Climate change and extreme weather events are ongoing threats to our food systems (Wheeler and Von Braun 2013; Krishnamurthy R et al. 2022; Ortiz-Bobea et al. 2021) and exacerbate chronic problems in food access and environmental justice (Wilson 2010; Gilson and Kenehan 2018). These challenges are more critical in the context of healthy food options such as tree nuts, which provide vital proteins and fats linked to a reduced risk of heart disease (Liu et al. 2020). The health benefits have led to higher demand illustrated by a threefold increase in consumption over the past 50 years (CDFA 2022). To meet this demand, California almond acreage increased 145% in the last two decades, making almonds the most commonly grown crop in the state. Almonds are grown on nearly 1.4 million acres by approximately 8,000 farms, producing about $7 billion in value, closely following the grape and dairy industries. However, California is subject to extended periods of drought and climate extremes. In fact, an ongoing water crisis due to recent prolonged and severe droughts (Wahl et al. 2022) has highlighted the need to use water more efficiently. In this context, we introduce the Tree-Crop Remote Sensing of Evapotranspiration Experiment (T-REX) as an interdisciplinary project aimed at developing crop sensing technologies across scales to inform and support 1) agricultural management decisions with an emphasis on irrigation and sustainable cultural practices, 2) local- to regional-scale water management, and 3) state policies related to water, greenhouse gases, and agro-environmental sustainability.
In response to drought conditions and the overdrafting of aquifers for irrigation, the state of California enacted the Sustainable Groundwater Management Act (SGMA) to curb groundwater overexploitation in 2014 (Roberts et al. 2021). Water supplies are also affected by the loss of water storage as snow in the Sierra Nevada (Siirila-Woodburn et al. 2021; Rhoades et al. 2022). Additionally, the increased risks of megafloods and precipitation whiplash events (Swain et al. 2018; Huang and Swain 2022) are also challenging California’s water management. Therefore, current and future water availability scenarios suggest increased uncertainty in creating and managing irrigated cropping systems in California.
New crop management practices for California woody perennial crops increasingly aim to balance crop yield and quality while considering water scarcity. Most woody perennial crops in California use micro-irrigation systems, such as sprinkler, drip, or, increasingly, subsurface drip irrigation (Jha et al. 2022). Moreover, most fertilization programs are linked to irrigation systems; therefore, irrigation management must not only optimize crop water demands but also prevent the accumulation of minerals in the first few centimeters near the soil surface while avoiding leaching minerals into groundwater. In addition, a comprehensive understanding of soils’ fragility has led to the concept of soil health—the continued capacity of soil to function as a vital living ecosystem that sustains plants, animals, and humans (Lehmann et al. 2020). In this context, the use of cultural practices (e.g., cover crops, the integration of livestock, and reducing or eliminating tillage) under the umbrella of regenerative agriculture are being incorporated to improve soil health, sequester carbon, and increase biodiversity in agroecosystems (Newton et al. 2020). As a result, new tools for irrigation management need to be developed to respond to the complexities of modern agricultural management beyond a sole yield maximization goal.
Woody perennial cropping systems are particularly vulnerable to droughts given their inflexible water demands during the summertime and the long return time on initial investments. For solutions to these issues, E&J Gallo Winery, the largest family-owned winery in the United States, partnered with the U.S. Department of Agriculture’s Hydrology and Remote Sensing Laboratory (HRSL) with the vision of developing practical tools based on satellite and/or airborne systems to guide irrigation decision making. The goal was to develop accurate maps of actual evapotranspiration (ETa) at daily to weekly increments and subfield spatial resolutions to reduce water use and enhance crop quality by improving irrigation efficiency (Kustas et al. 2018). This effort is known as GRAPEX (Grape Remote Sensing Atmospheric Profile and Evapotranspiration Experiment), an ongoing project and partnership with an interdisciplinary research team joined by multiple USDA Agricultural Research Service (ARS) laboratories; universities; the National Aeronautic and Space Administration (NASA); international research partners from Chile, Spain, Italy, and Israel; and an expanding network of industry partners. The success of GRAPEX has translated into a wealth of scientific knowledge (e.g., more than 65 peer-reviewed publications), more than 100 extension talks and seminars, and training of a new generation of researchers (six postdoctoral researchers and six graduate students) tackling the most pressing challenges in ETa monitoring and modeling, crop physiological modeling, and irrigation management.
Consequently, farmers, commodity groups, and state and federal agencies are joining efforts to support the development of the next generation of tools to improve water resource management in California. For instance, T-REX shares the ideas and motivations of the GRAPEX project, yet acknowledges that almonds are commonly grown in areas facing greater water insecurity due to increasing demands and variable supply. In addition, new farming technologies are rapidly becoming mainstream, agro-environmental issues are evolving, and water issues are worsening in the state. Almond growers have started to implement cultural practices such as using cover crops, integrating livestock, and reducing or eliminating tillage to promote soil health and carbon sequestration. Therefore, the vision of T-REX is more ambitious than what has been accomplished with the GRAPEX project. The project aims to develop the science and technology necessary to create the next generation of remote sensing–based farming decision support tools to sustainably grow perennial woody crops. T-REX also incorporates deliberate partnerships with local cooperative extension advisors, a network of almond growers, the Almond Board of California [ABC; 501(c)(3) nonprofit], and expertise in knowledge coproduction regarding contentious water challenges (Nocco et al. 2020). Our collaborative team also works across scales to develop cutting-edge scientific knowledge to accelerate the transition toward sustainable water and soil management. The T-REX project team aims to combine innovative satellite, uncrewed aerial systems (UAS), and other proximal sensing technologies to retrieve key parameters to model surface fluxes and other biophysical variables (Fig. 1). Tools developed under T-REX will advance current “business as usual” approaches by providing near-real-time and forecasted ETa information with greater spatial and temporal detail, guiding water and soil management decisions in orchard systems with more accuracy. Thus, our approach aligns with the strategy outlined by Novick et al. (2022) to inform Nature-based Climate Solutions (NbCS) by combining water and carbon fluxes across spatial and temporal scales.
