A review of evapotranspiration estimation methods for climate-smart agriculture tools under a changing climate: vulnerabilities, consequences, and implications

Climate change (CC) is the quick reply of nature resulting from ongoing human activities on Earth, notably the burning of fossil fuels, deforestation, and industrial processes that lead to significant and long-lasting changes in global weather patterns over time. CC has many adverse effects on Earth's environment, including regional temperature variations, increasing sea levels, more frequent and severe storms, changes in regional precipitation patterns, and the onset of rapid heat waves and drought events. These changes have profound implications for Earth's environment and in several fields as diverse as agriculture, public health, water use, energy production, and biodiversity. Therefore, CC is one of our biggest global challenges, and urgent action must be taken to reduce the usage of burning fossil fuels, deforestation, and industrial processes (Phan et al. 2024; Rawat et al. 2024).

Water is the most fundamental resource in agriculture. Studies reported that in the coming time, agricultural water users can face numerous challenges in accessing cost-effective and equitable water services based on their needs. Most countries with significant water resources could face water shortages shortly (Mishra et al. 2021; Salehi 2022; Lakhiar et al. 2024b). Because water insecurity is one of the essential effects of CC, which can negatively affect food security, agricultural productivity, and prices of several farm inputs, these factors can ultimately affect SDGs and increase poverty and inequality. Some figures suggest that the requirements for freshwater are projected to rise to 80% by 2050 compared to present figures (Flörke et al. 2018; World Bank 2020).

The water cycle in the atmosphere is one of the significant interdisciplinary, integrated, and balanced approaches that maintain water availability on the Earth. It is a vital biosphere component and is susceptible to CC. It has a significant role in retaining the balance of the natural ecosystem and promoting human development globally (Zhu et al. 2018). Abed-Elmdoust et al. (2016) and Sarker (2022) indicated that spatiotemporal patterns of climate variables have severely disrupted the geomorphology of watersheds and reshaped the water bodies networks due to land erosion and the emergence of the fluvial regime. Therefore, early information on climate parameters is crucial for understanding and dealing with the long-term effects of CC on watershed management. This early data accessibility can provide higher forecasting efficiency, distribution of resources, early mitigation of flood risks, and ultimate watershed stability. Generally, the balance between rainwater and evapotranspiration (ET) in watersheds controls the water available to replenish groundwater, streamflow, and reservoir storage. The accurate information about this water balance could help early forecast and alleviate water scarcity challenges, including agricultural and ecological water availability, flood risk assessment and planning, and efficient watershed infrastructure development (Abed-Elmdoust et al. 2016; Sarker 2022; Singhal et al. 2024). In addition, studies stated that research on the water cycle engages all the elements of agro-meteorology and hydrology parameters, together with precipitation (P), runoff, evaporation (E), transpiration (T), groundwater, and soil moisture condition (Tang et al. 2016; Cui et al. 2018). ET is considered an essential part of the continental water cycle, and its process is opposed to P. It was shown that ET returns about 60% of the land's rain to the environment (Trenberth et al. 2009; Lahoz & De Lannoy 2014). It is also a water replacement practice among the atmosphere, biosphere, and hydrosphere layers. Its accurate assessments are essential for observing drought conditions and protecting the ecological environment in any region. Thus, investigating ET and studying its variations through research studies hold both practical and theoretical significance in understanding how CC may impact the water cycle system in the future (Yu et al. 2021). Furthermore, the knowledge of ET can support growers in meeting their cultivation targets, improving water use efficiency (WUE), increasing crop yields, reducing energy consumption, and reducing associated environmental emissions.

The above discussion was the primary motivation for writing this review article. CC is currently related to its worldwide reach and significantly disturbing many areas. Thus, we wanted to investigate and see the effect of CC on ET and the significance of accurately estimating ET. To achieve the proposed study object, we posed the below specific questions:

• (1) How is CC affecting the world water cycle?

• (2) How does CC affect agro-metrological parameters and primary crop yields?

• (3) How can we deal with the increasing CC issue?

• (4) What is the current status of ET estimation techniques, and why is accurate estimation necessary for efficient irrigation water management?

Before answering the above questions, we searched the online literature database for studies discussing similar subjects. We saw that ET-based studies are currently widely conducted to understand the mechanisms and opposite effects of environmental parameter variations on the water cycle (Dong et al. 2020). Previously, numerous review articles were published on ET estimation using different concepts. Still, no study has briefly discussed and comprehensively summarized an overview of the importance of ET estimation in the context of climate-smart agriculture (C-SA) tools and its relevance in addressing the challenges posed by CC. For example, Anderson et al. (2024) provided a brief history of the thermal Infrared (IR)-based two-source energy balance (TSEB) model –diagnosing ET from plant to global scales. Derardja et al. (2024) reviewed the development and history of temperature-based ET models for remote sensing estimations. Amani & Shafizadeh-Moghadam (2023) published a review study and summarized the results of machine learning models and influential factors for ET estimation using remote sensing and ground-based data. Pinos (2022) summarized empirical and semi-empirical-based ET estimation models. Aryalekshmi et al. (2021) published a review that considered various surface energy balance remote sensing-based ET estimation models. Wang et al. (2021) presented an overview of the main influencing parameters that affect ET in existing ET models for controlled environment agriculture. Ren et al. (2021) summarized the results of irrigation based on daily weighted ET methods.Ghiat et al. (2021) reviewed the major mechanistic and empirical models (five empirical models) for estimating ET for both open and closed fields. Xiang et al. (2020) provided a comprehensive overview of the concepts, developments, equations, applications, and similarities between potential ET and reference crop ET. Cascone et al. (2019) reviewed the cooling effect of green roofs due to the ET process. Ebrahimian et al. (2019) investigated the role of ET in runoff volume reduction of green roofs and rain gardens through a comprehensive literature review. Hu et al. (2018) presented an overview of the recent developments and applications of surface renewal for ET measurements. Zhang et al. (2016) published a review of actual ET estimation based on remote sensing.

However, Section 2 reports the methodology of the review. Section 3 discusses the results and discussion, including the current status of CC and its effects on agro-metrological parameters and primary crop yield, strategies for how to deal with environmental change, and the role of C-SA technology in agriculture, the importance of accurate ET estimation in agriculture water management, research progress, and the status of ET estimation techniques, and information about partitioning the E and T from ET. The final section, Section 4, provides the conclusions and recommendations for the present study.

This section briefly discusses the results of the proposed study and underscores the importance of precise ET estimation in improving crop productivity and resilience in changing climatic conditions. The discussion highlights the review's key findings, emphasizing the role of ET estimation as a crucial tool in climate-smart agriculture.

