國立臺灣大學理學院大氣科學研究所 碩士論文
Graduate Institute of Atmospheric Sciences College of Science
National Taiwan University Master Thesis
灌溉的濕化與冷卻作用對於近地表微氣候的綜合反應 Integrated Responses of Irrigation Moistening and Cooling Effects to the Near-Surface Microclimate
葉亭佑 Ting-Yu Yeh
指導教授：羅敏輝 博士 Advisor: Min-Hui Lo, Ph.D.
中華民國 112 年 1 月 Jan, 2023
誌謝
能完成⼀篇碩⼠論⽂，其實需要的耐⼒以及膽量是真正投⼊研究後才懂 的。⼀路⾛來，無論是學校⽅面或是⼤氣系的老師們都給予我莫⼤的幫助。首 先，我要感謝敏輝老師⼀路的栽培我，第⼀次與老師多次交流是在製作統計投 影片，到後續碩⼠班多次 meeting 的交流，都感受到老師的學識淵博與對學⽣
的關愛，讓我在⼤氣系的研究路途上不僅學習到⼤量的陸氣交互知識，老師也 給我很多勇氣繼續往前前進。這份勇氣得來不易，讓我有機會提出我的見解，
有機會在腦中思辨，更有機會踏進⼤公司實習，這份恩情實在難以⾔喻。或許 我的誌謝在論⽂庫的茫茫⼤海中如同⼀根細針，但這份感謝在我⼼中佔有不可 或缺的地位。
另外我也要感謝研究室的夥伴們，在專題討論前給予我很多有意義且實質 的勘誤與想法，也協助我模式的輸出，若沒有⼤家的幫助，這篇論⽂將會漏洞 百出且內容簡陋。我還要謝謝鴻基老師，在我碩⼀時期的照顧，當年的⼀字⼀
句的推導淺⽔⽅程式，都化為我後續對於研究的數學基礎，碩⼀時的學長姐也 是對我如家⼈般的呵護，實為感動，希望在美國與氣象局發展的你們，都能成 為未來的閃亮之星。
最後我要感謝我的家⼈，有你的陪伴才有今天的我，住在雨都的我們，冬 季是如此的寒冷，但我們的⼼卻是永遠在⼀起，面對歡笑與淚⽔，⼀起把我送 出社會。
亭佑謝謝你，你無所畏懼的，完成你的⼼願。
中⽂摘要
近年來許多研究探討灌溉導致近地表微氣候改變的程度，但目前尚未有⽂
獻提出在乾濕季下，濕球溫度對於溫度及濕度改變的敏感度，且⾄今尚未了解 灌溉對濕球溫度氣候平均態的影響。本⽂探討灌溉的冷卻效果與濕化作用對於 濕球溫度的互補關係以及乾濕季下濕球溫度的改變特徵。前⼈已發現⼀年中最 熱月份的平均最⾼溫隨著灌溉的擴⼤⽽降溫（冷卻效果），⽽濕化效果主要是強 調灌溉增濕近地面空氣將提升濕球溫度與降低該地區的舒適度，本⽂將探討此 兩⼤作用並結合不同的背景濕度條件，討論濕球溫度變化的主導因素。此研究 分析美國國家⼤氣研究中⼼(National Center for Atmospheric Research)發展之耦 合氣候模型(Community Earth System Model)以及非耦合之陸地模型(Community Land Model)所輸出的兩公尺⾼的日最⾼溫、日平均相對濕度與混合比，並計算 該地區的濕球溫度。我們同時使用多重變數線性回歸技術從多重訊號中分離出 單⼀強迫項，得出冷卻效應與濕化效應各自對濕球溫度的影響。
灌溉比例在美國中部、歐洲、南亞與華北地區在過去百年有顯著擴⼤，模 型分析結果發現隨著灌溉範圍擴張，所有地區的最⾼乾球溫度皆下降。在分析 影響濕球溫度的因⼦後，發現混合比對於濕球溫度較為敏感。本研究總結兩種 情形，如果背景相對濕度較低時，則灌溉的濕化效果較⾼，可能會主導濕球溫 度上升的過程。另外，如果背景相對濕度接近飽和，因為蒸發機制不顯著，導 致灌溉冷卻與濕化效應無顯著發⽣，進⽽對濕球溫度無顯著影響。由於暖化下 的氣候平均態改變，會改變灌溉的效應，因此未來討論乾季濕熱之熱傷害，需 同時考量到灌溉與暖化下的共同效應。
關鍵字：灌溉、濕球溫度、濕化效應、冷卻效應、熱傷害、混合比。
Abstract
Irrigation practices can have significant biogeophysical effects on the climate.
Previous studies have shown that the change in average daily maximum temperature during the hottest month of the year has warmed less in regions with irrigation expansion in the past 100 years. Furthermore, the irrigation’s moistening effect may cause higher wet-bulb temperature due to higher near-surface water vapor from excess evaporation. However, the effects that dominate the change of the wet-bulb temperature in the dry and wet seasons are not well understood. This study
investigates the competing effects of cooling and moistening on the wet-bulb temperature. We use the meteorological variables of daily maximum temperature (T2m); daily mean relative humidity (RH) and daily mean surface pressure are used to calculate the specific humidity (or mixing ratio) and wet-bulb temperature from NCAR CESM coupled climate model and the offline NCAR Community Land Model. The linear regression technique isolates an individual forcing from a lumped signal and analyzes the temperature change through irrigation cooling and moistening effects. The irrigation fraction expanded in the central USA, Europe, South Asia, and North China in the past 100 years, so the maximum temperature decreased over those regions. We further differentiate the wet-bulb temperature from the dry-bulb
temperature and the mixing ratio, which is very sensitive when the mixing ratio changes. The results show that when the background relative humidity is low, the mixing ratio could change a lot, which means the amount of mixing ratio change has a
high probability of staying in the dominant region. The wet-bulb temperature is non- linear with the T and RH. We conclude with two scenarios. If background RH is low, the irrigation moistening effect most likely dominates. On the other hand, if the background RH is high, the evaporation is less from the lower water gradient.
Therefore, there is no apparent cooling or moistening effect to alter the wet-bulb temperature. In a nutshell, irrigation can worsen comfort and increase the danger of heat stress, especially in dry conditions. This is an essential factor needed to be considered in the future.
Keywords: Irrigation, Wet-bulb temperature, Moistening effect, Cooling effect, Heat stress, Mixing ratio.
