Discussion of “SWAT-Based Evapotranspirative Water Conservation Analysis Performed on Irrigated Croplands to Determine Potential Regional Water Savings” by Andrew Gayley

The discussers appreciate the author’s investigation of the utilizing soil and water assessment tool (SWAT) as a geographic information system (GIS)-based modeling platform for quantitative measurements of potential water savings across row crops grown in South Central Texas. The discussers would like to present the following points of view, which the author and potential researchers should consider. 1. Under the subheading “Modeling Platform” on page 457, the author has mentioned about using Priestly-Taylor method for computation of potential evapotranspiration (PET). At the same time, the author stated about Penman-Monteith equation for data quality evaluation of evapotranspiration in the subheading “Regional Irrigation Deficit” on page 458. These statements seem to be contradictory. Clarification about the methodology considered in this study is required. 2. The method used to estimate potential evapotranspiration was either the Priestly-Taylor method or the Penman-Monteith equation, which is an empirical method. It is well known that use of this empirical method in different areas yields different results and these methods are fitted to specific regions due to climatic variables (Xu and Singh 2002). Therefore, the evapotranspiration equation needs more scrutiny to determine an empirical method that fits better in the San Miguel Creek watershed, Texas. To obtain accurate values of evapotranspiration, perhaps additional work such as use of lysimeters are needed. The comparison of results of evapotranspiration measured by lysimeters and results employing various empirical methods (including Penman-Monteith and Priestly-Taylor method, among others) are assisted by using the best possible method to estimate evapotranspiration. On the other hand, the estimated evapotranspiration equation which is closer to lysimeter measurements is the best equation. Moreover, author has not provided essential details of either PriestlyTaylor method/Penman-Monteith equation in this study. 3. The author stated “Typical San Miguel flows observed at Tilden, Texas (USGS gauge 08206700), ranged from 0.17 (December) to 8.7 (July) m=s during the 10-year study” on page 457. It seems to discussers that only one gauge station data is used for the calculation of hydrological variables on page 457. In many research studies, it has been found that one point source of gauge station data is insufficient to calculate hydrological variables of larger area watershed since the only one outlet point station data cannot represent the truth characteristics of whole watershed (Cho et al. 2009). The author has not provided enough insight about this important aspect of hydrological modeling. This needs an explanation. 4. Under the subheading “Regional Irrigation Deficit,” the PenmanMonteith equation was used for the data quality evaluation process in an ArcSWAT platform. Arnold et al. (2012) and Zhang et al. (2008) reported that SWAT requires current landuse/ landcover (LULC) data and other hydro-meteorological parameters viz daily rainfall, minimum and maximum temperature, humidity, wind speed, and solar radiation data for calculation of the hydrological variables like PET, evapotranspiration (ET), actual evapotranspiration (AET), water yield, and surface runoff at the HRU (hydrological response unit) level. However, in this study, the author has not discussed about these parameters. In the absence of these parameters, SWAT cannot perform well especially for this kind of larger area watershed. The discussers wonder that how the author coped with this problem. 5. Under the subheading “Error Analysis,” the author compared simulated results with previously generated 2005/2006 raster images krigged from precipitation (PPT), PET, ET, and irrigation needs (IN) data from 22 Texas ET stations using the Penman-Monteith equation. The author has not provided necessary information of the methodology of calculating these variables. The distributed model has been shown to be sensitive to the locations of the rain gauges within the catchment and hence, to the spatial variability of the rainfall over the catchment (Bell and Moore 2000). Failing to consider adequately the spatial variability of rainfall will lead to errors in the values of the model parameters, which will be wrongly adjusted to compensate for errors in the rainfall input data (Schuurmans and Bierkens 2007). This needs an explanation. 6. Under the subheadings “Water Conservation Analysis,” the author has calculated irrigation deficit using SWAT. However, this section does not give sufficient information on the irrigation deficit used to compute optimal water conservation. It would be helpful for the readers if the author presented information regarding resolution or grid size crop data used, temporal changes in crop yield, methodology opted to analyze actual irrigation deficit, as well as changes that occurred in ET, PET, and IN. 7. On page 459, the author states, “Results were compared with previously generated 2005/2006 raster images krigged from PPT, PET, ET, and IN data from 22 Texas evapotranspiration stations using the Penman–Monteith equation for ET and a cotton/winter-wheat crop rotation across EAGCDAGRR land holdings.”Author has not provided any raster or gridded maps/ information in this article. If this 2010/2011 raster images were used, then the results from the study may be different. 8. The author has not provided information on rain gauges distributed over this large basin. According to Bell and Moore (2000), rain gauge density over the catchments is one of the main factors in attaining accuracy during analyzing hydrological variables for large area watershed. This needs an explanation. 9. The author provided calibration results of comparing modeled output with published values in Table 1 on page 460. The SWAT model has a tendency to simulate the outcomes of the model on a daily, monthly, and yearly basis. The discussers wonder about the time scale (i.e., daily/monthly/annually) used by the author while computing parameters presented in Table 1.