CC is considered one of the most significant present and future challenges (Saddique et al. 2022). It is mainly caused by actions of several ongoing anthropogenic activities (such as burning fossil fuels, cutting down forests, and farming livestock) on Earth and sometimes caused by natural processes, such as changes in the sun's energy and volcanic eruptions. These activities ultimately initiate the absorption of negatively affecting gases in the atmosphere (Stern & Kaufmann 2014). Montzka et al. (2011) reported that the gases that lead to CC are mainly caused by man-made environmental problems, greenhouse gas emissions (GHGs) (such as carbon dioxide, nitrous oxide, methane, chlorofluorocarbons, and water vapor), and other man-made chemicals. The main driver of CC is the greenhouse effect. Some gases in the Earth's atmosphere act like the glass in a greenhouse, trapping the Sun's heat and stopping it from leaking back into space and causing global warming. National Academy of Sciences (2020) reported that concentrations of the key greenhouse gases have all increased since the Industrial Revolution due to human activities. Carbon dioxide, methane, and nitrous oxide concentrations are now more abundant in the Earth's atmosphere than ever in the Last 800,000 years. These GHGs have increased the greenhouse effect and caused the Earth's surface temperature to rise. Moreover, Trenberth (2011) reported that CC directly influences precipitation. Increased temperature may lead to greater evaporation and, thus, surface drying, thereby increasing the intensity and duration of drought. However, the water-holding capacity of air increases by about 7% with a 1 °C temperature increase, which leads to increased water vapor in the atmosphere. In addition, current climate models have predicted that rising temperatures will intensify the Earth's water cycle, increasing evaporation. Increased evaporation will result in more frequent and intense storms but also contribute to drying over some land areas. As a result, storm-affected regions are likely to experience increased precipitation and increased risk of flooding. In contrast, areas far from storm tracks will likely experience less precipitation and an increased drought risk.

To sum up, temperature is a key parameter used to characterize climate, and it can have various impacts on ecosystems and human existence in different locations. For instance, rising air temperatures can cause heat waves to be more extreme, disturb a variety of natural processes, and harm the agricultural sector. Accordingly, the World Metrological Organization (2024) has warned that global temperature is likely to exceed 1.5 °C above the pre-industrial level temporarily in the next 5 years. This indicates that we are getting closer to the CC goals outlined in the Paris Agreement, which speak of long-term temperature increases over decades rather than 1–5 years. According to another analysis, the global mean near-surface temperature is expected to rise by 1.1–1.9 °C between 2024 and 2028 compared to the 1,850–1,900 baseline. Similarly, the National Climate Center, China, stated that during October 2023, the collected climate data from the 140 different meteorological stations showed that the average air temperature was increased by 1.6 °C and average precipitation was decreased by 5.2% compared to the average data of previous years (Li 2023). NORA (2023) informed us that in October 2023, the global air temperature hit a record high amid climate goal concerns globally. The global average air temperature reached a record high, surpassing the previous record by approximately 0.3 °C. A study by Copernicus Climate Change Service (2022) stated that the global average air temperature in August 2022 was 0.3 °C higher than in August 1991–2020. Studies have reported that the agricultural sector will face several problems with environmental change. The agricultural industry is one of the world's most openly exposed and vulnerable fields, spreading over the vast size and sensitivity to environmental factors. Thus, the rapid variations in average temperature and rainfall patterns can negatively affect cultivated crop yields by increasing and decreasing crop growth periods and disturbing plant–water relations and physiological properties (Reyer et al. 2012; Mahato 2014; Rezaei et al. 2023; Disasa et al. 2024). US Environmental Protection Agency (2024) stated that agriculture is susceptible to weather and CC. Furthermore, it is highly dependent on water, land, and other natural resources impacted by climate. Changes in temperature, precipitation, and the onset of frosts may expand the growing season or enable the cultivation of various crops in some areas, but they can also complicate agricultural practices in other areas. Crops, livestock, rural communities, soil and water resources, and agricultural laborers are all susceptible to the effects of CC. In addition, pesticide use drives CC, and research shows that the impact of changing climate will likely lead to increased use of synthetic pesticides. Farmers will need to apply more pesticides because of the accelerated rate of pesticide degradation caused by CC, which will shorten the effective time for pesticides. Also, rising temperatures and heat stress conditions will weaken plants' natural defenses and change their biology, leaving them more vulnerable to pests. The fluctuations in climate parameters can create the best environmental conditions for developing and surviving many harmful pathogens in crops (Elad & Pertot 2014). Studies reported that in humid conditions, there are many chances that the ratio of insects and pests would increase compared to the dry season. However, in warmer and more humid conditions, the chances of spreading different plant diseases are higher and more common (Laine 2023; Singh et al. 2023). Shrestha (2019) stated that it is projected that crop losses due to insects and pest infestations may increase by about 10–25% with changing climate factors. Additionally, rapid CCs can significantly increase the growth of various weeds, and herbicide efficacy is expected to change under future climate conditions. Moreover, due to their broad gene pool and more impressive physiological plasticity, weeds are anticipated to exhibit greater resilience and better adaptation to fluctuations in CO₂ concentrations and rising temperatures when competing with crops. The dynamics of the crop–weed competition may be impacted by the differences in responses that weeds with C3 and C4 photosynthetic pathways show to increased CO₂ levels and temperatures (Varanasi et al. 2016; Malhi et al. 2020).

In addition, FAO (2015) reported that climate-related challenges are a leading factor causing major crop yield reduction. This situation will become more critical with the increasing average global temperature. Similarly, increased precipitation may reduce the need for irrigation, but excessive rainfall may also negatively impact productivity, mainly if it occurs as infrequent and concentrated heavy downpours. This yield loss of major cultivated crops would significantly maximize the prices of several food products and adversely affect agriculture's overall efficiency (Mohammadi et al. 2023). Li et al. (2009b) informed that there are many possibilities that many world regions may see heavy drought events due to rapid CC. Also, it is predicted that the yield of major cultivated crops could decrease by 50 and 90% by 2050 and 2100 years, respectively.

Based on the above discussion, it could be stated that changes in the climate parameters result in significant changes in land and water regimes that can subsequently affect agricultural productivity worldwide. With the increasing world population and CC, the pressure on the agriculture sector to supply more food products can increase significantly. CC can likely aggravate the challenges the world faces in its food security in the future. The world's average annual temperature has risen considerably over the past decades, and the warming trend can continue under future projections (see Table 1).