Contents
誌謝……..……….. i
中⽂摘要……..………..ii
Abstract……….……iv
Contents……….…...vi
List of Tables……….………..viii
List of Figures………...ix
Chapter 1. Introduction………...1
1.1 Irrigation cooling effect……….1
1.2 Wet-bulb temperature………3
1.3 Scientific questions and hypotheses………..6
Chapter 2. Data of the coupled climate model and method………….8
2.1 Data………8
2.2 Method……….10
Chapter 3. Results of coupled climate model………...…13
3.1 Global land………...13
3.2 Central USA……….16
3.3 North China……… 17
3.4 Europe………..18
3.5 South Asia………18
Chapter 4. Results of the offline land surface model………...25
4.1 Data………..25
4.2 Comparison between offline and coupled model………....26
4.3 The scatter plot in South Asia………..27
4.4 The diurnal difference in the dry and wet seasons (South Asia).28
4.5The yearly trend of wet-bulb temperature in South Asia and North China………..29Chapter 5. Discussion……….33
Chapter 6. Conclusions………..………37
Chapter 7. References………39
List of Tables
Table 1. CESM simulation production. ………...44
List of Figures
Figure 1. Observed warming rates affected by irrigation. Boxplots of the total (ΔTXm, a, b) and irrigation-induced (ΔTXmirr, c, d) change in average daily maximum temperature during the hottest month of the year for global land (a, c) and South Asia (b, d) between 1901 and 1930 and 1981 and 2010. This figure and legend are taken from Wim et al. (2020) figure 1. ………45
Figure 2. Irrigation driven cooling in the Indo-Gangetic Plain. This figure and legend are taken from Vimal et al. (2020) figure 1. ……….46
Figure 3. Change in probability of hot extremes from expanding
irrigation and other forcings. This figure and legend are taken from Wim et al. (2020) figure 2. ………...…47
Figure 4. (a) Relationship between human mortality and area affected by extreme dry and moist heat stress (%) in India with 3-day maximum Tw greater than 27 during the heatwave. (c) same as (a) but for 3-day
maximum T2 greater than 45. Mortality data was obtained from EM-
DAT (https://www.emdat.be/ ) for the 1979-2016 period. This figure and
legend are taken from Vimal et al. (2020) figure S6. ………..48
Figure 5. Spatial distribution of highest daily maximum wet-bulb temperature in modern record (1979-2015). This figure and legend are taken from Im et al. (2017) figure 1. ………...49 Figure 6. Twmax (°C), and maps of the ensemble averaged 30-year Twmax. This figure and legend are taken from Im et al. (2017) figure 2. ………..50 Figure 7. The role of irrigation on summer heat fluxes, temperature
humidity SLP, and PBL height. This figure and legend are taken from Vimal et al. (2017) figure 3. ………51
Figure 8. Changes in three days maximum heat indicators in India during April to May for 1979 to 2018 period. This figure and legend are taken from Vimal et al. (2017) figure 2. ………...52
Figure 9. Isopleths of Tw versus RH% and T. This figure and legend are taken from Roland et al. (2011) figure 2. ………53
Figure 10. Irrigation fraction difference (%) between the early 20th
century reference period (1901-1930) and present-day (1981-2010). ….54
Figure 11. Change in Δ𝑇𝑋𝑚 versus change in the irrigated fraction in Global land, South Asia, North China, and the Central
USA. ……….…...55
Figure 12. Change in temperature (K), specific humidity (kg/kg), and maximum wet-bulb temperature (K) in global land. ………...56
Figure 13. Same as figure 12, but for Central US. ……….57 Figure 14. The equivalent effect of increasing mixing ratio on dry-bulb temperature change. ……….58 Figure 15. Same as figure 12, but for North China. ………...59 Figure 16. Same as figure 12, but for Europe. ………60 Figure 17. Same as figure 12, but for wet seasons in South Asia. ……..61 Figure 18. The specific humidity (kg/kg) changes due to other
forcing. ……….62 Figure 19. Same as figure 12, but for dry seasons in South Asia………63 Figure 20. Irrigation effect of one month average in dry South Asia
(1981-2010). The criteria are RH lower than one standard deviation. …64 Figure 21. Same as figure 20, but the criteria are RH higher than one standard deviation which represents as wet South India. ………65 Figure 22. The probability density function of (a) temperature change (K), (b) mixing ratio (kg/kg), (c) latent heat flux (W/m
2) in wet and dry conditions. ……….…..66
Figure 23. Comparison between the coupled and offline models in daily
maximum and minimum in South Asia. ………..67
Figure 24. Irrigation effect of one month average in wet South Asia of the offline model (1906-2014). The criteria are RH higher than one
standard deviation. ………...68
Figure 25. Same as figure 24, but the criteria are RH lower than one standard deviation which represents dry South India. ……….69
Figure 26. The probability density function of (a) mixing ratio change, (b) temperature in wet (blue line) and dry seasons (red
line). ………...70
Figure 27. Each variable difference due to irrigation from 1906 to 2010 in the dry season of South Asia. ………....………..71 Figure 28. The same as figure 27, but for the wet season in South
Asia. ……….72 Figure 29. Variables difference affected by irrigation in South
Asia. ……….73
Figure 30. The same as figure 29, but for North China. ……….74
Figure 31. The specific humidity of different layers in North China over
time. ……….75
Figure 32. South Asia profile in 1981-2010 for dry and wet seasons in
control run and the difference between irrigation and control
Figure 33. The summary of wet and dry conditions of wet-bulb
temperature. ……….77
Appendix A1. The probability density function of dry and wet events from January to March in South Asia. The blue and red lines represent the wet and dry conditions. ….………...78
Appendix A2. Same as A1, but for April to June. ………..79
Appendix A3. Same as A1, but for July to September. ………..80
Appendix A4. Same as A1, but for October to December. ……….81
Chapter 1. Introduction
1.1 Irrigation cooling effect
Irrigation practices have a large geophysical effect on climate (Wada et al., 2013).
Researches show that the temperature can be affected by evaporation, which is a
cooling effect and especially in highly irrigated regions (Chou et al., 2018). These
cooling effects can be investigated by observation data and CESM model simulations.
Wim et al. (2020) investigated the expansion of irrigation and found a negative
correlation between daytime summer temperatures and irrigation extent. The
mechanism comes from adding water to soil makes it moisturized. Then, evaporation
takes heat away and the temperature drops. Figure 1 shows the change in average daily
maximum temperature during the hottest month of the year, named, which has warmed
less with irrigation expansion. The delta symbol represents the difference between
1901-1930 and 1981-2010. Figure 1a explains the Δ𝑇𝑋𝑚 in all forcing scenarios.
When the change in irrigated fraction increases, Δ𝑇𝑋𝑚 decreases. Also, when the x-
axis number is small, Δ𝑇𝑋𝑚 is larger than zero. This signal comes from anthropogenic
warming. To get the irrigation effect, Wim et al. (2020) used the regression technique to
distinguish a lumped signal from all forces. Figure 1c shows that irrigation is a cooling
effect. The more changes in irrigated fraction, the lower the temperature difference
becomes.
In South Asia, Indo-Gangetic is a classical area to investigate the effect of
irrigation. Vimal et al. (2020) found the heat stress in highly irrigated areas decreased
by the cooling effect using in situ observations, reanalysis, and high-resolution Weather
Research and Forecasting (WRF) model driven by ERA5 with and without irrigation. In
figures 2a and 2b, the irrigation hot zone is located in the red contour which is cooler
than other places influenced by global warming. In figure 2d, the surface temperature
gets lower, when the irrigated area gets higher.
It is becoming difficult to ignore the effect of global warming and extreme events.
Wim et al. (2020) also explored the probability of hot extremes in 3 kinds of scenarios:
all forcing without irrigation, irrigation, and all forcing. Figure 3 displays the
probability ratio of over the 99th percentile of high temperature. In these figures, note
that irrigation is an alleviated factor in the hot extreme, especially in Indo-Gangetic
Plain. Therefore, almost all regions suffer from a high probability of hot extremes
because of global warming except for Indo-Gangetic Plain. According to their
simulation, around one billion people were less exposed to high temperatures due to the
irrigation cooling effect.