Illustration of the components, interactions, and outcomes of the T-REX project.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
T-REX project scientific scope
ET is a critical component of the hydrological cycle and the terrestrial energy budget. While direct ET measurements are challenging, several methods based on residual water balance, lysimetry, or micrometeorology have advanced to use point measurements representing the release of water to the atmosphere from water bodies, vegetation, and other land surfaces (Pozníková et al. 2018). The eddy covariance technique, one of the most common micrometeorological methods, estimates fluxes between a flat and homogenous surface and the overlying atmosphere by solving the covariance between turbulent vertical wind fluctuations and the quantity of interest, allowing the measurement of heat, mass, and momentum exchanges (Aubinet et al. 2012). Other estimation methods are based on models with different levels of complexity. Simple ET modeling approaches include using reference ET (ETo) estimates and then adjusting to crop ET (ETc) using crop or ecosystem coefficients (Allen et al. 1998). ETo can be estimated through various semiempirical relationships (Li et al. 2016). These approaches require meteorological data, empirical parameters, and a crop coefficient (Kc) to derive ETc. Also, there are more sophisticated methods for modeling ET based on turbulent exchange, which can vary from basic bulk transfer paradigms to more advanced techniques such as higher-order closure or Lagrangian descriptions (Paw U et al. 2003; Wang et al. 2023; Pozníková et al. 2018).
Satellite remote sensing technology offers the ability to provide routine spatial information on plant conditions and water use. Satellite retrievals of land surface temperature (LST; derived from thermal infrared imagery) have been shown to be particularly useful when estimating ETa due to the sensitivity of soil and canopy temperatures to variable soil moisture availability (Moran 2004). Several LST-based ETa models have been developed over the past few decades, including the Mapping Evapotranspiration with Internalized Calibration (METRIC; Allen et al. 2007), the Surface Energy Balance Algorithm for Land (SEBAL; Bastiaanssen et al. 1998), the operational Simplified Surface Energy Balance model (SSEBop; Senay et al. 2007), and the Atmosphere–Land Exchange Inverse model (ALEXI; Anderson et al. 1997) paired with the associated flux disaggregation technique (DisALEXI; Norman et al. 2003). In the context of T-REX, we focus on the paired ALEXI–DisALEXI modeling framework, which is based on the Two-Source Energy Balance (TSEB) approach.
The ALEXI–DisALEXI model has yet to be evaluated in detail in almond orchards, which provide a uniquely structured canopy and specific management practices that present intriguing modeling challenges, including 1) assessing satellite-based ET model uncertainty using the OpenET platform; 2) evaluating the ET model’s ability to partition bulk flux from remote sensing products (resolution of 30 m) between tree canopy and interrow; 3) addressing the structural characteristics of the canopy that will influence heat exchange and turbulent flow below, at, and above the canopy; 4) understanding the complex radiation transport through the canopy that will likely lead to variations in shadowing, soil surface fluxes, and ultimately thermal signatures; and 5) evaluating the impact of almond tree physiology, phenology, and agricultural management on modeling ETa.
Identifying key factors within the ET modeling framework that can be modified to optimize performance over these systems requires a deep understanding of the physical processes and management activities occurring within the orchard. Consequently, the T-REX project combines in situ data, ground-based information acquired during intensive observational periods (IOPs), and high-resolution information on surface conditions collected from UAS (Fig. 2). Continuous in situ micrometeorological measurements focus on 1) advancing the understanding of inherent uncertainties related to the eddy covariance technique (e.g., surface energy imbalance, local- and large-scale advection, fluxes fetch and footprint, sensor deployment and accuracy); 2) developing new low-cost robust and reliable techniques for continuous ETa monitoring; and 3) linking leaf-to-canopy plant physiological processes controlling water and CO2 gas exchange across water and heat stress levels observed in commercial orchards (Fig. 2). We also collect UAS-based imagery during and outside of the IOPs with off-the-shelf and custom UAS. Outside of the IOPs, we fly off-the-shelf UAS to develop, evaluate, and refine high-resolution ETa models and remotely sensed estimates of plant water stress with instrumentation that would be immediately accessible to applied researchers, extension advisors, crop consultants, and larger growers.
Synergies (a) across spatiotemporal scales, (b) between uncrewed aerial systems (UAS) and meteorological and biophysical measurements, and (c) intensive observational periods (IOPs) for the T-REX project.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
In this paper, we provide an overview and preliminary analysis of modeling efforts and measurements collected during the T-REX project to date. Results focus on the evaluation of modeled ET estimates based on thermal satellite remote sensing retrievals against eddy covariance measurements. We discuss the opportunities, challenges, and uncertainties due to inherited limitations in both observational and modeling approaches. We also describe the synergies between data collected via in situ, UAS, and satellites as a means to better inform irrigation management decision-making in almond orchards.
Sites and data description
Study sites.