United Nations Climate Change (2021) reported that the C-SA tools could build resilience against CC by creating evidence, building local institutions' efficiency, and endorsing new agricultural policies. It is an efficient approach for developing and planning several agricultural strategies for securing sustainable food security under CC. The C-SA tools mainly cover three main dimensions of sustainable development: economic, social, and environmental. In simple words, it can offer sustainable solutions and conditions for escalating agricultural efficiency and incomes, adapting and building resilience to rapid environmental changes. However, introducing and working on different C-SA tools aims to fight the threads of CC. With the use of a set of technologies and practices like carbon-smart initiatives, real-time data and information-gathering applications, sophisticated climate parameter monitoring techniques, precision nutrient and water management applications, and carbon-smart activities, the C-SA umbrella approach offers a chance to revolutionize existing farming systems. Also, using these C-SA practices and technologies can improve climate resilience while mitigating adverse environmental effects (Rodríguez-Barillas et al. 2024). Zheng et al. (2024) said that farmers across different agriculture-climate zones and countries have adopted various C-SA practices to enhance the sustainability of agricultural production and build the resilience of farming communities to climate shocks. By implementing C-SA practices, farmers can achieve higher and more stable yields and improve their income status, contributing to food security and economic stability. C-SA practices adopted by farmers, including crop rotation and integrated soil management, help farmers adapt to climate risk and contribute to a reduction in greenhouse gases (GHG) emissions, increasing farm income and productivity (Tanti et al. 2024). Increased intensity of seven C-SA practices, including the use of water-saving irrigation, organic fertilizer, farmyard manure, zero tillage, fallow cropping, crop rotation, and crop straw mulch, has been shown to increase household income, net farm income, and income diversity (Sang et al. 2024). Adopting efficient irrigation systems can also help farmers cope with water scarcity and erratic rainfall patterns (Davila et al. 2024), thus stabilizing food production and livelihoods. Moreover, C-SA practices reduce reliance on chemical inputs like pesticides and fertilizers, decreasing environmental pollution and improving ecosystem health (Tey et al. 2024). Promoting and adopting C-SA practices is crucial to improving smallholder farmers' capacity to adapt to CC, mitigate its impact, and help achieve the United Nations Sustainable Development Goals (SDGs) (Ma & Rahut 2024). Mushore et al. (2021) stated that selecting suitable C-SA technology approaches may vary from location to location (even within the same country). Therefore, it is vital to understand each C-SA technology separately for particular areas. Finding the best tool needed and employed in specific areas is essential for modifying systems in other regions. Therefore, future research studies should examine area-specific responses of the C-SA tools necessary for creating national and global databases for environmental change adaptation and mitigation strategies (Andrieu et al. 2021; Ariom et al. 2022). For decades, scientists have been searching for alternative approaches and technologies to enhance the effectiveness of food production systems. Previously, many technologies have been recommended as imperative strategies to overcome rising CC, energy costs, population, and water shortage issues (see Table 2 and Figure 2).

The water balance system of Earth is shifting due to many factors, including rising temperatures, erratic annual rainfall patterns, intense downpours, and fluctuations in periods of dry weather without precipitation. These factors can cause wetland ecosystems to deteriorate or even completely collapse. Studies reported that ET is the water loss from the soil and plants to the atmosphere. Therefore, it is a vital water cycle component in managing energy balance and the planet's water cycle. Presently, it is significantly impacted by CC (Gurara et al. 2021). Consequently, the change in surface ET is one crucial indicator for measuring how the global water cycle has changed due to CC. ET is a combination of two processes, T (from plants through the photosynthesis activity/leaves of the plants) and E (loss of water from the land surface) to the atmosphere (Wang & Dickinson 2012; Yan et al. 2021a; Li et al. 2024a). The amount of solar radiation, temperature, wind, soil moisture, and air vapor pressure are some variables that influence the rate of ET. A study by Wang et al. (2022a) stated that with rapid CC, the ET values might follow an increasing and decreasing trend globally. This review study has validated the statement of Wang et al. (2022a), as can be seen from the results of previous studies in Table 3. The ET value has significantly changed in recent years due to variations in CC parameters. Another study by Allen et al. (2003) stated that an increase in air temperature could cause an increase in crop water requirement (CWR). The increasing air temperature stimulates the plant transpiration, hence decreasing the WUE. It mainly happens due to reduced leaf photosynthesis and increased stomatal conductance to water vapor. Moreover, the increasing temperature and decreasing rainfall patterns would raise the CWR of various crops. Also, it can change the cropping patterns and growing season length, negatively affecting water productivity and crop yield (Linderholm 2006; Lobell & Field 2007). Snyder et al. (2013) stated that increasing temperature could also increase the temperature of the water bodies and provide more energy to evaporate more water from water bodies than transpiration. Due to this, the total share of E and humidity in the atmosphere can likely be increased globally. The increase in moisture/humidity will tend to reduce plant T and ET processes, thus reducing the overall crop yield. In addition, the rise in CO₂ concentrations could reduce both ET processes. Accordingly, it is crucial to understand how accurate information on ET can minimize and help cope with CC's effects.

ET is a complex and nonlinear phenomenon depending on several other interacting factors such as humidity, wind speed, radiation, crop type, and growth stage. Therefore, in CC scenarios, precise information about ET can play a vital role in sustainable water management, and it can be utilized as a boundary condition in several atmospheric and soil water modeling studies. As discussed in the above sections, it can be stated that CC significantly affects the overall productivity of the agriculture sector and could trigger numerous severe and complex impacts on crop production and water availability (Sun et al. 2018). It has the potential to change the soil water balance system of the Earth, which would cause significant variations in average E and T values (Gul et al. 2022). The results of E and T variations can be extreme changes in total crop yield and could cause an increase in the frequency of severe droughts and floods worldwide (Hardelin & Lankoski 2015).

The ET combines two processes: E from soil and other surfaces and T transpiration from plants. Studies reported that both the terms E and T, which combine evaporation and transpiration, were coined because sometimes it was challenging to separate these two processes in crops (Tanny 2022). It is a fundamental variable in the water cycle and pivotal in water resource management, agricultural planning, and climate modeling. At present, various methodologies exist to measure the ET values from the field, ranging from direct measurement techniques to sophisticated modeling approaches. Generally, direct and indirect methods based on hydrological approaches, plant physiological approaches, micrometeorological approaches, and modern techniques are used to quantify the ET values from the field (Aschale et al. 2022; Hobeichi et al. 2022). A brief discussion of different ET estimation methods is also given in the following sections.

Direct ET estimation methods measure the ET values directly from the fields. These methods involve directly measuring the amount of water transferred from the land surface to the atmosphere, either through evaporation from the soil and water bodies or transpiration from plants. These methods aim to quantify ET without relying heavily on models or empirical equations, making them crucial for accurate assessments in specific environmental conditions. These methods are typically more accurate than indirect methods, but the field application of these methods is more expensive and time-consuming. The most commonly used direct ET measuring methods are lysimeters (LYM), eddy covariance (EDC), Bowen ratio-energy balance (BREB), and scintillometers (SCI) (Ezenne et al. 2023). These methods are mainly used as standard tools for evaluating field-scale ET estimations worldwide (Wang et al. 2024b; Dai et al. 2022). Additionally, the sections and Table 4 discuss each direct ET estimation method.