1.2 Wet-bulb temperature
More and more researchers highlight the mortality attributed to the heat wave. For
instance, Vicedo et al. (2021) found that 37.0% of heat-related deaths are related to
anthropogenic climate change, which it is significant in every continent. They use the
cutting-edge time-series regression to get the relationship between mortality and
temperature at 731 stations, then project to the natural-only scenario. In this study, they
don’t apply specific humidity in their regression.
On the other hand, Vimal et al. (2020) calculated the mortality using dry-bulb
temperature and wet-bulb temperature in India as figure 4. The correlation coefficient of
mortality over the Tw plot is larger than mortality over the T plot, which means Tw is a
better indicator of understanding the linkage between heat stress and mortality in India.
In addition, Steven et al. (2010) argued that a resting human body (except for
absorbed solar heating) generates about 100W of metabolic heat that three main
processes must carry out, that is, evaporation cooling, heat conduction, and infrared
cooling. However, no matter how the object is wet or well-ventilated, the second law of
thermodynamics does not allow the heat to be taken away from an object when the
environment Tw is higher than the object’s temperature. As far as humans are
concerned, humans maintain a core body temperature of about 37℃ and the skin
temperature must be lower than the core temperature (about 35℃) to take away the
metabolic heat (McNab et al, 2002). According to the above discussion, they conclude
that humans cannot stay in an environment where Tw is over 35℃ for a long period.
This is also explored by Dunne et al. (2013), who indicated the wet-bulb temperature
could also affect the labor capacity. Besides, Krakauer et al. (2020) conducted climate
simulations to show that irrigation not only increases humidity but also worsens heat
stress.
Therefore, based on Steven’s idea, Im et al. (2017) plot the spatial distribution of
the highest daily maximum wet-bulb temperature which is defined as 6 hours window
average. Most of the high Tw is located in Asia. They show that Tw in the Persian Gulf
and the Red Sea can get over 31℃, and in Indo-Gangetic can get to 28.5℃, as figure 5.
In addition, the above findings are similar to Raymond et al. (2020) using ERA-Interim
reanalysis.
Focusing on the Indo-Gangetic area, Im et al. (2017) projected the daily maximum
wet-bulb temperature using Representative Concentration Pathway (RCP) 4.5 and 8.5
(Figure 6). In this figure, Twmax in the Indo-Gangetic area might get to 31℃ under RCP
4.5 and 35℃ under RCP 8.5. That is to say, around 4% of people will experience Twmax
exceeding 35℃ by 2100. However, this effect doesn’t exist only in Indo-Gangetic
Plain. Kang and Eltahir (2018) conducted a Business-As-Usual (BAU) scenario of
greenhouse gas to show that the wet-bulb temperature would increase by 3.4 (K) under
irrigation run, which is larger than the control simulation by 1 (K) in the North China
Plain.
Vimal et al. (2020) included the idea of wet-bulb temperature and irrigation
cooling effect in their study. They calculated the differences in the summer energy
budget, temperature, specific humidity, sea level pressure (SLP), and boundary layer
height (PBL height). Figures 7a to 7c explain the energy flux changes. Because of more
soil water, the evaporation flux goes up and takes away the heat. At the same time, the
temperature gradient between surface air and soil decreases. Therefore, sensible heat
flux decreases. The energy anomaly goes away from land totally because of the larger
flux of evaporation. Thus, increased latent heat flux and evaporation lead to increased
evaporation cooling and reduced land surface temperature and 2m temperature.
Moreover, Vimal et al. (2020) proposed a mechanism to explain the increase in
specific humidity. As mentioned above, irrigation changes the heat flux. The reduction
in sensible heat flux makes surface air cool, and descending air causes an increase in sea
level pressure and the development of anticyclonic circulation (figure 7g). Therefore,
the PBL collapses (figure 7h) and a shallower PBL leads to an increase in the low-level
moist enthalpy and specific humidity (figure 7f). According to their calculation, the
Indo-Gangetic area temperature decrease. Heat index and wet-bulb temperature increase
due to irrigation (figure 8 d-f). They concluded that the decline in dry heat and increase
in moist heat in Indo-Gangetic Plain can be attributed to the combination of large-scale
climate warming and the localized effect of irrigation.
1.3 Scientific questions and hypotheses
According to the discussion above, we can conclude some important characteristics
of T and Tw affected by irrigation. First of all, irrigation is a cooling effect on T. Also, it
can moisturize the surface air by PBL collapse and the specific humidity goes up. Thus,
irrigation on wet-bulb temperature is a competing effect. Irrigation is an increasing
factor in Tw from April to May (Vimal et al. (2020)). However, what scenario can cause
the moistening or cooling effect to dominate? Will irrigation worsen or alleviate the
extreme on Tw?
To understand the characteristic of Tw, we take figure in Roland et al. (2011) as a
reference (figure 9). We can see that when RH is low and T is high, the Tw curve is
more horizontal than in other conditions. That is to say, when increasing RH, it is easier
to increase Tw. Because of this characteristic, the cooling effect is not easy to apply on
Tw. On the other hand, if RH is high, the Tw curve is more vertical, so the moistening
effect is not apparent as the cooling effect.
In a nutshell, our hypotheses are as below. If background RH is low and
temperature is high, the irrigation moistening effect dominates. However, if background
RH is high, the irrigation cooling effect dominates.
Chapter 2. Data of the coupled climate model and method
2.1 Data
We use the same data as used in Wim et al. (2020), which is a fully coupled CESM
model (version 1.2). Also, version 4.0 of the Community Land Model (CLM) represents
the land surface in CESM1 (Wim et al. (2020)). Vegetation state and soil moisture
content in one soil column for irrigated crops calculate the irrigation module. In CLM
4.0, greenhouse gas concentrations and satellite-derived vegetation phenology are
considered. The resolution of all simulations is 0.9°x1.25°. Here, we use three main
variables. Daily maximum temperature on reference height (Tref), daily mean relative
humidity on reference height (RHref), and daily mean surface pressure to calculate the
specific humidity (or mixing ratio) and wet-bulb temperature. In Community
Atmosphere Model (CAM), the reference height is 2m height. Note that the 2m height
temperature and specific humidity (or relative humidity) in this model are calculated by
the Monin-Obukhov similarity theory (David et al. 2019). The theory indicates that the
temperature and humidity profiles in the boundary layer are determined by the Monin-
Obukhov length (L) which is related to the stability of the atmosphere and the shear
wind. In specific, the 2m height temperature is related to the gradient of potential
temperature between the atmosphere and surface and the Monin-Obukhov length (L)
(David et al. 2019).
The reason why we use model data is that only the model can isolate the single
process, while observation data is a lumped signal.
The transfer of T, RH, and P to Tw is as below:
(1) Use the c-c equation to get saturated vapor pressure.
𝑒!(𝑇) = 𝑒!"exp .−𝐿# 𝑅#21
𝑇− 1 𝑇"45
( 1 )
(2) Use the definition of relative humidity.
𝑒 = 𝑅𝐻 ∗ 𝑒!
( 2 )
(3) Use the definition of mixing ratio.
𝑤 =0.622𝑒 𝑝 − 𝑒
( 3 )
(4) Calculate Tw using 15 iterations of bisection method.
𝑇$ = 𝑇%− 𝐿#
𝑅 .0.622
𝑝 𝐴𝑒𝑥𝑝 2− 𝐵
𝑇 − 𝑤45
( 4 )
where A and B are constants.