The data used to evaluate physical processes unique to almond orchards and to refine and evaluate the models were collected in three almond orchards located in the Central Valley of California (Fig. 3). Orchards include VAC (Solano County near Vacaville, CA), WWF (Yolo County near Woodland, CA), and OLA (Madera County near Madera, CA). VAC is a 70-ha 7th leaf (i.e., number of years since trees have had leaves in the orchard after planting) orchard (year 2022 tree height of 6 m and a fractional cover of 60%) planted with 100% of the Independence variety in a southwest–northeast row orientation with drip irrigation installed. WWF is a 60-ha 9th leaf orchard (a tree height 6 m and a fractional cover of 80%) with 50% Nonpareil and 17% Butte, Monterey, and Carmel varieties, with a north–south row orientation and microsprinkler irrigation installed. OLA is a 20-ha 8th leaf orchard (a tree height of 6.5 m and a fractional cover of 90%) with 50% Nonpareil, 37% Wood Colony, and 13% Supareil varieties, with north–south row orientation and microsprinkler irrigation installed. Hereafter, we refer to each site using a three-capital-letter abbreviation as detailed in the previous description (i.e., VAC, WWF, and OLA).
(left) Location of T-REX almond orchards within the state of California. (center) Black lines represent field boundaries and red dots represent flux tower locations. (right) Pictures from the flux-tower view-angle at the three T-REX sites: (from top to bottom) WWF, VAC, and OLA, respectively.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
OLA and VAC orchards have a naturally occurring cover crop that is present in the alleys (not the berms) during the nongrowing season (November–March). Once conditions become dry in the early summer, the cover crop is either left to senesce or removed via mowing, as a common practice used to maximize the irrigation water available for the trees. WWF has a seeded mix of legumes and grasses as cover crop present from winter through spring and then mowed after senescence. As is common in almond orchards, each site experiences a sharp decrease in applied irrigation during the phenological stage of hull split initiation, which typically occurs in early August. When the almond fruit splits and exposes the soft shell inside (hull split), it leaves the almond vulnerable to diseases. To limit damage, growers are recommended to implement regulated deficit irrigation to promote the drying out of the nuts and to shorten the window of time where the orchard is vulnerable to damage (Girona et al. 2005). Over the course of 3 weeks following hull split initiation, growers slowly increase irrigation back to estimated ETc until preharvest, when the orchards are dried down. Following harvest, orchards receive full estimated ETc requirement until senescence.
Meteorological and biophysical monitoring.
In late spring and early summer 2021, we installed towers and initiated measurements of surface energy fluxes, meteorological variables, and phenological changes at the three T-REX commercial almond orchard sites (as described above). These tower sites included micrometeorological instrumentation to estimate the net exchange of water (LE), carbon (Fc), and surface energy components [i.e., net radiation (Rn), sensible heat flux (H), and soil heat flux (G)] based on the eddy covariance (EC) technique, as well as other methods such as surface renewal, flux gradient, and Bowen ratios (Foken 2017). An array of soil heat flux, soil temperature, and moisture sensors in a diagonal cross-row transect of five measurement points was deployed to estimate the soil heat flux corrected by heat storage. A detailed list of all the sensors and deployment height or depth in the soil is presented in Table 1.
Micrometeorological sensors and deployment height at each experimental site. See footnotes for expansions of corporation names. Please note that the use of trade, firm, or corporation names in this article is for the information and convenience of the reader. Such use does not constitute official endorsement or approval by the U.S. Department of Agriculture or the Agricultural Research Service of any product or service to the exclusion of others that may be suitable.
The EC technique was used to measure exchanges of energy, momentum, and atmospheric trace gases (e.g., CO2 and water vapor) between the T-REX almond orchards and the atmosphere. The EC method relies on measuring the fluctuations in wind velocity, temperature, and trace gases at high frequencies to estimate the surface fluxes (i.e., H and LE), while the available energy is directly estimated as the difference between Rn and G. The H and LE values were calculated as functions of 60-min average covariance of the corresponding variables sampled at 20 Hz. Anomalous observations in the high-frequency time series for each variable were removed following the median absolute deviation method introduced by Mauder et al. (2013). Wind velocity components were rotated into the mean streamwise flow following the 3D coordinate rotation method described by Paw U et al. (2000). The air temperature was derived from sonic temperatures and the atmospheric water vapor density estimations were based on Schotanus et al. (1983). The resulting fluxes were adjusted by the Webb, Pearman, and Leuning (WPL) density corrections (Webb et al. 1980). Wind components and scalar quantities were adjusted in time to account for sensor displacement, and also, spectral corrections based on Moore (1986) were applied to address the attenuation of fluxes, especially at high frequency. Since the ideal cospectra for latent heat flux is not well defined, we used an idealized cospectra for sensible heat as a reference (Moore 1986; Wang et al. 2023). This assumes a cospectral similarity between heat and moisture fluxes (Massman 2000). Therefore, a transfer function was applied to derive spectral correction factors for both heat and moisture fluxes. Spectral factors were derived separately for different times of the year (i.e., December–February, March–May, June–August, and September–November) and stability cases (i.e., unstable, neutral, and stable). Flux footprint analyses based on Kljun et al. (2015) were performed, and only observations when at least 90% of fluxes’ source and sink area are expected within the orchard area were considered on daily analysis.
