The role of irrigation in the earth system: Past, present, and future

As the biggest freshwater-use practice, irrigation is an important climate forcing, with
its cooling effects on the near-surface temperature being highlighted by previous studies.
Since the year 1901, area equipped for irrigation has experienced pronounced expansion,
from 0.5 x 106 km2 to a bit less than 2.8 x 106 km2 in the year 2014, and observations
revealed a global irrigation water withdrawal from 2500 to 3000 km3 per year currently. Ir-
rigation practices reallocate the water resources from surface and groundwater to cropland,
leading to changes in the water cycle, energy balance, and carbon exchange, which in turn
cause alterations in the near-surface temperature. Constrained by the temporal and spatial
limitations of observations, Earth system models are widely used to quantify these impacts,
but their irrigation representations vary. There is consistency among previous studies that
irrigation could alleviate heat extremes, but its impacts on other variables lack agreement,
which may be attributed to different study regions, periods, and irrigation parameterizations
in Earth system models. This study aims to address three research frontiers in irrigation-
related studies: (i) huge inconsistency among Earth system modelling-based studies; (ii)
over-simplified irrigation representation in Earth system models; and (iii) limited explo-
rations of irrigation-induced impacts under future scenarios, to achieve a comprehensive
understanding of irrigation’s role in the Earth system.
First, a model intercomparison project is initiated that aims to quantify the impact of ir-
rigation expansion on historical climate (IRRMIP). Three Earth system modelling groups
participate in this study by simulating two experiments: one historical set of simulations
with transient irrigation extent and one where the irrigation extent is fixed at the level of the
year 1901. The results are analysed to isolate the impact of irrigation expansion on moist-
heat stress. Although many efforts have been made to assess irrigation’s influence on the
Earth system, most studies assume static irrigation extent and none considered multiple
models. In other words, most of the studies were conducted in a single model context with
a static irrigation extent, making the comparison between these studies complicated. All
Earth system model groups (us included) performed simulations with the same temporal
coverage, input surface data, and output variables. The results indicate that the considered
Earth system models tend to underestimate the irrigation water withdrawals. Based on the
simulated 2-meter air temperature, relative humidity, and wind speed, we calculate several
moist-heat metrics including apparent temperature and wet bulb temperature. Multi-model
results reveal that irrigation reduces the frequency of heat events (exceeding 99.0%, 99.5%,
vii
viii
and 99.9% percentile value of 2-meter air temperature) by more than 4 times over the grid
cells with substantial irrigation expansion, and even reverses the warming trend by reducing
the hours/year exposed to these events. In contrast, the impact of irrigation expansion on
moist-heat events (apparent and wet bulb temperature) is negligible or even tends towards
increased heat stress. These results reveal a general over-optimism regarding the irriga-
tion’s cooling impacts in the context of human health, and highlight the need to critically
scrutinize irrigation as a possible future heat extreme adaptation strategy.
Second, we developed a new irrigation module for the Community Earth System Model
version 2 (CESM2) consisting of four different irrigation techniques, to address the un-
derestimation in irrigation water withdrawal indicated by IRRMIP outputs. While CESM
has been frequently used in irrigation-related studies and has shown a satisfying skill, its
representation of irrigation water demand is idealised, causing the model to underestimate
withdrawals. To address this issue, four irrigation techniques are designed: drip, sprin-
kler, flood, and paddy, each with its own irrigation trigger, water amount, water application
technique, and surface water ponding option. The new parameterisation is tested in a set of
simulations that employs existing spatial information on the occurrence of these techniques.
Simulated irrigation water withdrawal is higher in the new irrigation scheme, thereby sub-
stantially reducing the bias over the USA (-10.58 to -0.03 km3 year−1) and other countries
(-64.74 to -7.67 km3 per year) but causing a slight overestimation over China. At three
cropland sites, results show that sprinkler and flood irrigation have little impact on surface
fluxes, but paddy irrigation substantially increases model’s performance in terms of evapo-
transpiration estimation. When comparing irrigation-induced impacts on water and energy
fluxes at a global scale, the new irrigation scheme shows a more pronounced influence for
most variables, making it a more suitable tool for assessing irrigation-related impacts.
Finally, the climatic impact of irrigation is projected under a range of climate and land-use
scenarios. To our knowledge, none of the previous studies assessed irrigation’s climatic
impacts under future scenarios while updating irrigation extent as well as the irrigation
method, but the creation of socioeconomic scenario-related datasets, including greenhouse
gas emission and land use, makes it possible. To this end, (i) a new dataset of future
irrigation techniques share under varying shared socioeconomic pathways is generated, (ii)
CESM2 is further developed to represent multiple irrigation techniques for one crop type
within one grid cell, and (iii) fully-coupled simulations are performed with the improved
CESM2 model both in historical (1985-2014) and future (2015-2049) periods. The design
of projected irrigation techniques share follows the general assumption: a country with
higher socioeconomic capacity and more severe water scarcity issues holds a faster speed
in irrigation system update. Based on this dataset, 2/3 of the global irrigation extent will
be equipped with drip or sprinkler irrigation under SSP1-1.26, and this value for SSP3-
7.0 and SSP5-8.5 is less than 1/2 and less than 2/3, respectively. Necessary modification
is made in CESM2 to account for various irrigation techniques in one grid cell based on
the irrigation module developed before, and then simulations with and without irrigation
are conducted both for the historical (1985-2014) and future (2020-2049) period. Outputs
reveal that irrigation still has cooling impacts on 2-meter air temperature, but the impacts
on apparent and wet bulb temperature become less pronounced. Even for the 2-meter
air temperature, irrigation is unable to reverse or alleviate the warming signals caused by
other forcings. These results highlight the limited potential of irrigation as a local climate
adaptation strategy and underscore the importance of reducing greenhouse gas emissions
as a prime mitigation strategy to reduce escalating heat stress.