A LYM is a device used to measure the water balance of a soil and plant system in natural conditions. It mainly consists of a large container filled with soil embedded in the ground. The container has a drainage outlet at the bottom, which collects the water that percolates through the soil from crops. In addition, it is also used to study a wide range of hydrology, hydrogeology, ecology, environmental protection, and agronomic processes, such as ET, infiltration, salt leaching, soil water content, and crop water use (Sołtysiak & Rakoczy 2019; Tamimi et al. 2022). Previously, Johann Baptist Van Helmont was the first scientist to design the LYM in 1620 in the Netherlands (Howell et al. 1991). He designed the first LYM to conduct field experiments. After that, the LYM technique gained researchers' intention to study the soil–water and plant relation. At present, researchers propose several types of LYM structures. However, it should be noted that due to its specificity, no universal methodology is recommended for the design of the LYM setup. Generally, two types of LYM: (1) weighing and (2) non-weighing or classic LYM are used in field applications. Classic LYM calculates ET through water budgets, where excess water is removed by drainage or vacuum and is subtracted from a known water volume applied to the soil surface. While weighing LYM, measure the ET directly through the changes in the mass of the container. The weighing LYM is the most accurate method and can quantify ET over the shortest intervals. The advantages of this method are that the measurements are direct and require no interpretation or scaling, and ET can be determined over intervals more concisely than in a day (Beeson 2011). A study by Moorhead (2018) stated that LYM is one of the most accurate ET measuring techniques, which accurately weighs to estimate the quantity of water lost or gained per unit of time for the specific crop and area. The physical restrictions on plant size and substrate mass in LYM are subject to the capacity and sensitivity of the measurement devices. The LYM application procedures are costly in construction, time-consuming, and require technical expertise to perform the experiments (Liu & Xu 2018; Hess et al. 2021). Allen et al. (2011) stated that the representativeness of the ET measured by LYM is apt to be suspected for different crop trait parameters, water and nutrient conditions, and soil properties in and outside the system. Differences in soil type, vegetation, and microclimate outside the LYM can lead to discrepancies and may not accurately represent the variability of ET over a larger field. Moreover, how an LYM method can upscale the ET estimation at the plot or field level is still unclear. Some studies suggested that the small pot-based or micro-scale LYM could replicate the ET estimations from the same cropping area. These types of LYM are economical and have lower settlement costs than larger ones. The micro-scale LYM systems are easy to operate and do not require technical knowledge to run the experiment. Several researchers have used this ET estimation method to determine soil E and plant T with varying growth and dry biomass production and allocations in real-time conditions (Ruiz-Peñalver et al. 2015).

In addition, pan evaporation is one type of direct ET estimation method. It is commonly used to measure evaporation from open-water bodies or land surfaces in most parts of the world for determining rainfall-runoff modeling, crop water requirements, and irrigation scheduling (Irmak et al. 2002; Majhi et al. 2020). Generally, a pan is made of galvanized steel or other non-corrosive material and placed on a leveled land surface filled with water (usually 2–8 inches deep) (Wu et al. 2020; Dehghanipour et al. 2021). The water level in the pan is served and measured daily, typically at a fixed time, using a graduated measuring gauge. The difference in water level between the initial and subsequent readings represents the amount of evaporation. After that, the measured pan evaporation is multiplied by a pan coefficient, a factor that accounts for the difference between evaporation from the pan and actual ET from a vegetated surface or water body. The pan coefficient values vary depending on local climate conditions and vegetation type (Al-Mukhtar 2021; Emadi et al. 2021; Elbeltagi et al. 2023). The pan evaporation method is easy to implement and requires minimal equipment, making it suitable for field applications and areas with limited resources. The equipment and maintenance costs are relatively low compared to other ET measurement techniques (Majhi & Naidu 2021; Abed et al. 2023; El Bilali et al. 2023). This method has several limitations that must be considered when interpreting and applying pan evaporation data. Sometimes, it provided overestimated results because it is openly exposed to direct sunlight and wind, which increase evaporation rates compared to vegetated surfaces, canopies, and soil moistures. Its measurements require regular maintenance, including refilling the pan to maintain a consistent water level and cleaning to prevent algae growth. Despite these limitations, the method is considered valuable for estimating ET globally, particularly in data-scarce regions.

EDC, also known as eddy correlation or eddy flux, is a sophisticated micrometeorological technique used to calculate the vertical turbulent fluxes of gases and measure the turbulent flux of momentum, heat, water vapor, trace gases, and aerosol concentrations among the land surface and the atmosphere (Vekuri et al. 2023). It is based on the principle that the turbulent flux of a quantity is proportional to the covariance of the vertical wind velocity and the different atmospheric gas concentrations. The EDC method uses a three-dimensional sonic anemometer and a gas analyzer. The sonic anemometer estimates the components of the wind and velocity, and the gas analyzer measures the concentration of the quantity of interest (e.g., carbon dioxide, methane, water vapor) (Hill et al. 2017; Baldocchi 2020; Liu & Qiao 2023). Swinbank (1951) proposed the EDC method first. He said that this method can be used experimentally to study spatially varying factors in a variety of research and operational settings, including plant functional and structural properties, irrigation management, nutrient management, ET, infiltration, leaching losses, and air quality (Rafi et al. 2019; Buttar et al. 2019; Wang et al. 2024c; Jiang et al. 2024). A study by Vekuri et al. (2023) reported that the attractiveness of this technique is manifested in the FLUXNET network worldwide, which has over 900 EDC sites globally. FLUXNET network presents continuous data on the short-term net ecosystem exchange of CO₂ with the atmosphere and several other parameters. The measured data are integrated temporally to determine the related carbon balance. In addition, EDC is widely used in measuring ET over several ecosystems, offering real-time data on the amount of water vapor being transferred from the surface to the atmosphere. It is also extensively used to monitor CO₂ exchange, aiding in studies of carbon sequestration and GHGs. Additionally, the technique can measure sensible and latent heat fluxes, contributing to energy balance assessments in different environments. This method offers several advantages, including direct, real-time measurements of gas fluxes without the need for empirical models or assumptions. It can capture fluxes over a relatively large area, representing the average conditions of the landscape, and is versatile enough to be applied in various ecosystems, such as forests, grasslands, wetlands, agricultural fields, and urban areas. Chatterjee & Anapalli (2023) stated that the EDC method requires high-frequency sensor measurements with simultaneous data processing. However, evaluating the EDC method is challenging due to the large fetch area with a highly variable open boundary layer. The EDC method also has some disadvantages. It requires sophisticated, costly equipment and significant setup, operation, and data analysis expertise. The accuracy of the measurements can be affected by several factors, including terrain complexity, atmospheric stability, and instrument calibration. Additionally, the measured energy balance often needs to be perfectly close, leading to potential uncertainties in the data. However, the technique works best in homogeneous, flat terrain with sufficient wind speed, making it less effective in complex landscapes.

