Comment on “Improved Water Savings and Reduction in Moist Heat Stress Caused by Efficient Irrigation” by Anukesh Krishnankutty Ambika and Vimal Mishra

A recent study by Ambika and Mishra (2022, https://doi.org/10.1029/2021EF002642) asserted using simulated evidence that a shift towards drip irrigation from channel irrigation in the Indo‐Gangetic Plains will result in alleviation of moist heat stress and water savings. The assumptions, parameterizations, and approach adopted in this study do not adequately represent fundamentals of irrigation science, management, and engineering, as well as realities of local agricultural water management in the Indo‐Gangetic Plains. This comment attempts to highlight these inadequacies of the paper and related literature, by focusing on four major aspects related to irrigation process, local land‐use, timing of irrigation and moist heat stress, and water accounting. These weaknesses in irrigation processes modeling result in overlooking of indispensable aspects that should be accounted for to comprehensively assess impacts of irrigation method choice on regional weather and water resources.


Introduction
In a recent article in this journal, Ambika and Mishra (2022), hereafter AM22, employ modeling capabilities to investigate how a shift toward efficient irrigation (surface drip irrigation, specifically) will change moist heat stress regimes and water savings over the Indo-Gangetic Plains of India. AM22 continues a useful stream of research oriented toward thermal comfort in a densely populated region globally, and how agricultural water management practices dictate these regimes in the region (Ambika & Mishra, 2019, 2021Mishra et al., 2020). However, there are several flaws with how AM22 has conceptualized, assumed, and implemented the two irrigation systems and their biophysical impacts in their modeling framework. In this comment, I have revisited AM22's assumptions, analysis, and inferences from an irrigation science, engineering, and management standpoint, and the potential misrepresentations that can stem from these. By doing so, I anticipate that appropriate consideration of lesser understood mechanisms of irrigation dynamics in the region will result in improved and robust future undertaking of such investigations.
Various elements of irrigation systems transition and their impacts on soil water balance, surface energy balance, and moist heat extremes have been overlooked or misrepresented by AM22, and are detailed one by one in the following sections.

Wetted Soil and Canopy Surfaces
AM22's analysis has misassumed different components of the soil-canopy system that will be wetted and act as evaporation sources in channel versus drip irrigation regimes. The specific assumptions and parametrizations used by AM22 are inconsistent with the real-world manifestation of wetting and evaporation processes, as well as their original source of irrigation parameterization (Valmassoi et al., 2020). According to AM22, their implementation of irrigation schemes is based on Valmassoi et al. (2020), however, I find stark differences among the two. While AM22 mentions that their drip irrigation scheme only accounts for soil evaporation, and channel irrigation considers both soil evaporation and canopy interception, this is entirely opposite to what has been asserted by Valmassoi et al. (2020). Both schemes, in addition to being inconsistent with each other, are also contrasting to what is observed in irrigated agricultural fields (described in Table 1).
Channel and drip irrigation differ primarily in their soil wetting attributes, and not canopy interception. A primary difference in soil wetting attributes of the two methods is the extent to which soil surface is wetted by water inputs, represented by a fraction (f w ). As defined by ASCE Task Committee on Revision of Manual 70, f w should be considered as 1.0 for precipitation, sprinkler, and basin irrigation, 0.6-1.0 for furrow irrigation, and 0.3-0.4 for drip irrigation. Drip irrigation only wets a small portion of the soil surface adjacent to the plant row, and this portion can be even smaller when drip tape is installed in alternate crop rows. The fraction of soil surface from which most evaporation occurs is the minimum of f w and fraction of soil surface not shaded by vegetation. The interplay among these two processes is changed with source of wetting (irrigation vs. precipitation) and crop growth (leaf area index or ground canopy cover). AM22 did not consider the difference in f w among the two irrigation systems, and its seasonal dynamics with canopy growth. Contrary to what is suggested by Valmassoi et al. (2020) and AM22, drip irrigation does not apply water right above the canopy, but at the soil surface. Moreover, under both channel and drip irrigation, water does not considerably wet the canopy (as in sprinkler irrigation or precipitation events), except for soon after emergence.