Data processing for gap-filling and energy balance closure correction to estimate ET for satellite-derived ET evaluation used an open-source Python toolkit following Volk et al. (2021). Initial hourly fluxes of H, LE, G, and Rn were interpolated for gaps up to 2 h during the daytime or 4 h at night. Any remaining gaps led to the omission of the total daily estimate. Following gap-filling, fluxes were summed to daily periods. Energy balance closure corrections were applied by distributing the residual between (Rn − G) and (H + LE) while conserving the Bowen ratio (H/LE) (Bambach et al. 2022a; Mauder et al. 2020; Volk et al. 2021). All statistical measures for satellite-based ET comparisons were against this corrected ET data (closed energy balance). Only dates maintaining a daily energy balance closure of 0.75 or above were included.
During shorter periods within each growing season, additional micrometeorological instrumentation was deployed at each site. More specifically, at the Vacaville site, a profile of sonic anemometers and infrared gas analyzers were mounted at 11.0-, 7.4-, 6.0-, 2.2-, 1.4-, and 1.0-m height on the tower to investigate effects of the canopy structure on near-surface turbulence. Wind, temperature, water vapor, and temperature high-frequency measurements were also carried out within the tree canopy and just above the interrow middle surface (1.5 m above ground) to characterize turbulence characteristics in relationship to the canopy structure.
Sap-flow sensors were installed in the Vacaville site using custom heat pulse probes per the design of Deng et al. (2021). These double ratio method (DRM) sensors consist of upstream and downstream temperature probes and a heater needle used to deliver heat pulses once every 15 min. The DRM technique measures velocities across a wide range, including very high, very low and negative velocities (reverse flow). One sensor was placed on each of 12 trees located within the eddy covariance mean flux footprint and relatively close to the flux tower.
Tree phenology and interrow changes (e.g., cover crop presence and stages) were continuously monitored using phenocams positioned on each tower and programmed to collect RGB and thermal images at 0800, 1200, 1600, and 2000 Pacific standard time (PST).
Intensive observational periods (IOPs).
On 39 days during the 2021 and 2022 growing seasons, we performed IOPs at each of the T-REX sites (Table 2), and plans are to continue these intensive data collection campaigns in subsequent seasons. IOPs are planned on Landsat 8 or 9 satellite overpass dates for each site, often with UAS flights, and throughout different orchard/tree phenological stages. The first IOP in each growing season would typically occur in spring after flowering and leaf out, when tree canopy leaf area is low and middle/cover crop biomass is higher. Since the towers were installed in late spring 2021, the IOPs started later in the first growing season.
Number of intensive observational periods per site and year.
Several ground-based spot measurements within the area represented by the flux footprint were collected during the IOPs. To determine variability in tree biomass, water use, and stress, measurements of leaf area index (LAI; LAI-2200, LI-COR, Lincoln, Nebraska), leaf stomatal conductance, photosynthesis, chlorophyll fluorescence (using portable gas exchange systems LI6800 or LI-600F, LI-COR), and leaf water potential (using pressure chambers PMS Instrument Company, Albany, Oregon) were collected in six, five, and seven trees at WWF, VAC, and OLA, respectively. Leaf-level hyperspectral measurements (TREK and FieldSpec 4 Spectroradiometer, ASD Inc., Boulder, Colorado) were collected to evaluate stress conditions and relate remotely sensed multispectral retrievals to in situ canopy conditions. The reflectance values are being used to evaluate and calibrate stress detection tools scalable to airborne and satellite spectral observations.
Satellite remote sensing.
Model schematic of the ALEXI–DisALEXI ET modeling framework adapted for tree crops from Anderson et al. (2004).
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
The ALEXI model is run operationally over the continental United States (CONUS) at 4 km spatial resolution by using two morning LST observations from the GOES satellites (approximately 1.5 h after local sunrise and 1.0 h before local noon). The TSEB model is applied at both times to simulate fluxes and atmospheric boundary layer (ABL) growth and warming over the morning period (Anderson et al. 1997, 2007b). Daily ET (mm day−1) is estimated by upscaling the instantaneous latent flux retrieved at the time of the second LST measurement to a daily value (MJ m−2 day−1) by conserving the ratio of latent heat to insolation during daylight hours (Cammalleri et al. 2014a,b), and then converting to mass flux using the latent heat of vaporization. ALEXI inputs include meteorological forcing information (wind speed, air temperature, solar radiation, air pressure, and vapor pressure) from the Climate Forecast System Reanalysis (CFSR) dataset at 0.25° resolution and hourly to 3-h time steps (Dee et al. 2014). Land cover classification was from the University of Maryland global land cover dataset at 1 km resolution based on observations from AVHHR (Hansen et al. 2000). Landcover information is used to assign values to hc (canopy height), d (displacement height), z0 (roughness length for momentum transfer), leaf absorptivity, and nominal leaf size (Anderson et al. 2007a).
The DisALEXI algorithm (Norman et al. 2003; Anderson et al. 2004) was developed to map ET at higher spatial resolution by spatially disaggregating ALEXI ET through the utilization of higher-resolution thermal infrared (TIR) imagery. DisALEXI operates by running TSEB over each ALEXI pixel using higher-spatial-resolution vegetation cover and LST information from polar-orbiting satellites. To ensure consistency between the ALEXI and DisALEXI flux fields, air temperature is set at a nominal blending height of 50 m and is iteratively adjusted at the ALEXI pixel scale until the DisALEXI daily ET fluxes converge to the ALEXI value over that pixel. At present, DisALEXI high-resolution (30 m) ET maps are created on Landsat overpass dates using Landsat thermal and reflectance data. DisALEXI input includes the same meteorological forcing information used by ALEXI and a land cover classification from the 30-m National Land Cover Dataset (NLCD; Fry et al. 2011). DisALEXI also uses Landsat 7, 8, and 9 Collection 2 TIR and Surface Reflectance (SR) band information as well as Harmonized Landsat and Sentinel-2 SR datasets (HLS). Note that Landsat 7 and 8 were used during 2021 and Landsat 8 and 9 were used in 2022 (following the launch of Landsat 9 in December 2021).