BREB is another direct ET method used to estimate latent and sensible heat fluxes based on air temperature measurements, humidity gradients, net radiation, and soil heat flux (Spittlehouse & Black 1980; Fritschen & Simpson 1989). It is based on the Bowen ratio technique, which is the ratio of the sensible heat flux to the latent heat flux. The Bowen ratio is calculated from simultaneous vertical air temperature and humidity gradients measured at two heights above the surface (Bowen 1926; Euser et al. 2014). The energy balance equation is then used to solve the latent and sensible heat fluxes using the Bowen ratio and the net radiation flux. The BREB method is relatively simple and inexpensive for measuring surface energy fluxes. It is a comparatively robust method and is less sensitive to measurement errors than other methods (Buttar et al. 2018, 2022). Without requiring any knowledge of the aerodynamic properties of the surface of interest, the BREB approach relies on direct, elementary measurements. It can integrate latent heat fluxes over vast distances (hundreds of meters) and measure the variations in climate parameters over short intervals (such as 10 and 30 min) while allowing for continuous, unsupervised assessment. Its drawbacks include the instrument's biases in measuring gradients and energy balance terms (Liu & Xu 2018). In addition, there's a chance that the data will be discontinuous, and a sufficient upwind source area is needed to create an equilibrium boundary layer with horizontally constant temperature and vapor gradients (Todd et al. 2000). To detect even minor variations in temperature and vapor pressure at two heights, this method's major disadvantage is that the two sensors must precisely agree with one another (Fuchs & Tanner 1970; Angus & Watts 1984; Fritschen & Qian 1990).

A SCI is a scientific device used to measure turbulent fluctuations of the refractive index of air caused by variations in temperature, humidity, and pressure. It consists of an optical or radio wave transmitter and a receiver at opposite ends of an atmospheric propagation path. Both the instruments are separated by several hundred meters to kilometers. Its transmitter and receiver measure the area-averaged sensible and latent heat fluxes from the field. The Monin–Obukhov similarity theory serves as the foundation for this method (Beyrich et al. 2021; Ward 2017). Moreover, its transmitter emits electromagnetic radiation scattered by the turbulent atmosphere, transferring it toward the receiver. The transmitted signal's intensity changes are picked up and assessed by the receiver. The structural parameter (Cn2), which is the spectrum amplitude of the refractive index fluctuations in the inertial sub-range of turbulence, is typically used to quantify the magnitude of refractive index variations (Hill & Ochs 1978; Wang et al. 1978; Moene et al. 2004). The SCI approach has the following benefits: it does not require direct contact with the surface being measured, making it ideal for measuring ET over large areas. It can estimate ET over several kilometers, making it suitable for regional-scale studies. It can provide continuous measurements of ET, which helps monitor changes in ET over time. However, the disadvantage of the SCI method is that its efficiency can be affected by atmospheric conditions, such as cloud cover and wind speed. It has a limited spatial resolution, so it cannot be used to measure ET at fine scales. The tripod vibrations caused by high wind speed can increase SCI-based heat fluxes to reach unrealistic values, consequently, ET. Besides, high wind speed can alter SCI data through (1) tripod vibrations, (2) misalignments, and (3) electronic noise and absorption contributions to the signal. Therefore, reprocessing SCI data through spectral analysis is essential for computing SCI-based heat fluxes, and studies should focus on this issue (Aguirre et al. 2022).

The sap flow method is another technique of ET estimation; in this technique, the movement of fluid in the stems, trunks, tillers, roots, stems, and branches of plants are measured in the xylem tissues of plants using several sensors (Bouamama-Gzara et al. 2022; Dukat et al. 2023). Water movement in plants is correlated with sap flow, which is also correlated with transpiration and, on occasion, ET. Therefore, it is not called water flow or velocity because the fluid within a plant's stem is not entirely water. It is helpful for adequately managing irrigation water (Jones 2004; Vandegehuchte & Steppe 2012; Esteves et al. 2015; Kumar et al. 2022). Fernández et al. (2001) and Dix & Aubrey (2021) stated that sap flow measurements can be categorized into heat pulse and constant heating: heat pulse velocity, heat balance, thermal dissipation, and heat field deformation. According to Sun et al. (2021), research has been done to determine how much water plants use by connecting sap flow to the scalar variables of plants. By establishing a correlation between sap flow and leaf surface area, one could accurately estimate the transpiration of the plant canopy. Nonetheless, several investigations have endeavored to associate sap flow with meteorological parameters, and the results have demonstrated a favorable correlation between sap flow and air temperature (Juice et al. 2016). The sap flow method offers several advantages for studying plant–water relations and physiology, such as (1) direct measurements of sap flow, which is a more accurate representation of plant–water use; (2) providing continuous measurements, which allows for detailed analysis of plant–water use patterns over time; and (3) providing an opportunity to conduct nondestructive measurement, which is allowing us to make repeated measurements on the same plant over time. The disadvantages of the sap flow method are: (1) it consists of several sensors (heat pulse velocity sensors) or probes that are inserted into the plant stem, which can be invasive and may affect plant growth; (2) its measurements can be affected by environmental factors (temperature, humidity, and wind speed), which can affect its efficiency; and (3) it requires careful calibration to ensure accurate measurements (Köstner et al. 1998; Uddin et al. 2014; Bittencourt et al. 2023).

The indirect ET estimation method is a set of models commonly used to estimate the amount of water lost to the atmosphere based on the collected meteorological data, calculation of the empirical relationships, and modeling approaches. Currently, several empirical-based ET estimation models are proposed and used to estimate the ET. These ET estimation models have been categorized into the following types: (1) radiation-based, (2) temperature-based, (3) relative humidity-based, (4) water balance/mass transfer, and (5) a combination of all models (Muhammad et al. 2019; Dai et al. 2022). A study by Djaman et al. (2015) and Lang et al. (2017) informed that each ET estimation model has its perspectives and concepts and is often developed for broad applicability across different climatic regions. Some models are designed or calibrated for particular climatic conditions or vegetation types. Thus, they can only provide more accurate ET estimates within their intended application range but may not perform as well in other regions or for different vegetation types (see Table 5). As can be seen in Table 5, many research studies have been conducted globally to select the most suitable and accurate ET estimation model for a particular region (Song et al. 2019; Hamed et al. 2022). Studies reported that choosing the best model ET estimation for a specific area depends on several factors, such as data availability, the type and volume of the water body, and the cropping area. However, until today, it is still challenging to choose and recommend the most appropriate or optimal model for a particular region. Because of the lesser availability of observed field data, most models have not been evaluated against direct ET measurement methods across various areas and different climate zones (Kiefer et al. 2019; Hamed et al. 2022).

In addition, from 1800 to 1947, non-Dalton theory-based E estimation models were also developed, mainly based on aridity analysis and energy as radiation, degree days, or atmospheric turbulence (McMahon et al. 2016). Researchers that applied atmospheric turbulence to estimate the E processes from 1920 to 1939 are provided in a study by Thornthwaite & Holzman (1939). Besides, the Bowen ratio and BREB equations were introduced in 1926 and 1927, respectively (Bowen 1926; Cummings & Richardson 1927).

