This is an interesting article describing effects of high temperature and drought on maize yield and yield components in Nebraska. The authors used remote sensing to detect high temperature stress and drought stress and also tested whether four different crop models can reproduce the effects detected by remote sensing. The article is well written, good to understand and figures are of high quality. However, I cannot recommend to publish the present version of the article in HESS. My major criticisms are:

1) The major source to describe high temperature and drought stress in maize are land surface temperature and ET detected by remote sensing. I think that the temperature based indicators LST and EDD are highly determined by the ratio ET/PET which was used to describe drought impacts. Which factor different from drought can explain canopy temperature differences between well watered and rainfed maize fields? Or in other words: can differences in LST and EDD at the same location happen independently of drought stress? I don't think so. If so, for example because of different LAI, then this is likely an affect of drought in previous growth stages.

It is well understood that transpiration cooling is directly controlled by the stomata conductance and vapor pressure deficit, which are again controlled by drought. This is also the reason why canopy temperature differences are often used as indicator for drought stress or even for irrigation scheduling. Consequently I think that EDD differences or LST differences between irrigated and rainfed maize in the same region are just another manifestation of differences in drought stress between irrigated and rainfed fields. From that perspective I cannot understand why the collinearity tests performed for the variables included in equation 7-9 did not show critical values.

2) The authors showed that there are considerable differences in the growing season length of irrigated and rainfed maize and suggest that the differences are mainly an effect of cooler canopy temperature under well watered conditions (lines 322-337). Another potential reason could be the so called drought escape effect. It is known that many crops speed up their phenological development under drought to make sure that grains reach physiological maturity before the stress becomes so strong that the crop has to die. Again, in that case it would be a drought effect and not an effect of higher temperatures. I agree that it is not so easy to find out which effect really matters. I suggest to test the GDD computed in equation 3 for years with similar canopy temperature but different drought stress (ET/PET ratio). For example, a year that is warm and wet should result in similar canopy temperatures compared to a year that is a bit cooler but dry. Important is that the test has to be made for the same location (county) to avoid that cultivar differences between warmer and cooler regions disturb the relationship. If for years with similar canopy temperature but different ET/PET ratio the GDD is similar, then the shorting of the growing period is independently of drought and the drought escape mechanism can be excluded. If GDD is, for similar canopy temperatures, positively correlated with the ET/PET ratio, then this would point to the drought escape mechanism.

Specific comments:

Line 175 (equation 3): Why was it decided to set the high temperature threshold to 30 dC? In the literature heat stress thresholds for maize are typically higher, about 34 dC (Sanchez et al., 2014).

Line 262 (equation 7): How was LST and ET/PET computed? As mean for the whole growing period? In the variable explanation (line 265) you call LST "local crop temperature stress" but shouldn't you then better use EDD here?

Lines 280-292: Any reason why delta EDD and delta ET/PET are NOT highly correlated?

Lines 363-365: "As shown in Figure 7, we found that temperature sensitivity of yield was significantly weakened from − 6.9%/℃ to −1%/℃ in irrigated vs. rainfed areas ..."

=> shouldn't this be vice versa (lower sensitivity in irrigated maize)?

Lines 438-442: The assimilation of satellite derived LST might in fact reduce crop model uncertainty but this helps only when LST data are available. Crop models are also often used for climate change impact analysis but for simulation of potential futures LST is not available. Another disadvantage could be that LST is sensor and satellite specific, for example due to the different overpass times. Therefore another recommendation could be to improve crop models so that they can reproduce the effects that were found in the present study and use remotely sensed LST for validation.

Figure 8: It seems that there is also considerable drought stress in irrigated maize because the ET/PET ratio is often much lower than 1. Any explanation why yield under irrigated conditions is often much higher for similar ET/PET ratios? Because irrigated maize is more often grown in cooler regions?

References:

Sanchez, B., Rasmussen, A. and Porter, J.R. (2014). Temperatures and the growth and development of maize and rice: a review. Global Change Biology 20, 408–417.

Irrigation benefits crop yield mainly through water stress and high temperature stress mitigation. Although water stress alleviation via irrigation has been addressed intensively, a further understanding of high temperature stress alleviation is still required. This paper attempts to separate the irrigation effects on maize heat stress from that on water stress using satellite vegetation, temperature, and evaporation products. The paper is well written and structured overall. The conclusions are drawn based on solid analyses and interpretations of the results. The pathway provided to improve the state-of-the-art crop models is of great interest to the community.

The only major concern for me is the collinearity between LST and ET considering LST is directly impacted by ET through surface energy balance. Thus, it should be careful to disentangle the heat stress and water stress. More illustrations would be required.

Some minor comments are as follows.

Lines 190-194, I could not find the citations ‘Senay et al., 2013’ and ‘Velpuri et al., 2013’ in the reference.

Lines 195-196, Is the PET also available in the SSEBop ET product? It would be better to use consistent ET and PET to calculate the water stress index.

Lines 256-257, I would suggest to simply describe the uncertainties of AGE rather than just include the reference for better readability.

Lines 286-287, what is the added value by using daytime LST difference considering the relative contribution of water and high temperature stress alleviation to yield benefit has been analyzed using Eq. 8.

Lines 317-318 and lines 346-347, could some explanations be found for the different performances between LST and air temperature?

This paper proposes, based on a series of LST measurements derived from Modis, to analyse the effects of thermal and water stress by considering a big data set over a large territory and a long time series 2003-2016. The originality of this work, beyond the considerable corpus of data, is to analyse the contributions of both types of stress.

The approach nevertheless presents an important methodological flaw in the separation of thermal and hydric effects. Indeed, water stress leads to a decrease in photosynthesis and therefore in yields, but also to an increase in temperature, which itself can have an effect on yield. Therefore, to separate the thermal effect it is necessary to be able to control the water stress. This is the case with irrigated conditions and Figure 8b does not show a clear effect of heat stress under these conditions. The authors use an empirical model (eq 8) which is not at all suitable for separating the temperature and water effects or it should be demonstrated. Therefore i found that the conclusions (as the 65% and 35%, heat) cannot be supported by such methodology and probably the author give too much importance to the heat stress.

In fact this study does not refer to existing knowledge in ecosphysiology on heat stress on field crops which would have allowed a better understanding of the periods and impact of heat stress on yields. This is reflected in the choice of crop models, which is documented in a much too summary manner. Because of the strong link between water stress and temperature, on the one hand, and the pre-eminence of water stress on yields, it seems difficult to isolate the effect of heat stress on yield with a simple statistical analysis as is done in the paper. Moreover, the choice of explanatory variables aggregated over two parts of the cycle does not help to analyse phenomena that occur over short periods of time linked to climatic variability and the sensitivity of the yield to heat stress.

Because of the unsuitable approach to adress thermal stresses which would have been the true originality of this work, I do not recommend the publication of this article. Moreover, it has some formal defects:

The authors could describe a little better the sources of performance data

L126 what is a MODIS sinusoidale projection

L141-144 : better describe how phenology is retrieved. Site observations in Figure 4c show that the phenology was not well characterized (gap of 20 days with VP this gap might have commented

L176-180 : I guess that met data are obtained hourly, why using sine function (is the fact of using sine function has an impact on GDD and EDD

L301  not clear

L328-329: are irrigated and non irrigated using varieties. Probably not and we can expect that phenological characteristics might be different. This can explain shorter cycle with non irrigated crop.

L390-403 : in Agmip there several models that compute crop temperature (STICS for instance), wihy not using them. Are sure that LPJ-guess do no not compute crop temperature.