Data were acquired during both years on cloud-free days for Landsat scenes p044r033 and p043r034, covering sites WWF/VAC and OLA, respectively. High-resolution LST maps were generated using Landsat Collection 2 Surface Temperature (ST) sharpened from a native resolution of 100 to 30 m using a data mining sharpening (DMS) approach developed by Gao et al. (2012). LAI and NDVI at 30-m resolution were created using HLS band information following a procedure in Gao et al. (2020) based on machine learning, which has been recently modified to include ground-based LAI measurements from vineyard sites (Kang et al. 2022). Instantaneous latent heat flux retrieved at Landsat overpass time is upscaled to daily ET (mm day−1) using the same approach as ALEXI. Interpolation of ET between Landsat overpasses follows a similar conservation approach. The ratio of daily ET (mm day−1) and reference ET (ETo; mm day−1) for each Landsat overpass date is first calculated and then linearly interpolated on a daily basis between clear overpass dates. These values were then multiplied by ETo to compute a daily time series of ET, accounting for changes in vegetation cover. Hence, the reintroduction of daily reference ET aims capturing the effects of daily weather changes on ET rates. Reference ET for both Landsat overpass dates and interim dates are obtained from a local meteorological station (CIMIS network; https://cimis.water.ca.gov/). For more information on the ALEXI–DisALEXI modeling framework in the context of California specialty crop application, the reader is referred to Knipper et al. (2019).
Airborne remote sensing.
Most almond orchards in California have precision irrigation systems (microsprinklers, drip) to increase water application efficiencies (Jha et al. 2022). The majority of these systems do not have the flexibility to irrigate at a tree-level precision. However, new technologies such as variable drip irrigation systems offer granularity that could lead to potential water savings, increase yields, and improve fruit quality. Furthermore, UAS technologies have been growing their presence in agriculture and irrigation management, aiming to provide information to support irrigation decisions at the tree level. Within the T-REX project, UAS missions were designed 1) to test and further develop airborne remote sensing thermal and multispectral data collection, which are referred to as “basic missions,” and 2) to develop a workflow for timely production of ET maps for irrigation management and develop product types of interests for growers, which are referred as “applied missions.”
During basic missions, we fly custom UAS platforms to capture aerial data using advances in UAS technology and the highest possible precision and accuracy in thermal and multispectral imagery to evaluate and improve procedures to model high-resolution ET, ET partitioning between soil and plant, and water stress. Ultimately, basic missions aim to expand our fundamental knowledge of the spatial energy balance and to better understand the limitations of the tools developed from applied missions. For both basic and applied missions, we are acting as a model-agnostic group with regard to adapting spatially explicit energy balance models for UAS-based imagery. We are working with UAS-based versions of the TSEB model (Nieto et al. 2019; Nassar et al. 2021), High-Resolution Mapping of Evapotranspiration model (HRMET; Ebert et al. 2022), and the new Three-Source Energy Balance model (3SEB; Burchard-Levine et al. 2022a,b) formulated for landscapes with heterogeneous canopy elements. The dual-canopy nature of tree cropping systems, high value of each tree, and longevity of production (25 years) make high-resolution ETa partitioning and deployable tree-based estimates of ET and stress a critical remote sensing goal for UAS. When using the TSEB approach, ET is derived using a similar version of the model as described for satellite remote sensing, yet special considerations are made to account for high-resolution LST data (Nieto et al. 2019). Due to the nature of the UAS information, the main crop and the interrow conditions can be characterized in terms of reflectance, temperature, canopy height, and fractional cover at tree row distance (Fig. 5). This characterization allows for the quantification of TSEB energy balance components [Eq. (1)], including the contributions of the trees and interrow to ET. The TSEB model frequently incorporates advances based on new formulations that better characterize the soil–canopy–atmosphere interactions observed through field experiments; model updates are regularly available in the open-source TSEB repository (https://github.com/hectornieto/pyTSEB). Previous research on vineyards using TSEB provides a context for the data collection efforts and challenges to address when monitoring tall, clumped canopy environments (fruit and nut trees) (Nassar et al. 2020; Gao et al. 2023).
Model schematic of the TSEB ET modeling framework describing tall, clumped canopy vegetation and interrow (as almond orchards) for unmanned aerial vehicles. Illustration adapted from Nassar et al. (2021).
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
During applied UAS missions in 2021 and 2022, we flew a quadcopter (DJI Matrice 210RTK, Shenzhen DJI Sciences and Technologies Ltd., Shenzen, Guangdong, China) equipped with a combined multispectral and thermal camera (Micasense Inc., Seattle, Washington) every 2 weeks at the OLA and WWF sites close to Landsat overpass dates. Applied aerial missions are flown at an elevation of 90 m with 70% overlap. At a 90-m flight altitude, the applied system captures multispectral imagery (475, 560, 668, 717, 842 nm) at a 3.88-cm resolution and thermal imagery (forward leaning infrared centered at 11,000 nm) at a resolution of 60.9 cm. All data from applied missions are processed into geotiffs using the Micasense Altum photogrammetry workflow developed for use with Agisoft Metashape Professional (Agisoft LLC, St. Petersburg, Russia).