After the development of Equation (5), the science of E was established through a range of theoretical analyses that culminated in models for estimating E in different environmental conditions, water sources, agricultural lands, and soil types. Based on Equation (5), various world researchers have developed several E estimation models by modifying Equation (5) according to their requirements.

In addition, recent decades have witnessed an almost exponential increase in indirect ET methods. Since 2000, ‘ET’ has become more frequent in the scientific literature than ‘E’ when referring to the integrated land surface latent heat flux. Moreover, some of the most frequently used indirect ET estimation models are shown in Table 1 in the Supplementary File.

In addition, RS is one of the most advanced technologies. It is mainly used to acquire information about an object without physically touching it. It is typically done by detecting and measuring the reflected or emitted electromagnetic radiation from an object or area. It can be effectively used to evaluate, monitor, and study the variations of the water cycle and its main components, such as ET, T, and P, for regions at different temporal scales (e.g., hours, days, months, and years) (Cui et al. 2018). With the advent of RS in the near past, data collection sources have fundamentally improved by introducing satellite sensors with higher spatial and temporal resolution on space-borne platforms. Also, most RS datasets are freely available online, and anyone can get them. It has been further advanced by developing open-source RS services, spatially distributed hydrological models, and data processing, analysis, and visualization software. Several RS-based sensors and models have been developed to estimate the ET. The developed sensors include the Moderate Resolution Imaging Spectroradiometer, Landsat, Sentinel-2, Landsat 8, the WorldView series, the Visible Infrared Imaging Radiometer Suite, the Advanced Very High-Resolution Radiometer, and the Advanced Spaceborne Thermal Emission and Reflection Radiometer. However, presently available RSG-based algorithms comprise TSEB, atmosphere-land exchange inverse, triangle evaporative fraction model, trapezoid interpolation model, the surface energy balance algorithm for land model (SEBAL), the surface energy budget system, the simplified surface energy balance index, Mapping ET at high resolution with internalized calibration (METRIC), four-source surface energy balance model, time domain triangle model, the surface energy balance algorithm for land-improved, S-SEBI, two-source triangle evaporative fraction model, two-source trapezoid model for ET hybrid dual-source scheme and trapezoid framework-based ET model (HTEM), surface energy balance system model, endmember-based soil and vegetation energy partitioning, and others (Li et al. 2009a; Amani & Shafizadeh-Moghadam 2023). Li et al. (2009a) study reported that RS is an alternative approach to computing ET on local, regional, and global scales. It is the only viable means to map ET's regional and mesoscale patterns on the Earth's surface globally consistent and economically feasible (Weigand & Bartholic 1970; Moran et al. 1989; Caselles et al. 1992). Another study by Engman (1999) revealed that the RS has several marked advantages over conventional ‘point’ measurements: (1) it can offer broad and nonstop spatial coverage within short time intervals; (2) it is not costly; and (3) it is more suitable for un-gauged areas where man-made measurements are challenging to conduct. A recent study published by Amani & Shafizadeh-Moghadam (2023) informed that the RS cannot provide the ET measurement directly, but it can help to estimate the retrievable surface parameters data associated with ET. Hatfield (1983) said that through the RS, anyone could measure the surface temperature, vegetation indices, and soil moisture data from a resolution of a few cm² to about several km² by downloading satellite images from certain satellites. RS can offer spatial distribution and temporal evolution of many other parameters such as the digital elevation model, normalized difference vegetation index (NDVI), leaf area index (LAI), enhanced vegetation index, photosynthetic active radiation, soil adjusted vegetation index, surface albedo, normalized difference water index, and land use and land cover from visible and near-infrared bands (Mauser & Schädlich 1998; Amani & Shafizadeh-Moghadam 2023).

Studies reported that these RS models can accurately estimate the ET variations for extensive areas through the real-time collection of land surface information during satellite transit. Therefore, more and more studies are applying remote sensing techniques to simulate regional ET based on spatial modeling approaches (Karimi & Bastiaanssen 2015; Chen et al. 2023a). However, the limitations of RS methods can be categorized into three main areas: accuracy, spatial resolution, and temporal resolution. Meanwhile, its accuracy mainly depends on the quality of input data, such as satellite imagery and local meteorological data. Also, RS models require calibration and validation using ground-based measurements. The spatial resolution represents the accuracy of the selected satellite for downloading the imagery, and it can limit the ability of the software to estimate ET at fine scales. The mixed pixels of the imagery are another factor in heterogeneous land cover areas where a single pixel can represent a mix of different land cover types. Lastly, the temporal resolution of satellite imagery may not be sufficient to capture the dynamic nature of ET. This issue primarily arises during changing weather conditions. The cloud obstructs the satellite observations and limits the ability to monitor ET continuously. Despite the above limitations, ET estimation using RS is constantly increasing. Researchers worldwide are actively working on ET estimation using RS (see Table 6) and addressing these limitations by developing new algorithms, incorporating ancillary data sources, and improving calibration and validation strategies.

The emergence of modern technologies such as UAVs in agriculture holds immense potential in intelligent farming in a changing climate. Modern technologies enable us to collect field data and monitor ongoing agricultural activities in real-time (Rouchi et al. 2023). A study by Tang et al. (2019) stated that currently, UAVs are gradually employed in agriculture for ET estimation. It is a vital technique for monitoring ET from the agriculture fields. It provides many opportunities for researchers to cover and monitor larger cropping areas by taking advantage of new sensing techniques, mapping approaches, and data analytical methods. Another study by Niu & Chen (2022) reported that the RS satellite imageries can only provide temporally and spatially distributed measurements. Their spatial resolution is in the range of meters, often not enough for crops with clumped canopy structures. However, the spatial resolution of UAVs-captured imageries can be as high as centimeter-level, allowing for detailed mapping of ET variability within fields or small areas. Rozenstein et al. (2023) and Niu et al. (2020) stated that the limitations of the RS technology could be overcome by using the UAVs technology. For example, RS satellite imageries are prone to cloud cover, but UAVs images are not prone to cloud cover. UAVs are flying below the clouds and nearly closer to the ground or crops, reducing atmospheric interference and improving the accuracy of surface measurements (Ebert et al. 2022). UAVs imageries can be captured at any time if the weather is within operating limitations, but RS imageries have a fixed flight path. For the most significant accuracy, several other lightweight Red, Green, Blue (RGB), multispectral, and thermal infrared sensor cameras can be mounted on the UAVs to collect high-resolution filed images, enabling a more comprehensive assessment of surface energy fluxes. In addition, the UAVs imageries have higher temporal and spatial resolution images, relatively low operational costs, and nearly real-time image acquisition, making the UAVs an ideal platform for mapping and monitoring ET (Zhang et al. 2019; Shao et al. 2021). For ET estimation using UAVs, models developed for satellite remote sensing are used rather than being developed explicitly for UAVs imagery. These models include METRIC, high-resolution mapping of ET, ML, one source energy balance, the surface energy balance algorithm for land (SEBAL), dual-temperature-difference, artificial neural networks (ANNs), and two sources energy balance (TSEB) (Laliberte et al. 2007; Hardin & Hardin 2010; Mokhtari et al. 2021; Khanal & Barber 2023; Rozenstein et al. 2023; Tunca 2023). However, UAVs offer unique opportunities due to their high spatial resolution and flexibility, which has led to adaptations of these models for UAV-specific applications. Therefore, some researchers use high-resolution vegetation index-based models for UAVs imagery to estimate ET. These models utilize the detailed vegetation indices that UAVs can capture to estimate ET at a much finer scale than satellite imagery allows. The UAVs' ability to capture data at a high spatial resolution enables more accurate and localized ET estimates, which are instrumental in heterogeneous agricultural landscapes. Niu et al. (2020) reported that each UAVs model has merits and demerits. METRIC/SEBAL models are more recognized by researchers for ET estimation using UAVs, but they are based on satellite (Landsat) platforms. Therefore, significant modifications would be required to make them work best with UAVs images. However, the TSEB model is less widely known, but it offers more potential for UAVs applications in many crop conditions, especially tree crops. In addition, no existing models can fully satisfy the spatial, temporal, spectral, and accuracy requirements for ET-based science and applications. Therefore, innovative models for ET estimation using UAVs are required. Besides, previous studies reported that the optimal point or height where UAVs drones should fly to capture the field images and the data that best represents a crop's ET is still unclear. It is a big challenge associated with ET estimation using UAVs and needs proper attention. However, many researchers have adopted UAVs for ET estimation by flying the UAVs at different heights (usually setup as 30, 60, and 120 m), using specialized equipment, and controlled environments, and relying on data analysis expertise.