2.2 Method
Here, following Wim et al. (2020), we use the linear regression technique to isolate
an individual forcing from a lumped signal.
∆𝑇𝑋𝑚 = 𝛽&× ∆𝑓%''+ 𝛽(× 𝑙𝑎𝑡. +𝛽)× 𝑙𝑜𝑛. +𝛽*× 𝑒𝑙𝑒𝑣.
( 5 )
∆𝑓%'' represents the irrigation fraction difference between 1901-1930 and 1981-2010.
The longitude, latitude, and elevation terms represent other effects. The irrigation-
induced temperature change in the pixel ∆𝑇𝑋𝑚%'' can be obtained by taking out the
coefficient of ∆𝑓%'', which can be written as
∆𝑇𝑋𝑚%''(𝑖) = 𝛽&× ∆𝑓%''(𝑖).
( 6 )
We hypothesize that specific humidity and Tw may not depend on the locations. To
optimize the regression process and not take other variables into our calculation, we
need to investigate the CESM experimental simulations. Table 1 represents the CESM
productions.
Then, we can do some calculations between these four simulations. We assume the
linear responses when we decompose the irrigation and other effects for the multiple
linear regressions. The non-linear effect may exist in these analyses, but we make the
linear assumption to get the first-order characteristic of irrigation. The first one is
control minus 20cc, which represents the other forcing change between 1901-1930 and
1981-2010. The second one is irr minus 20cirr, which explains the other forcing plus
irrigation effect. These two results could be used to distinguish the lumped signal.
By doing so, we propose a new concept of regression technique as below:
𝛥𝑇𝑋𝑚%''+(",%'' = 𝛽&𝛥𝑓%''+ 𝛽(𝛥𝑇𝑋𝑚,-.+(",, + 𝑎&
( 7 )
𝛥𝑞𝑋𝑚%''+(",%'' = 𝛽&𝛥𝑓%''+ 𝛽(𝛥𝑞𝑋𝑚,-.+(",,+ 𝑎(
( 8 )
𝛥𝑇𝑤𝑋𝑚%''+(",%'' = 𝛽&𝛥𝑓%''+ 𝛽(𝛥𝑇𝑤𝑋𝑚,-.+(",,+ 𝑎)
( 9 )
where q represents specific humidity. The terms on the left-hand side are signals with
irrigation and other forcings at the same time. 𝛽&𝛥𝑓%'' is the irrigation effect and 𝛽(
terms are other forcings. That is to say, we can distinguish a lumped signal and use the
simulation without taking other variables into our calculation.
Chapter 3. Results of coupled climate model
Between 1901-1930 and 1981-2010, the irrigation fraction expanded in the Central
USA, Europe, South Asia, and North China. Figure 10 shows that Indo-Gangetic Plain
expanded up to 50-60 percent of irrigation, and North China expanded by 30-50
percent. Here, we divide these places as global land, Central USA (42.5°W-111.25°W,
31.57°N-42.88°N), North China (108.75°E-121.25°E, 27.80°N-41.00°N), Europe (0°E-
32.5°E, 35.34°N-56.07°N), and South Asia (68.75°E-90°E, 5.18°N-32.51°N).
Figure 11 shows when irrigation expands, Δ𝑇𝑋𝑚 decreases in all regions. If the
change in irrigated fraction equals 0, there is no irrigation expansion. We can see that
Δ𝑇𝑋𝑚 is about 0.5K which is the anthropogenic warming signal. If we focus on the
irrigation-induced effect (second column), Δ𝑇𝑋𝑚%'' decreases because of negative 𝛽&.
3.1 Global land
Figure 12 demonstrates the same idea as figure 11 except for the definition of
Δ𝑇𝑋𝑚. Here, we capture the change in average daily maximum wet-bulb temperature
during the hottest month of the year, named Δ𝑇𝑤𝑋𝑚. Then, we get the dry-bulb
temperature and specific humidity at the time of 𝑇𝑤𝑋𝑚. In the first row of figure 12,
global land Tw increases by 0.5K. T goes down as irrigation expands. As for specific
humidity, it increases by 0.0005 to 0.001 (kg/kg). The second row shows the irrigation-
induced effect. We can see that when Δ𝑓%'' equals 0.5, anomalous temperature
decreases to -0.5K. As for specific humidity, Δ𝑓%'' goes up and Δ𝑞 also increases
because of irrigation replenishing water into the soil and evaporating. The slope is about
5 × 10+*/0.5 (Δ𝑞/Δ𝑓%''). In this scenario, the Tw increases because the moistening
effect dominates.
To discern whether the moistening effect dominates, we introduce the
differentiating of the definition of Tw. Here, we assume that constant P is 1013 (hPa)
and 300 (K) into Tw in equation (13):
(1) The definition of Tw.
𝑇$ = 𝑇 −𝐿#
𝑅#.0.622
𝑝 𝐴𝑒𝑥𝑝 2− 𝐵
𝑇$− 𝑤45
( 10 )
(2) Differentiate it in both side.
𝑑(𝑇$) = 𝑑[𝑇 −𝐿#
𝑅#.0.622
𝑝 𝐴𝑒𝑥𝑝 2− 𝐵
𝑇$− 𝑤45]
( 11 )
(3) Use the chain rule to deal with the exponential term.
𝑑(𝑇$) = 𝑑𝑇 −𝐿#
𝑅#.0.622
𝑝 𝐴𝑒𝑥𝑝 2− 𝐵
𝑇$− 𝑤45 2𝐵
𝑇$(𝑑𝑇𝑤 − 𝑑𝑤4
( 12 )
(4) Distinguish each variable.
21 +𝐿# 𝑅#
0.622
𝑝 𝐴𝑒𝑥𝑝 2− 𝐵
𝑇$− 𝑤4 𝐵
𝑇$(4 dT/ = dT +𝐿# 𝑅#𝑑𝑤
( 13 )
The left-hand side of the factor is about 8.07. The error for the pressure is -0.97% when
decreasing 10hPa. It is so small that the assumption of constant pressure is reasonable.
As for all forcing, dw is about 0.001 (kg/kg). Then, when plugging these numbers into
equation (13) above then we get:
dT/ = 0.12dT + 0.67 ≅ 0.67~0.7
( 14 )
which is similar to the Tw in all forcing plots. Note that the effect of specific humidity
change on Tw is larger than the temperature change. The same concept can be applied in
irrigation induced plots to examine which one dominates when Δ𝑓%'' = 0.5, plug
𝑑𝑤 = 5 × 10+* given by figure 12e. Then we have:
dT/ = 0.12dT + 0.335.
( 15 )
The dT term cannot cancel out the dw term, so the moistening dominates and Tw
increases about 0.3 (K) at Δ𝑓%'' = 0.5.
3.2 Central USA
Figure 13 shows the change of Tw in the Central USA. In the first row, Tw
increases 0.5 (K) in all Δ𝑓%'' when the specific humidity increases as a constant and
dominates although dry-bulb temperature decreases about 0.5 (K) in 50% of irrigation
expansion. In the second row, we found that Tw is canceled out by the decreasing
temperature and increasing specific humidity. The main reason comes from the
irrigation fraction induced specific humidity change is lower than global land.