T-REX will explore the links across scales, and we will investigate the added value of spatial resolution within the context of agricultural management. For instance, Fig. 6b illustrates the potential of high-resolution ET maps depicting information in enough detail to inform lateral/control valve zones and diagnose irrigation management challenges in almond orchards. For example, one of the varieties in this orchard, Supareil, appears to have relatively higher ET fluxes (shown as “stripes” of high values every six rows). Variety-based diagnostics and management may be desirable for leaf-out, hull-split, and postharvest irrigation needs. These phenological stages can occur at slightly different times across almond varieties and could benefit from differences in irrigation management (Drechsler et al. 2022). At current spatial resolution, each pixel corresponds to an orchard row, and zone diagnostics are possible to identify canopy gaps or trees that are underperforming.
Spatial maps of (a) modeled ET (mm day−1) derived using the ALEXI–DisALEXI scheme and (b) TSEB applied to thermal imagery collected via a UAS-based mission over the OLA modeling domain during the Landsat overpass (1230 PST) on 5 Aug 2021.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
Future work will refine models at the 7-m scale that encompasses both interrows and trees, while developing even higher-resolution models that separate individual trees from bare soil and cover crops (Burchard-Levine et al. 2022a) to take advantage of the highest thermal resolution from off-the-shelf tools (currently 0.3–0.6 m at 90-m flight altitude). Nevertheless, a visual comparison against modeling outputs based on the ALEXI–DisALEXI scheme reveals similar spatial patterns, highlighting the need to further advance the development of fusion approaches that could exploit the consistency, frequency, and large-scale spatial coverage obtained from satellite-based ET products and the granularity from UAS-based imagery.
Preliminary results and analysis
In this article, we introduce the general framework and scope of the T-REX project; thus, this section introduces some preliminary results and analysis with a focus on the evaluation of model ET using the ALEXI–DisALEXI scheme and EC flux measurements.
EC flux measurements.
A fundamental assumption when assessing the quality of eddy covariance flux measurement is that turbulent fluxes (H + LE) should equal the available energy for turbulent transport (Rn − G) (Foken and Napo 2008). However, there is observational uncertainty associated with the measurements or assumptions made in calculating the terms of the energy balance equation, which can lead to a lack of closure or energy imbalance (Leuning et al. 2012; Mauder et al. 2020). Usually, EC accuracy is assessed based on the slope of the regression line between available energy and EC fluxes: (H + LE)/(Rn − G). Thus, the slope represents the overall energy balance closure, and as slope values approach 1, it is assumed that there is less uncertainty in estimating the surface fluxes. The T-REX sites EC flux observations for the year 2022 have a mean ratio of H + LE to Rn − G of ≈0.88 and ≈0.90 based on the subdaily (60 min) and daily estimations, respectively (Fig. 7). The magnitude of this imbalance is comparable to the lack of closure reported by other studies measuring EC fluxes on woody perennial orchards and vineyards (e.g., Bambach et al. 2022a,b; Volk et al. 2023; Prueger et al. 2019).
Scatterplots and linear least squares regressions of (left) hourly and (right) daily available energy (Rn − G) and EC turbulent fluxes (H + LE). Rows represent the observations for a given T-REX site. In each plot, R2 is the coefficient of determination, which is followed by the linear regression equation for each relationship.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
From a practical standpoint, our results highlight the challenges and limitations regarding using crop coefficients to estimate crop water demands. Several studies use the EC technique to estimate ETa, and Kc is derived as the ratio of ETa to reference ET (ETo) (e.g., Drechsler et al. 2022; Bellvert et al. 2018). We observe that the magnitude of ETa and related lack of closure can vary across almond orchards leading to a 17% uncertainty in maximum seasonal Kc values across a comparison for the three T-REX sites and the two growing seasons analyzed (2021 and 2022). Unfortunately, several studies estimating Kc for almond orchards based on the EC method do not report if any correction was performed to account for the lack of energy balance closure. Therefore, it was challenging to compare these results with previous studies reporting almond orchards’ water demands.
The almond orchard at Vacaville, California (VAC), has the largest energy imbalance. Given the location of this study site, we hypothesize that surface fluxes are significantly influenced by advective fluxes, which are not accounted for through the EC technique. VAC is surrounded by extensive pasture areas interrupted by small residential neighborhoods and some orchards. Thus, the “oasis” advective effect by which the cropland evaporative cooling leads to horizontal transport of wet and cold air toward surrounding dry and hot areas, while intense vertical convection at the almond orchard, in this case, might lead to a mesoscale low pressure system. Then, the warm and dry air from the surrounding areas moves above the orchard, creating pseudostable atmospheric conditions, which reinforce the surface cooling, enhance ET from the orchard, and might drive a net flux of water vapor from the orchard toward the surrounding hot and dry areas (Ruehr et al. 2020; Potchter et al. 2008; Douglas et al. 2009).
On 18 July 2022 [day of the year (DOY) 199] at VAC, the air temperature and vapor pressure deficit observed at the orchard reached maximum values of 34.3°C and 3.5 kPa, respectively. Prevailing winds were consistently from the south and southeast direction. Observations over a senescent pasture upwind from the almond orchard reached air temperature and vapor pressure deficit maximum values of 37.2°C and 5.5 kPa, respectively. Figures 8a–d illustrate vertical profiles of turbulence kinetic energy (TKE) at the VAC almond orchard for the onset of advective conditions preceding 18 July 2022. The air temperature and moisture differences observed between the almond orchard and the surrounding area on 18 July 2022 are coupled with a strong inversion in the TKE parameter above the almond orchard. These observations illustrate the oasis effect by showing indications that a hot, dry, and stable air mass above the canopy enhances convection within the canopy, enhances ET, and very likely leads to a net horizontal flux not accounted for by the EC method. The H and LE time series from observations at different heights for this case show some indications of the expected biases in surface fluxes under strong advective conditions (Fig. 8e).