Recently, ML technology has emerged as a promising tool for ET estimation, offering a noninvasive and spatially explicit approach to assess water cycle management. ML represents input–output relations without understanding the physical process, making them practical tools for modeling nonlinear systems (Yildirim et al. 2023). Also, it provides powerful tools to model complex relationships between meteorological variables, vegetation characteristics, and soil properties based on a minimum number of weather input parameters. ML algorithms can learn from large datasets of historical observations and remotely sensed data to predict ET with improved accuracy and spatial resolution (Raza et al. 2023a). By analyzing images captured by sensor cameras or UAVs, different ML algorithms can extract valuable information about vegetation characteristics, canopy temperature, and soil moisture conditions, which are crucial inputs for ET models. Generally, the ML approaches are categorized into TA and deep learning (DL)-based algorithms. The TA is further classified into random forests (RF), multi-layer perceptron, SVM classification, and regression trees approaches. The DL algorithms are further classified into ANN, genetic programming, long short-term memory (LSTM), linear regression (LR), artificial hummingbird algorithm, SVM, multi-layer perceptual neural network, convolutional neural networks, self-organizing mapping neural network, and others. Studies reported that the above algorithms could be used to estimate the ET. However, compared to other algorithms, the ANN tool is appropriate for performing nonlinear modeling processes, such as crop ET estimation (Raza et al. 2023b; Dimitriadou & Nikolakopoulos 2022; Makwana et al. 2022; Tejada et al. 2022; Mostafa et al. 2023). Pagano et al. (2023) reported that it is worth noting that the inputs to these models lend themselves well to sensitivity analyses and obtaining optimal structures. They do not rely on mathematical relationships even for complex phenomena. Although these models have been trained and tested in the literature for many meteorological stations and climatic conditions, preparing them for every station in a large region is impractical. Therefore, regional-based ML ET estimation models need to be created. Currently, ML has emerged as a promising tool for ET estimation, but it also faces several challenges and limitations. For example, ML models require high-quality data for training and validation. However, ET data are often limited and may be subject to errors and uncertainties, particularly in regions with sparse meteorological and vegetation monitoring networks. Some models, such as the RF, are fast to implement, while other models, such as ANNs, take longer time; by increasing the study area and influential factors, the ML models might be time-consuming mainly when parameter tuning is considered and input data come from the high-resolution satellite imagery. Therefore, it is suggested that future studies should examine the limitations and accuracy of ET estimation using DL models and RS data for local and large-scale studies. Also, studies should be planned by performing the comparison between different MAL algorithms with TA methods under other climate conditions for particular crops at general and regional levels (Amani & Shafizadeh-Moghadam 2023; Hendy et al. 2023; Zouzou & Citakoglu 2023).

Partitioning ET into E and T is essential for understanding the global water cycle and improving water resource management under CC (Paul-Limoges et al. 2022; Petrík et al. 2022). In addition, a growing awareness of the importance of eco-hydrology has motivated efforts to partition ET into its components as a key to unraveling processes underlying ecosystem water use and its response to CC. Rothfuss et al. (2021) summarized the published articles on partitioning ET into E and T. They reported that the T-to-ET ratio is a critical parameter in landscape hydrology and ecology. Hence, separating the T flux in ET is essential for many applications because of its link to the plant–water relationship. At the global scale, the uncertainty of the ET into E and T estimates remains high. Therefore, many research studies have been focused on the differentiation and quantification of ET components. One of the most essential eco-hydrological challenges and ET processes coincide, and there is no easy way to distinguish between them (Chen et al. 2023b; Nguyen & Choi 2023). A study by Gibson et al. (2021) and Kool et al. (2014) mentioned that the history of partitioning ET into E and T dates back to the 1970s when scientists developed models and methods (such as micro- LYM and sap flow measurements) and conducted experiments. Several ET partitioning physical models are designed to estimate the E and T. The developed models include the Shuttleworth–Wallace two-source ET model, the Priestly–Taylor jet propulsion laboratory model, and the diagnostic biophysical model (e.g., PML-V2) (Yan et al. 2019; Lu et al. 2023). Studies reported that Shuttleworth & Wallace (1985) were the first scientists to publish the analytical model combining E and T by formulating the different media through which evaporative flux travels as resistances. Besides, the Shuttleworth-Wallace proposed approach which caught the interest of many other world scientists. Later various numerical and analytical models (energy and water balance, soil water energy and transpiration, TSEB, FAO dual-Kc, isotope model, and satellite-based estimations) were developed to determine the E and T separately at regional and global scales (Lascano et al. 1987; Norman et al. 1995; Daamen & Simmonds 1996; Allen et al. 1998; Yepez et al. 2003; Gao et al. 2022; Raghav et al. 2022; Yan et al. 2022b). A review study published by Kool et al. (2014) on partitioning the ET into E and T mentioned that micro-LYM, soil heat pulse, chambers, micro Bowen ratio-energy balance, and EDC methods are used to determine the E from the soil surface and water bodies. They reported that chambers, biomass–transpiration relationship, and sap flow methods are thermal-based techniques to compute the water flow through the stem of the plants. These techniques can broadly be classified into (1) heat balance, (2) heat pulse, and (3) constant heater methods. Another study by Liebhard et al. (2022) said that several experimental and modeling approaches exist based on the isotope composition of soil water and vapor to partition E and T. Various practical methods comprise techniques to sample and analyze water vapor above or at the soil surface. However, theoretical approaches to partitioning E and T are often based on the free water evaporation Craig-Gordon, analytical, or numerical isotope transport models.