We can plot a diagram to discern whether dT or dw is dominated. Assuming that
constant P is 1013 (hPa), figure 14 shows the equivalent dry-bulb temperature effect on
Tw when the mixing ratio changes. That is to say, if the mixing ratio increases
0.2 × 10+) (kg/kg), it equals to an increase of 1 (K) of dry-blub temperature. If we
need to cancel out the moistening effect, the dry-blub temperature must decrease by 1
(K) so that the right-hand side of equation (13) equals zero. Another aspect of this
picture tells us how to discern which one is the dominant factor. The area above the blue
line means the temperature effect is over the moistening effect. Thus, if we want to get
the result that Tw is dominated by the cooling effect, the moistening of dw must as
small as possible. This can make the cooling signal not be canceled out by the
moistening effect.
3.3 North China
In the first row of figure 15, we can see the same effect as above. The specific
humidity change is about 10+) (kg/kg), so the Tw is dominated by the moistening
effect although the temperature decreases when irrigated fraction goes up. On the other
hand, the specific humidity induced by irrigation is small so the Tw has the same trend
as temperature. The specific humidity changes almost come from anthropogenic
forcing, not from irrigation expansion. This may be due to the idea of the c-c equation.
When anthropogenic forcing causes greenhouse gas emissions to increase, the
temperature goes up. Therefore, the air parcel could hold more water vapor than before,
which makes the specific humidity increase homogeneously.
3.4 Europe
Europe is a special case in these areas. In figure 16, we can find that the wet-bulb
temperature increases a lot because of the large change in specific humidity. This
change comes from irrigation induced effect; that is to say, adding water to soil is much
more effective in evaporating into the atmosphere, which causes the rapid increase of
specific humidity and decreases the temperature. The evaporation allows water vapor to
take away the heat and increase the enthalpy of the near-surface.
3.5 South Asia
In South Asia, it is a good case to discuss the irrigation effect on heat waves
because of the high temperature of the dry season and the wet season, from April to
May and July to August, respectively. In addition, the irrigation area in Indo-Gangetic
Plain is also larger compared to other places. Therefore, we discuss this place more
comprehensively to figure out the effect of irrigation on wet-bulb temperature.
First of all, we do the same analysis on the change in the wet seasons. In the
nutshell, during the wet seasons, the wet-bulb temperature increases by 0.5 (K) in figure
17c because of the change in specific humidity. But why the specific humidity doesn’t
go up when the irrigation fraction increases in figure 17b?
To figure out the reason, we distinguish the irrigation effect and other effects in the
map. Figure 18 shows the specific humidity change induced by other forcings. The
contour represents the irrigation fraction which illustrates that Northeast India is a
highly irrigated area. Interestingly, the specific humidity decreases in high and increases
in low irrigated areas. The mechanism comes from the irrigation cooling effect. When
the irrigation makes the surface temperature goes down, that means the air parcel cannot
hold as much water vapor according to the c-c equation. On the other hand, global
warming makes the temperature in other places rise. In summer, the monsoon brings
lots of moisture into India, so the higher temperature place can hold much more water
vapor. Therefore, the increasing moisture from irrigation and the decreasing moisture
from the cooling effect cancel out. That’s why the specific humidity in figure 17b stays
still as the irrigated fraction increases.
For the dry season in figure 19, we analyze the Tw change in South Asia from
April to May. We can see that the Tw change over 1 (K) as the irrigated change fraction
equals 0.3, which is much higher than the wet season. The reason also comes from the
increased specific humidity, which almost results from the irrigation. In this part, it is
important to understand that the range of the humidity can change a lot in the dry season
and it is from the irrigation expansion. On the other hand, the 10m high humidity is low
so the gradient of humidity increases, which makes the evaporation work effectively.
Although it effectively cools down the surface temperature, the cooling effect is little
compared to the moistening effect. Therefore, the heat wave in the dry season might be
dangerous for humans because the air parcel can have much more water vapor and
make the atmosphere humid leading to a higher Tw.
Figure 20 shows the irrigation effect of one month's average in South Asia. The
irrigation effect is the simulation from irr minus control in table 1. The criteria is RH
lower than one standard deviation. Note that the red dots represent temperature change
is the dominant factor. That is to say, the absolute value of the temperature difference is
larger than the mixing ratio.
According to our calculation, when the mixing ratio change is small compared to
the temperature change, the Tw change and T change are linear. The regression slope of
the red dots is 0.20 which is close to the theoretical value of 0.125. On the other hand, if
the mixing ratio change is over about 0.0003 (kg/kg), the moistening effect dominates
the region. This means the wet-bulb temperature change is positive no matter how the
dry-bulb temperature change is induced by the irrigation effect, so most of the black
dots are situated in the first and second quadrants in the left diagram due to the
moistening effect. Note that the red-to-black ratio is low so the possibility of
temperature change dominating in the dry condition is low. In addition, this is evidence
that the Tw is more sensitive to mixing ratio change.
Same as above, the red region represents the temperature change dominating the
area over the wet conditions (Figure 21). Comparing figure 20 and figure 21, the red
dots of dry-bulb temperature in dry condition skew to the negative value. However, in
wet conditions, the red dots of dry-bulb temperature do not skew to the negative value,
which means the temperature cooling effect does not exist in this scenario. This
phenomenon can be seen in figure 22a which is the probability density function of
temperature change. The definition of dry and wet conditions is the same as in figures
20 and 21. To confirm the significance, we set the hypothesis as below:
𝐻": 𝜇0'1 ≥ 0
𝐻&: 𝜇0'1 < 0
( 16 )
The one-tail test tells us that the z-value is low enough to reject the null hypothesis, and
the p-value is lower than 0.01. Therefore, the irrigation cooling effect affects the
temperature in dry conditions significantly. The mechanism might come from the
gradient of humidity between the soil and the atmosphere increasing so evaporation
works effectively compared to wet conditions.
To find the possibility of the temperature change dominating region, we do the
possibility density function of wet and dry conditions and the result is shown in figure
22b. First of all, the probability of wet conditions of mixing ratio changes close to 0
(kg/kg) is higher than in dry conditions. Also, we do the one-tailed z-test to verify the
red line skews to the positive value. The hypothesis is as below:
𝐻": 𝜇0'1 ≤ 𝜇$2-
𝐻&: 𝜇0'1 > 𝜇$2-
( 17 )
The z-value equals 5.0661 and the p-value is lower than 0.01, which means we have
99% confidence to reject the null hypothesis and the mean value of mixing ratio change
is significantly skewed to a positive value.
In addition, the variance is also important when we investigate the moistening
effect. Here, we do the one-tailed F-test to verify whether the variance of dry conditions
is larger than wet conditions. The hypothesis is as below:
𝐻": 𝜎0'1( ≤ 𝜎$2-(
𝐻&: 𝜎0'1( > 𝜎$2-( .
( 18 )
The F value is 6.9501 and the critical value is 1.0956, which means we have 99% of
confidence in rejecting the null hypothesis. In this analysis, we can conclude that the
distribution of dry conditions is wider than wet conditions. The probability of suffering
from an extreme event is higher than in wet conditions because of the skewed and wide
distribution. Therefore, the ratio of dots on the left-hand side of figure 20 in the first and
second quadrants to the third and fourth quadrants is larger than that in figure 21. Also,
some wet-bulb changes can reach up to 6 (K) in figure 20, which results from the
extreme value of the change of mixing ratio.