(a)–(d) Turbulence kinetic energy (TKE) profile observations from measurements at the T-REX Vacaville, California, from days of the year (DOY) 196 to 199. (e) Sensible (H) and latent (LE) heat flux estimations based on EC analysis of observations at six different heights (z) from the ground surface. (f),(g) Air temperature profiles representing (f) nonadvective and (g) advective conditions. (h) Illustration of sensors, and corresponding height (z), deployed at the T-REX Vacaville flux tower.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
First, there is a more significant contribution of afternoon and night surface fluxes decoupled from the available energy components. Careful consideration is required when assessing methods to achieve EB closure during advective conditions. Second, instrumentation commonly deployed to measure fluxes above the roughness sublayer might be subject to greater influence by the dry and hot air above the canopy, which might not necessarily represent the strong convective conditions driving surface fluxes. For example, on DOY 199 during the afternoon, LE estimations at 1.5 m above the canopy are greater than LE estimations at near twice the canopy height (z = 11.0 m). Third, LE estimations from observations in the interrow space converge to LE derived from measurements within the canopy, adding other evidence of the enhanced within-canopy turbulence resulting from these advective conditions, which could be of particular importance in sparse, row-oriented canopy structures. Figures 8f and 8g depict air temperature profiles for DOY 196 and 199, which represent nonadvective and advective conditions, respectively. The warm layer above the canopy on DOY 199 is observed, yet the temperature difference with the air temperature below is only about 1.5°C. These measurements were performed using aspirated fine-wire thermocouples, and we suspect that misleading results could be found if less sensitive air temperature observations were considered. While the energy imbalance observed across the different T-REX sites is probably explained by several compounding factors, we will continue and expand extensive micrometeorological observations to understand the role of advection as a source of uncertainty in ET estimations based on the EC method. A distributed temperature sensor (DTS) along with an optical and microwave scintillometer system will be deployed to investigate the influence of advection across a horizontal gradient, and a Streamline XR Doppler scanning lidar to monitor vertical and horizontal atmospheric characteristics. We argue that advection might be of significant importance when aiming to quantify surface fluxes over the agricultural landscape in semiarid regions such as California, particularly if changes in water supply contribute to fallowing of agricultural land near orchards. Unfortunately, limited progress has been made on quantifying the influence of advection on ET and other surface fluxes based on micrometeorological or land surface modeling approaches. Thus, the impact of advection on ET has been identified as a research priority for the T-REX project.
Satellite remote sensing.
Evaluation of the accuracy of DisALEXI daily ET largely depends on the accuracy in EC flux tower measurements achieved on clear Landsat overpass dates. To evaluate the performance of DisALEXI over almond orchards, we derive statistical measures against observed closed flux ET estimates. However, visual analysis, including time series and scatterplot evaluation, includes both energy balanced closed and unclosed observed ETa estimates to provide context in the range of observed ETa at these sites. Figure 9 shows a time series of DisALEXI-derived ETa estimates on Landsat overpass dates (red diamond) and interpolated daily ETa values (red line). Observed closed flux ETa estimates following Volk et al. (2023) are shown as filled blue markers (closed; Fig. 9), while unclosed ETa is shown as white markers with blue edges (unclosed; Fig. 9. As with previous DisALEXI evaluation efforts, modeled fluxes have been averaged over a 3 × 3 pixel area (90 m × 90 m) shifted to include the EC tower footprint.
Time series comparison of DisALEXI ETa on Landsat overpass dates (LS ETd; red diamonds) and fused daily DisALEXI ETa (LS ETd Fused; red line) against observed closed ETa (blue markers) and unclosed ETa (white markers) for sites (top) WWF, (middle) VAC, and (bottom) OLA.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
Time series evaluation indicates DisALEXI produces reliable ETa estimates both on Landsat overpass dates and when interpolated to the daily time steps for WWF and OLA, with values tracking observations closely during most of the months during both years (RMSE = 1.20 mm day−1 for both Landsat and daily ETa on average for the sites; Table 3). At WWF, DisALEXI ETa aligns well with observations during the entirety of 2021, but fails to capture peak ETa values during the month of June for 2022. Similarly, a negative bias is reported for DisALEXI ET during the peak months (JJA) of both years for site VAC. This negative bias is exacerbated in 2022 when observed ET reaches values greater than 8 mm day−1 for a larger portion of the growing season, while modeled values remain between 4 and 6 mm day−1. Differences between observations and DisALEXI modeled ETa are likely due to advective conditions present at VAC partially captured by eddy covariance ETa estimates. The VAC almond orchard neighbors senescent grassland and barren/nonvegetative land (airport) causing advection of hot/dry air across the orchard, increasing the evaporative demand, and causing latent heat fluxes to increase substantially, yet not represented by the DisALEXI modeling scheme. Similar results were found at a vineyard site subject to strong advective conditions (Knipper et al. 2020). Specific evaluation and further research into the effect advection is having on EC measurements at VAC is currently underway.