Irrigation Application Rate, Frequency, and Depth
AM22's analysis does not account for the modified dynamics of application rate, irrigation frequency and irrigation depth under channel versus drip irrigation. Being a low-pressure system, drip irrigation operates with a low water application rate, and thus, requires longer to apply a predetermined irrigation depth, and is ideal for high frequency applications. It is not uncommon to see drip systems running continuously through the peak water demand periods in semi-arid and arid regions. These are critical differences that represent the distinction of drip systems from channel irrigation in terms of water influx and were overlooked by AM22. Moreover, AM22 has assumed that irrigation depth applied remains unaffected from choice of irrigation, which is contrary to the fact that drip irrigation has higher irrigation efficiency. A higher irrigation efficiency from reduced soil evaporation means that the required effective irrigation depth should be lower in drip irrigation compared to channel irrigation.

Evaporation Versus Transpiration
Crop water use is sum of (a) soil and canopy evaporation (E) and (b) canopy transpiration (T). The impacts of irrigation choice on E can also indirectly can impact T. As long as well-watered conditions are maintained in the root zone, plant water uptake will be optimum and limited by available energy only. However, a reduction in E under drip irrigation is likely to increase vapor pressure deficit owing to a modified microclimate, and in turn, increase T, offsetting water saving from reduced E (Busari et al., 2015;Irmak & Kukal, 2022;Irmak et al., 2019;Q. Li et al., 2019;Tolk et al., 1995;Uddin & Murphy, 2020). AM22 does not decompose the impacts of irrigation choice on E and T individually, which might be due to inability of WRF to do so. However, it is critical that impacts on T are evaluated as well to ensure that optimum carbon assimilation is maintained, and no yield penalties are encountered (Doorenbos & Kassam, 1979;Yang et al., 2020). This is especially important because T represents dominant ET component in crops with near-complete to complete canopy closure during peak water demand periods.

Disconnected Nature of Irrigation Scenario From Local Cropping Systems
AM22 considers a (hypothetical) scenario where channel irrigated acreage is brought under drip irrigation scheme. Simulated findings are presented as policy-oriented proof of concept demonstrating dual benefits of this transition. The idea of this transition is highly detached from local agricultural cropping systems, current practices, and realities that do not adequately justify irrigation infrastructure upgrades. There are several reasons as to why current cropping systems hinder smooth adoption of drip irrigation. Rice is a major crop grown in the IGP region, in terms of both economic value as well as water consumption. Here, rice is grown under the puddled transplant regime, in which anaerobic conditions are induced by puddling and continuous flooding. Substantially high irrigation application rates are required to saturate the root zone prior to transplanting as well as establishing ponded conditions, which cannot be fulfilled by low-pressure irrigation. Although there is a possibility that at least some of the regional rice acreage can be directly seeded, however, there are enormous challenges for adoption given higher weed infestation, lack of varieties, lower yields, poor establishment under flooded soils, and Table 1 Elements of Drip and Channel Irrigation Schemes, as Assumed by AM22 and Valmassoi et al. (2020)  Accounted for canopy interception and soil evaporation Drip irrigation is applied at the soil surface and does not wet the canopy.
Evaporation is from unsaturated soil only Channel Accounted for canopy interception, canopy evaporation, and soil evaporation Accounted for soil evaporation and evaporation of ponded surface water Channel irrigation does not wet the canopy either. In some crops like rice, water will be ponded at the surface creating saturated surface conditions.

Note.
For each element, the physical manifestation of the irrigation scheme (column 4) is also provided.
iron deficiency (Jat et al., 2020). Given regionally suitable varieties become available, water savings can be achieved using science-backed and cheaper management (sound irrigation scheduling) options while still using channel irrigation (Y. Li et al., 2020), rather than investing in drip irrigation. Thus, water conservation under current situation is not hindered by channel irrigation or lack of higher irrigation-efficiency technology, but by regional crop and management regimes. Bulk of the increased irrigation efficiency associated with adoption of drip irrigation is due to avoidance of losses (percolation and evaporation) while water is delivered to the field, which is not accounted for by AM22 owing to model incapability.