As the discussion above, it is interesting to investigate the mechanism of
evaporation influenced by the gradient of soil and air moisture. Here, we explore the
difference in latent heat flux change. As shown in figure 22c, the dry condition curve
skews to the positive value. This means the net effect of irrigation for dry conditions can
bring more evaporation, especially from April to May (as appendix A2). Note that the
appendix figures point out the dry seasons and wet seasons in South Asia. The number
in the parentheses denotes the number of dry or wet events from 1981 to 2010. From
April to May, the climate is hot and dry, which means the background relative humidity
is low. If we add water to the soil, the gradient of moisture goes up and makes
evaporation effective. On the other hand, we find that there is no apparent difference in
other seasons. In summary, the cooling effect of irrigation on dry-bulb temperature is
sensitive due to the effective evaporation. This is an important conclusion to discuss the
characteristic of temperature in each condition.
Chapter 4. Results of the offline land surface model
4.1 Data
In this chapter, we use version 5.0 of the Community Land Model (CLM) to
represent the land surface in CESM and apply offline simulation which spins up in 1901
for 5 years and duration between 1901 to 2014 (David et al. 2019). The resolution of all
simulations is 0.9°x1.25°. Here, to demonstrate the mechanism between evaporation
and irrigation, we use seven variables, which are ground temperature (Tg), 2m high
temperature (T2m), 2m high specific humidity (q2m), atmospheric specific humidity
(QBOT), sensible heat flux (SH), latent heat flux (LH), and wind speed (U). The reason
why we use the offline model is that it is clearer to know the mechanism of evaporation.
Compared to coupled model, the reaction on the ground would not influence back to the
atmosphere, so the change on the ground is simple to learn the pure process.
In the offline model, the irrigation is made once per day at 6 AM local time (David
et al. 2019). The model would calculate each pixel whether the irrigation moisture is
over the threshold. If it is over the threshold, the model keeps the pixel in the same soil
moisture. However, if the pixel’s moisture is under the threshold, the model irrigates to
the available moisture (David et al. 2019).
4.2 Comparison between offline and coupled model
Figure 23 shows the difference between the offline and coupled models. First of
all, the red line represents the coupled model temperature on the ground and the black
(dash) line represents the ground (2m high) temperature. The red line has higher annual
variability compared to the black line. This comes from the characteristic of the offline
model which is calculated by bounded atmospheric information. That is to say, the
irrigation and control run have the same forcing in the atmosphere to calculate the
variables on the ground. However, the coupled model means that the surface
information can influence the atmosphere by land-atmosphere interactions, so the
atmospheric information is not the same in each time step.
Also, the change of the black dash line is not as large as the solid line. This is
because the 2m high temperature is interpolated by the surface and bounded atmosphere
(David et al. 2019). Therefore, it includes the information of higher atmosphere which
is the same in control and irrigation run. Finally, during the daytime, evaporation works
more compared to night, so both offline and coupled models have a larger decrease in
temperature during the daytime. Again, in the CLM model, the 2m high temperature and
specific humidity are calculated based on the similarity theory assuming that the surface
heat between the surface and the bottom layer of the atmosphere model is a constant
with height (David et al. 2019).
4.3 The scatter plot in South Asia
Here, we plot the scatter plot as same in section 3.5. In figure 24 and 25, the
absolute value of the difference in mixing ratio is smaller than those in figure 20 and 21,
which comes from the same reason as section 4.2. The bounded atmosphere makes the
2m high mixing ratio change not as large as the coupled model. In addition, because the
surface information cannot be a forcing to the atmosphere, most of the mixing ratio dots
move to the positive axis when we turn on the irrigation due to the moisture going to the
atmosphere and not dissipated by the coupling process.
Overall, we can get the same result that the drier season in figure 25 makes
evaporation work efficiently to increase the mixing ratio and decrease the temperature
at the same time. However, most of the dots are black, which means the mixing ratio
change is a dominant factor as far as wet-bulb temperature is concerned.
Figures 26a and 26b are the conclusions above. The temperature and mixing ratio
change are narrowed to zero and shifted to the positive value compared to the coupled
model. Also, dry conditions have a larger tail of moisturized and temperature decrease
compared to wet conditions. Here, we get similar results as the coupled model. The
evaporation works efficiently in dry conditions so that the possibility density
distribution curve shifts to moistening and cooling values. However, evaporation
doesn’t work efficiently in wet conditions, so the mean value is close to zero and
symmetry to zero value.
4.4 The diurnal difference in the dry and wet seasons (South Asia)
Figure 27 shows the difference in temperature, wet-bulb temperature, and mixing
ratio in the dry season, respectively. On dry season days, the gradient of humidity
between irrigated surfaces and the atmosphere is large. Therefore, evaporation works
more effectively than the night and wet seasons. However, according to equation (13),
the moistening effect is much larger than the temperature decrease. Thus, the wet-bulb
temperature curve is similar to the mixing ratio change. Specifically, the curve of
mixing ratio change increased from 1906 to 2010 because the irrigation amount in
South Asia also increases and makes the evaporation go larger. During the nighttime,
the temperature decreases because of sunsets. Besides, the moisture in the atmosphere
goes higher from evaporation in the daytime. Therefore, the relative humidity is larger
at night than at daytime. Evaporation is less due to the lower gradient of the mixing
ratio between the surface and atmosphere and solar radiation. Thus, the red curve in
figure 27 has the same trend as the blue curve, but the absolute values are lower.
Figure 28 shows the difference in temperature, wet-bulb temperature, and the
mixing ratio in wet seasons, respectively. In these figures, the daytime and nighttime are
similar. This is because the background atmosphere is wet and the gradient between the
surface and atmosphere is small both day and night. Therefore, the wet-bulb
temperature doesn’t increase as large as in dry seasons but still has the same trend.
4.5 The yearly trend of wet-bulb temperature in South Asia and North China
In this section, we explore six variables to deeply understand the mechanism of
irrigation moistening by comparing two different regions, South Asia and North China.
Figure 29 shows the difference in variables due to irrigation. Figure 29d illustrates the
irrigation amount increase over time. This is due to the increasing population and the
food demand getting higher, particularly after the 1950s (Siebert et al. 2015, Guo et al.
2022). Because of these higher amounts of water in the soil, the specific humidity near
the surface gets higher, making evaporation efficient. Accordingly, we calculate the
correlation coefficient of irrigation amount to temperature, wet-bulb temperature,
mixing ratio, and latent heat flux. The coefficients are -0.993, 0.994, 0.997, and 0.991,
respectively. Therefore, the relations between these variables are highly correlated.
To analyze what makes the evaporation work, we investigate the bulk formula
which is the concept of the mixture in the boundary layer (Lawrence et al. 2019) as
below:
(1) The definition of the bulk formula:
𝐿𝐻 = 𝜌𝐿3𝐶4𝑈&"(𝑞!− 𝑞5)
( 19 )
(2) Differentiate it on both sides:
𝑑𝐿𝐻 ∝ 𝑈&"𝑑(𝑞!− 𝑞5).
( 20 )
Therefore, the difference in latent heat flux between the two models is in proportion to
the specific humidity difference. Figure 29f shows the decomposition of latent heat flux.
