conserve water in urban areas. We, however, observed no substantial water saving associated with restricting irrigation frequency (watering days). In fact, for both species and in both years, less restrictive watering days improved the VR values. We [7] also found no turfgrass quality improvement or water conservation associated with restricting the watering days in Southern California. This finding is attributed partly to sandy soils with low water holding capacity at both sites and the fact that most of the turfgrass roots are expected to be in the topsoil layer. Su et al. [31] reported that 86% of all root length of 'Dynasty' tall fescue in the field was in the upper 30 cm under well-watered condition and silt loam soil in Kansas, USA. Sinclair et al. [32] reported low amounts of root mass in the deepest soil layer during sod establishment for 'Tifway 419' hybrid bermudagrass grown in tubes filled with loamy sand soil in Florida, USA. The Weathermatic smart controller can dynamically adjust the irrigation frequency based on the actual ETo demand, so restricting irrigation frequency is unnecessary. When irrigation frequency is not restricted, programming a minimum deficit threshold is crucial to avoid light irrigation applications and prevent excessive evaporative loss.

The irrigation scenarios directly impacted the dynamics of near-surface soil moisture. However, the near-surface soil moisture fluctuation over time was not always adequate to explain the variation in turfgrass quality. For example, in 2019, the visual quality of tall fescue for the 83% ETo treatment decreased over time, but the soil moisture showed minimum fluctuations. Moreover, in 2019, the continuous late green-up of hybrid bermudagrass through June and a fertilizer application in mid-July caused the quality ratings to improve as the trial progressed despite a constant decrease in the near-surface soil moisture of the hybrid bermudagrass plots. Consequently, the near-surface soil moisture data should be interpreted carefully and in conjunction with other parameters such as turfgrass physiology and fertilizer applications. The in situ soil moisture values were higher than the

laboratory-measured water retention data (Figure 2). This could be related to differences between laboratory and field sensors and the small size of the laboratory samples. Note that the reported results in this study are for a pop-up sprinkler system with autonomous irrigation scheduling using the Weathermatic ET-based smart irrigation controller. Further studies are needed to determine whether lower irrigation amounts might be sufficient when irrigation scheduling is done based on actual root zone soil water holding capacity information and for irrigation systems with higher potential efficiency, such as underground drip irrigation systems.