Asynchrony Between Irrigation Season and Periods of Moist Heat Stress
There is incoherence between the considered timings of moist heat stress, irrigation season, and summer by AM22. For the purpose of their investigations, AM22 have considered April and May as the summer months, with the intention of representing pre-monsoon conditions. This assumption, however, does not seem to hold true for Indo-Gangetic Plains, where monsoon onset occurs after 15 June, on an average. South-Asia is divided into early and late monsoon regions (Raymond et al., 2020), and vast majority of Indo-Gangetic Plains lies within the latter. Raymond et al. (2020) reports that the station-weighted average climatological monsoon onset date for the late monsoon region of South Asia is around early July. Moreover, they show that extreme moist heat stress conditions are most severe around the time of monsoon onset, and they continue well into the monsoon season. Thus, limiting the investigations to April-May leads to critical periods of moist heat stress being overlooked in the Indo-Gangetic Plains.
Additionally, irrigation as a management practice has been only spatially represented in AM22's simulations, without considering and addressing the seasonality of irrigation water fluxes in the region. This is critical while attributing differences in moist heat stress to differences in irrigation schemes. The investigation months (April and May) are not especially significant from an irrigation standpoint, as they do not represent substantial crop water use period for any of the major regional cropping systems. April marks the time for the wheat crop approaching physiological maturity, and thus, irrigation is no longer required. Rice, the next major crop that holds significance from water consumption standpoint, is not sown until mid-to-late June. The extensive focus on the months April and May by large-scale irrigation modeling studies is most likely due to the fact that this period represents the annual maxima of water deficit (demand minus supply). However, a larger water deficit does not necessarily translate into need for irrigation water application. In fact, irrigation water application is substantially large (depth/volume-wise) during rice growing season, that is, June until September, as ponded conditions are maintained by rice growers, in an effort to avoid rice sensitivity to unsaturated soil conditions. Modeling perspectives do not appropriately address rice irrigation, being based on tracking water deficits, and do not account for cropping system-specific management practices. In fact, it has been shown that crop water requirements are large in September-October for northwest India (Huang et al., 2018), leading to peak irrigation water withdrawal in this period (Famiglietti, 2014;Rodell et al., 2009). Thus, irrigated production systems should instead show controls on atmospheric moisture budget during periods that extend beyond April and May, when significant amounts of water are applied to crops, and greater crop transpiration rates occur.

Higher Irrigation Efficiency Does Not Translate Into Water Savings
AM22's simulations conclude that a shift from channel to drip irrigation results in water savings. Such connections among irrigation technology and water savings are incomprehensive and are rarely achieved globally. It has been shown extensively that increased irrigation efficiency via use of advanced technologies (such as surface to drip irrigation shifts) rarely results in improved water availability (Grafton et al., 2018;Lankford et al., 2020;Sears et al., 2018;Ward & Pulido-Velazquez. 2008). AM22's assertions on linkages between irrigation efficiency and water savings overlook basin-scale water accounting and behavioral response of irrigators, both of which are critical determinants of actual consequences of a higher irrigation efficiency. Under the current channel-based irrigation scheme in the Indo-Gangetic Plains, the farm-scale "losses" of water are recovered and reused at watershed/basin scales. A typical shift from channel to drip irrigation across the globe has not only been associated with reduced soil evaporation, but also reduced return flows, reduced subsurface recharge, increased irrigated acreage, and shift toward crops with higher water requirements (Grafton et al., 2018). These unintended consequences have been also registered within AM22's area of interest (Rajasthan), where farmers have been incentivized to adopt drip irrigation to "save" water, but these policies have resulted in increased irrigated area and total