The main reason that causes the evaporation to work is the difference of specific
humidity in the atmosphere and surface, here we call (𝑞! − 𝑞5)!67 afterward and the
subscript means the simulation model. When irrigation turns on, the (𝑞!− 𝑞5)%'' goes
higher because the specific humidity gets larger near the surface. Also, the key factor
𝑑(𝑞!− 𝑞5) can be written as:
𝑑(𝑞!− 𝑞5) = (𝑞!− 𝑞5)%''− (𝑞!− 𝑞5),-. = (𝑞!)%''− (𝑞!),-.
( 21 )
Note that this is a bounded model, so the specific humidity of the atmosphere is the
same in both models. Note that figure 29f is similar to figure 29e but not the same. This
comes from the coefficient ahead in equation (19) is not the same when the model does
the irrigation.
In plain language, the mechanism is that evaporation makes the surface moisturize,
which causes evaporation to work efficiently. This gives rise to latent heat flux and the
mixing ratio goes high. Also, the temperature goes down due to the cooling effect. The
result of wet-bulb temperature obeys the mixing ratio curve because the moistening
effect is the key faction to dominate.
Figure 30 is the profile of North China, the irrigation amount is less than South
Asia by one order. Thus, the evaporation-related variables change less. However,
although the irrigation amount increases over time, latent heat flux decreases between
1920 to 1950 and 2000 to 2010. Figures 31a and b show the specific humidity at 10m
and 2m high. It grows about 10-3 (kg/kg) and 10-4 (kg/kg) from 1920 to 1950 and 2000
to 2010. That is, the background atmosphere is moisturized. According to our discussion
above, the evaporation is less because of the specific humidity difference in the two
models, that is, (𝑞!)%'' minus (𝑞!),-., gets lower. Therefore, the related variables (i.e.
wet-bulb temperature, mixing ratio, etc.) have the same trend as latent heat flux
although the irrigation amount remains nearly the same.
Also, we find that the irrigation amount increases rapidly from 1950 to 1960. This
might come from two factors at the same time. The first one is the green revolution
which is the food demand from the increasing population. The second one is the
background atmosphere becomes dry. Figure 31a shows the decreasing specific
humidity in the upper layer from 1950 to 1960. That is to say, the model is forced to
irrigate this region due to the increasing irrigation fraction and the dry climate.
Therefore, the latent heat flux increases rapidly because the specific humidity near the
surface gets much higher compared to the control simulation. Therefore, the mixing
ratio and wet-bulb temperature increase a lot over this period.
Chapter 5. Discussion
Here, we revisit our hypothesis. If the background RH is low, the irrigation
moistening effect would dominate. Due to the higher standard deviation of the mixing
ratio changing, the Tw might have a higher chance of extremes. On the other hand, if
the background RH is high, the evaporation is less from the lower water gradient.
Therefore, there is no apparent cooling or moistening effect to alter the Tw.
Some may argue that the Tw is so low in the dry condition that we should not
concern about heat stress here. We analyze the Tw mean and extreme value shown in
figure 32. Figures 32a and 32b illustrate the Tw average in dry and wet seasons in India,
respectively. Indeed, the heating center (figure 32e) is located in the apparently low Tw
area. However, we consider the most dangerous place to be located in the transient zone
(at the edge of the heating center) where the Tw is not low but the evaporation still
works. In this area, the moistening effect still exists to increase the Tw. On the other
hand, figure 32g shows the maximum Tw difference (irrigation run minus control run),
the edge of the heating center makes the extreme value in figure 32c increase nearly the
same as the Tw maximum in the wet season in figure 32d. Also, in figures 32f and 32h,
the difference between control and irrigation run in the wet season is small compared to
the dry condition. Therefore, this is evidence that the moist heat stress is a concern
when it comes to the mean and extreme value in the dry condition in the Indo-Gangetic
Plain.
In our simulations, the irrigation amount comes from the threshold of soil
moisture. If the soil moisture is lower than the threshold, the model irrigates the pixel to
the available amount which means the soil cannot hold any more water. Jha et al. (2022)
argue that the irrigation amount in the dry season in India is overestimated by the CLM
model because of the policy of the Indian government to preserve the water. They use
census-based data from satellite observation and find that the heat stress is
overestimated by 4.9 times in the model. In our perspective, we use the CLM model to
know that evaporation is a key factor controlling wet-bulb temperature and heat stress.
Irrigation can really make the regional wet-bulb temperature increase due to the higher
humidity. However, after exploring its mechanism, we could combine the census-based
irrigation amount to analyze the effect of irrigation in the real world. This can be
investigated in the future so that the climate in the irrigated world would not be
overestimated.
In addition, several studies estimate that irrigation inputs more water vapor into the
atmosphere, which makes the surface warming temperature be overlooked (Kenny &
Hodzic, 2019). Li et al. (2022) used two different assumptions to simulate with or
without global irrigation. Their results show that the temperature in both day and night
time drops due to irrigation in the first case. This comes from the reduction of
shortwave radiation by the increasing mid-level cloud, which is consistent with our
simulation and assumption. However, they argue that moisture in the atmosphere is a
greenhouse gas that makes the reflected longwave radiation increase and makes the
surface temperature goes up at the same time. From our viewpoint, this is truly a new
discovery that is different from most previous research. But for the wet-bulb
temperature, although the dry-bulb temperature goes up, the wet-bulb temperature
would not change a lot because the temperature effect is not large.
We consider the CTX method (Perkin and Alexander, 2013) to be a way to measure
the frequency and strength of heat waves when it comes to moist heat stress. They
define the threshold of each day by taking the 90th percentile of Tmax using a 15-day
running window for 46 years. The reason why they use the running window is that the
climate signal may advance or delay on some days. By doing so, a window average is a
better index to represent a climate condition. CTX gives us a benchmark to analyze the
characteristic of heat waves and this can be applied to how the Tw could be affected
extremely under the moistening effect.
For the farmers in the Indo-Gangetic Plain, the seasonal to annual projection is
very important to avoid exposure to heat stress. Climate Services Toolbox (CSTools) is
a useful tool based on R packages to visualize climate information (Núria et al., 2022).
They use state-of-art methods to downscale and plot the probability density function to
present the data. By using this tool, we can evaluate the moistening effect on Tw to alert
the farmer not to expose to the heat condition with higher resolution.
Ultimately, our results are consistent with Guo et al. (2022) using ERA-5
reanalysis data. They argue that the ERA-5 data is assimilated with observed datasets
every 6 hours so it is reliable data. We consider that the ERA-5 data is similar to our
offline control simulation. They explore the wet-bulb globe temperature to demonstrate
the heat stress due to shortwave radiation. This can be our future work to investigate the
wind speed and shortwave radiation in the CLM model combining census-based
irrigation data. By doing so, we can separate the irrigation effect from the lumped signal
and get more detailed information about heat stress and comfort.
Chapter 6. Conclusions
In this study, we use NCAR CESM and CLM models to assess the near surface
climate changes. The impact of irrigation on wet-bulb temperature from cooling and
moistening effects is explored. We use the technique of linear regression to distinguish
the effect of global warming and irrigation. Overall, irrigation is a cooling effect in the
area we choose. As for wet-bulb temperature, there are two competing effects. The
moistening and the irrigation cooling effects can increase and decrease the wet-bulb
temperature, respectively. However, the background climate conditions could be a key
factor in wet-bulb temperature change due to the amount of evaporation.