In addition, micrometeorological methods and process-based models for catchment scale are also used for ET partitioning (Gaj et al. 2016; Rothfuss et al. 2021). These models consider a watershed's entire water cycle, including isotope compositions of runoff, vadose, ground, stream, and xylem waters. Consequently, the isotope-based eco-hydrological models can be used for partitioning E and T and determining the different processes of the water cycle (Ala-Aho et al. 2017; Kuppel et al. 2018; Knighton et al. 2020). Ren et al. (2022) planned the study by selecting the sap flow and LYM methods for partitioning the ET into E and T using stable isotope techniques. The ET value estimated by the sap flow method was less than LYM, and the simulated value with LYM was higher than that of sap flow. This study concluded that LYM is the most accurate method compared to other methods. Li et al. (2024b) observed the changes in water vapor and carbon dioxide fluxes and the flux variance similarity theory based on five WUE algorithms for partitioning ET into its components from urban forest land. They combined oxygen and hydrogen isotopes to verify their results and discussed the algorithms' characteristics, uncertainties, and applicability in urban forest land. This study indicated that the partitioning of ET was partially consistent with the isotope-based approach. Yu et al. (2022) proposed a novel ET partitioning method using a unified stomata conductance model to estimate the ratio of T and ET by calculating the balance of the WUE of the ecosystem to leaf using half-hourly flux data. The model was consistent with other methods that were compared and showed high correlation values to the sap flow-based T approach. The developed method can provide a feasible and universal system for partitioning ET globally. Lai et al. (2022) proposed a cutting method for separating the ET. They concluded that the cutting method is applicable and could obtain a reasonable ratio of T to ET for winter wheat cropland. Quintela da Rocha et al. (2022) quantified the T and E using EDC flux measurements in tallgrass prairie using the underlying uWUE approach. They compared the results daily. They concluded that ET partitioning studies at the ecosystem scale are still scarce. Han et al. (2022) performed in situ measurements of water isotopes in atmospheric vapor, plant tissues, and soil pools for partitioning ET by estimating the stable isotopic compositions of ET. They combined high-frequency laser spectroscopy and chamber methods to constrain the estimates of T/ET. Their results provide insight into the regional water cycle and potentially benefit many applications, from forest ecosystem protection to paleo isotope archives. Sun et al. (2022a) used an isotope tracer coupled to a ‘multi-source’ model to estimate the ET and its components. The study reported that the R² was 0.78 and 0.74 between the simulated ET and total ET measured by the EDC and sap flow methods. These results illustrated that this model could reconstruct the water exchange in the desert riparian forest ecosystem. Liu et al. (2022b) proposed a method using RSG and ground observations for 30 sites encompassing 10 primary plant functional types and optimized the Penman–Monteith model to calculate T and T/ET. This study showed a high correlation (R² = 0.69) between the results of the Ball–Berry–Leuning model and the optimized Penman–Monteith model. The authors reported that the proposed method can be valuable for T and T/ET using RSG.

This review summarizes the results of previously published research studies on CC and CS-A technologies and discusses the water cycle's main components, especially ET. After a brief literature review, we found that CC has many adverse effects on regional temperature, rainfall, and ET variations. These changes have significantly impacted agriculture production (see Table 1), giving rise to considerable uncertainties and threats of food and water insecurity. Under increasing climate conditions, adopting CS-A technology tools (see Table 2) can help mitigate CC's adverse effects.

In addition, this study stated that ET, which maintains water availability on Earth, is highly sensitive to CC (see Table 3). Besides, several methodologies, tools, and approaches (direct, indirect, and modern methods) are available for efficient estimation of the ET under diverse conditions. Despite the availability of numerous methods, the estimation of ET is quite difficult. It becomes more complex due to the unavailability of all metrological parameter data at all stations or regions. Also, each method carries inherent complexities, limitations, and drawbacks that hinder accurate assessments. However, the following conclusions and recommendations were drawn for further research:

• (1) In recent years, significant progress has been made in ET estimation; however, until today, it is still challenging to choose the most appropriate method. Because of the following issues: (1) the data availability, (2) most models have not been evaluated against direct ET measurement methods across various regions and different climate zones, and (3) ET value uncertainties compared with direct, indirect, and modern methods. Therefore, finding a suitable method based on the availability of data and the performance of the ET estimate method is challenging.

• (2) Scaling ET approaches from fields to regional levels requires understanding and dealing with multiple difficulties related to gathering data, modeling methodologies, and the assessment and validation process. These limitations underscore the necessity for interdisciplinary collaboration and rigorous techniques to enhance the dependability of regional ET estimations. Thus, it is necessary to implement integrated approaches that combine remote sensing, advanced modeling techniques, and comprehensive ground validation to tackle the difficulties associated with scaling ET estimations.

• (3) The limitations of the RGS method are accuracy, spatial resolution, and temporal resolution. The cloud cover and atmospheric interference can further degrade the data quality, leading to inaccuracies in ET estimation. However, the optimal point or drone flying height where the data can best represent a crop ET remains unclear for UAVs and battery life, payload capacity, and flight duration.

• (4) Applying the ET estimation results to the smart irrigation systems can help to save and improve the irrigation timing and amount significantly. The approach utilizes ET data, soil moisture sensors, and weather forecast systems to improve and enhance water efficiency, minimize over-irrigation, and stimulate healthier crop development by connecting irrigation methods with plants' water needs in real-time.

We believe the information provided will help readers by comprehensively analyzing the accuracy and applicability of different ET estimation methods under various climate conditions. It highlights the vulnerabilities and consequences of inaccurate ET calculations, offering insights into selecting appropriate strategies for managing water resources efficiently. This can also enable flood and drought risk assessments, promoting enhanced planning and adaptive methods for sustainable water resource management within the watershed amid changing climatic scenarios. The article also addresses the challenges and limitations of ET estimation and provides recommendations for future research directions. It is a comprehensive resource for researchers, policymakers, and practitioners interested in leveraging ET estimation for sustainable agricultural practices in the face of climate variability and change.

This study was financially supported by the National Key R&D Program (2021YFC3201103); the Natural Science Foundation of China (U2243228,41830863, 1509107); the Key R&D Project of Jiangsu Province (BE2022351); the Yinshanbeilu Grassland Eco-hydrology National Observation and Research Station, Institute of Water Resources, Hydropower Research, Beijing 100038, China (YSS2022011) and the Jiangsu Funding Program for Excellent Postdoctoral Talent (No.2022ZB640).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.