As shown in figure 33, if the background relative humidity is wet, there are two
scenarios. (a) If the wet-bulb temperature is controlled by dry-bulb temperature, the
irrigation cooling effect doesn’t exist apparently because the mean of latent heat flux
change is nearly zero. (b) If it is controlled by the mixing ratio, the wet-bulb
temperature would be divided into two parts. As the mixing ratio increases, the wet-bulb
temperature also goes up, and vice versa.
From our simulations, if the background is dry, the wet-bulb temperature is more
likely controlled by the mixing ratio change. However, if the wet-bulb temperature is
still controlled by the dry-bulb temperature (the mixing ratio change is low enough), the
irrigation cooling effect exists because the latent heat flux change is positive, which
makes the wet-bulb temperature drop. On the other hand, if the wet-bulb temperature is
controlled by the mixing ratio, the wet-bulb temperature is more likely to increase
because the distribution of the mixing ratio change skews to a positive value and the
standard deviation is larger than that in the wet condition.
In the nutshell, the cooling effect and moistening effect are not apparent due to the
less evaporation in the wet condition. On the other hand, irrigation might worsen
comfort in view of the wet-bulb temperature, which is consistent with the result of
Mishra et al. (2020). Ultimately, the wet-bulb temperature is so sensitive to the mixing
ratio change that it is a critical factor when we focus on comfort and moist heat stress.
Chapter 7. References
[1] Chou, Chihchung, et al. "Irrigation-induced land–atmosphere feedbacks and their
impacts on Indian summer monsoon." Journal of Climate 31.21 (2018): 8785-8801.
[2] David M. Lawrence, Rosie A. Fisher, Charles D. Koven, Keith W. Oleson, Sean C.
Swenson, Gordon Bonan, Nathan Collier, Bardan Ghimire, Leo van Kampenhout,
Daniel Kennedy, Erik Kluzek, Peter J. Lawrence, Fang Li, Hongyi Li, Danica
Lombardozzi, William J. Riley, William J. Sacks, Mingjie Shi, Mariana Vertenstein,
William R. Wieder, Chonggang Xu, Ashehad A. Ali, Andrew M. Badger, Gautam Bisht,
Michiel van den Broeke, Michael A. Brunke, Sean P. Burns, Jonathan Buzan, Martyn
Clark, Anthony Craig, Kyla Dahlin, Beth Drewniak, Joshua B. Fisher, Mark Flanner,
Andrew M. Fox, Pierre Gentine, Forrest Hoffman, Gretchen Keppel-Aleks, Ryan Knox,
Sanjiv Kumar, Jan Lenaerts, L. Ruby Leung, William H. Lipscomb, Yaqiong Lu,
Ashutosh Pandey, Jon D. Pelletier, Justin Perket, James T. Randerson, Daniel M.
Ricciuto, Benjamin M. Sanderson, Andrew Slater, Zachary M. Subin, Jinyun Tang, R.
Quinn Thomas, Maria Val Martin, Xubin Zeng, 2019: The Community Land Model
version 5: Description of new features, benchmarking, and impact of forcing
uncertainty. Journal of Advances in Modeling Earth Systems, 11, 4245-4287. https://doi-
org.cuucar.idm.oclc.org/10.1029/2018MS001583.
[3] Dunne, John P., Ronald J. Stouffer, and Jasmin G. John. "Reductions in labour
capacity from heat stress under climate warming." Nature Climate Change 3.6 (2013):
563-566.
[4] Guo, Qiang, et al. "Irrigated cropland expansion exacerbates the urban moist heat
stress in northern India." Environmental Research Letters 17.5 (2022): 054013.
[5] Huachen Li, Min-Hui Lo, Donfryeol Ryu, Murray Peel, Yongqiang Zhang, Possible
increase of air temperature by irrigation.
[6] Im, E.S., Pal, J.S. and Eltahir, E.A. Deadly Heat Waves Projected in the Densely
Populated Agricultural Regions of South Asia. Science Advances, 3, No. 8. (2017).
[7] Jha, R., Mondal, A., Devanand, A., Roxy, M. K., & Ghosh, S. (2022). Limited
influence of irrigation on pre-monsoon heat stress in the Indo-Gangetic Plain. Nature
communications, 13(1), 1-10.
[8] Kang, Suchul, and Elfatih AB Eltahir. "North China Plain threatened by deadly
heatwaves due to climate change and irrigation." Nature communications 9.1 (2018): 1-
9.
[9] Kennedy, Ivan, and Migdat Hodzic. "Testing the hypothesis that variations in
atmospheric water vapour are the main cause of fluctuations in global
temperature." Periodicals of Engineering and Natural Sciences (PEN) 7.2 (2019): 870-
880.
[10] Krakauer, Nir Y., Benjamin I. Cook, and Michael J. Puma. "Effect of irrigation on
humid heat extremes." Environmental Research Letters 15.9 (2020): 094010.
[11] Lawrence, David M., et al. "The Community Land Model version 5: Description of
new features, benchmarking, and impact of forcing uncertainty." Journal of Advances in
Modeling Earth Systems 11.12 (2019): 4245-4287.
[12] McNab, Brian Keith. The physiological ecology of vertebrates: a view from
energetics. Cornell University Press, 2002.
[13] Mishra, V., Ambika, A.K., Asoka, A. et al. Moist heat stress extremes in India
enhanced by irrigation. Nat. Geosci. 13, 722–728 (2020).
[14] Pérez-Zanón, Núria, et al. "Climate Services Toolbox (CSTools) v4. 0: from
climate forecasts to climate forecast information." Geoscientific Model
Development 15.15 (2022): 6115-6142.
[15] Perkins, Sarah E., and Lisa V. Alexander. "On the measurement of heat
waves." Journal of climate 26.13 (2013): 4500-4517.
[16] Raymond, Colin, Tom Matthews, and Radley M. Horton. "The emergence of heat
and humidity too severe for human tolerance." Science Advances 6.19 (2020):
eaaw1838.
[17] Roland, S. Wet-Bulb Temperature from Relative Humidity and Air Temperature.
Science, 50, 2267-2269. (2011).
[18] S. C. Sherwood, M. Huber, An adaptability limit to climate change due to heat
stress. Proc. Natl. Acad. Sci. U.S.A. 107, 9552–9555 (2010).
[19] Siebert S, Kummu M, Porkka M, Döll P, Ramankutty N and Scanlon B R 2015 A
global data set of the extent of irrigated land from 1900 to 2005 Hydrol. Earth Syst. Sci.
19 1521–45
[20] Thiery, W., Visser, A.J., Fischer, E.M. et al. Warming of hot extremes alleviated
by expanding irrigation. Nat Commun 11, 290 (2020).
[21] Vicedo-Cabrera, A.M., Scovronick, N., Sera, F. et al. The burden of heat-related
mortality attributable to recent human-induced climate change. Nat. Clim.
Chang. 11, 492–500 (2021).
[22] Wada, Yoshihide, et al. "Human water consumption intensifies hydrological
drought worldwide." Environmental Research Letters 8.3 (2013): 034036.
Time series
Natural forcing
Irrigation effect
Other forcing
20cc 1901-1930 Yes No No
20cirr 1901-1930 Yes Yes No
Control 1981-2010 Yes No Yes
Irr 1981-2010 Yes Yes Yes
Table 1. CESM simulation production.