Statistical measures n (number of days included), meanP (mean predicted), meanO (mean observed), R2 (coefficient of determination), RMSE (root-mean-square error), MAE (mean absolute error), and MBE (mean bias error) for sites WWF, VAC, and OLA for Landsat overpass (LS), daily fused (Fused), weekly (Week), and monthly (Month) time steps. Note that weekly and monthly statistics have been converted to units of mm day−1 for comparison.
Site OLA, located the farthest south, shows the largest daily ET estimates, with values consistently above 6 mm day−1 for JJA. DisALEXI modeled ETa values tend to align best at OLA as day to day variability is less drastic. Despite high correlation (R2 = 0.88 for daily estimates; Table 3), DisALEXI ETa shows a consistent positive bias for the entire time series (MBE = 0.55 mm day−1 for daily estimates). This positive bias is increased for both years during the month of August and into September, aligning with hull split. Focusing on observed ET during this timeframe we find a sharp decline in ETa, with values going from roughly 8 to about 2 mm day−1 (in 2021) or 3 mm day−1 (in 2022) in 2 weeks’ time (Fig. 10). DisALEXI modeled daily ETa indicates a decline in ETa during the same timeframe. However, ETa values do not decrease at the rate of observed ETa, producing a larger positive bias during this time and largely contributing to the overall MBE.
Scatterplots of observed ETa (x axis) and modeled ETa (y axis) for sites (top) WWF, (middle) VAC, and (bottom) OLA and time steps Landsat overpass (white), fused daily (blue), weekly totals (green), and monthly totals (red).
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
Weekly total ETa estimates are most relevant to irrigation management as schedules are commonly created and delivered on a weekly basis. As such, we include a statistical analysis of weekly total estimates compared to observations (Fig. 10 and Table 3). Also included are daily (Landsat ETa values) and monthly values to determine modeling capabilities at varying time scales (Fig. 10). Figure 10 indicates that overall patterns found in daily estimates are also found in weekly estimates and, to a lesser extent, monthly. Statistical measures indicate a general improvement in MAE and R2 values when transitioning from Landsat ETa to monthly for sites WWF and OLA (Table 3). Although a statistical improvement from Landsat ET to monthly ET is present at VAC, daily and weekly estimates report higher errors (RMSE = 2.32 and 2.39 mm day−1 for daily and weekly, respectively, compared to 2.26 and 2.17 mm day−1 for Landsat and monthly, respectively). Although generalities exist when scaling up temporally, such as positive biases at OLA and negative biases at high ETa values at VAC, variation about the mean decreases. This result is expected due to the time averaging of random errors and suggests modeled results can be best used for applications where weekly or greater time scales are required.
A key advantage of satellite remote sensing—and by association the DisALEXI modeling framework—is the ability to monitor ETa over expansive areas that not only show within orchard variability but encompass surrounding agricultural commodities. Figure 11 provides a contextual look at the mapping of ETa using the DisALEXI modeling framework by providing a regional view of daily ETa derived on Landsat overpass date 22 July 2022 (Fig. 11, left and right panels). Also included are enhanced views of WWF, OLA, and VAC orchards on the same date (Fig. 11, top-center to bottom-center panels, respectively). Irrigated agriculture is made apparent in Fig. 11, with drastic variations in ETa values across each domain, including near each almond orchard. Figure 11 puts into context the challenges associated with obtaining both observed measurements and modeled fluxes at site VAC by showcasing the expansive fallow and senescent grassland landscape to the southwest of the orchard, coinciding with the predominant winds.
(left),(right) Spatial maps of DisALEXI ETa over each modeling domain and (center) focused view over each almond orchard on Landsat overpass date 11 Jul 2022.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
Finally, a key aspect of the T-REX project is identifying and quantifying uncertainty sources in ET measurements and model estimates. Moreover, identifying the timing, landscape, weather, and other environmental conditions that can bias observations or model estimates is critical to transitioning toward the broad adoption of a new generation of ET tools for water and other natural resources management. We recognize that validating numerical models of natural systems is impossible, as discussed in detail by Oreskes et al. (1994). For instance, all the assumptions and simplifications compound on the measurements from the eddy covariance technique, and the DisALEXI modeling scheme will always provide a limited yet useful depiction of ET from a complex agroecosystem. In Fig. 12, we illustrate summer ET maps (total ETa from June through August) for the T-REX sites in 2021 and 2022. On each map, black contours define the orchards under study and the eddy covariance flux footprint of daytime fluxes for the same summer period within each orchard. The box-and-whisker plots compare the variability of model ET estimates within the orchard and the flux tower footprint, while the ET value from EC measurements is shown as a horizontal black line across each subplot. Noteworthy is the spatial variability within VAC, with larger ETa estimates modeled for the northern portion of the orchard compared to the south, with trends in magnitude running northwest to southeast. The area of the orchard with the largest ETa values coincides with the tower location, yet ET estimates from EC are 10% greater than the ET estimates from the area with the largest fluxes. Site OLA shows the least amount of within-orchard variation, producing large uniform ETa estimates. WWF shows slightly more variation, with ETa estimates decreasing on the eastern edge of the orchard, and larger differences within the compared years. These results highlight the importance of disentangling the sources of ET uncertainty, and model and observational sensitivity to those factors. Moreover, future research will incorporate other sources of information, such as high-resolution thermal imagery and leaf-level physiological measurements, to assess the potential and limitations of such information to constrain ET modeling and observational uncertainty.
Summertime (June–August) spatial maps of DisALEXI total ETa (mm) over each T-REX site for 2021 and 2022. Black contours represent the orchard area under evaluation and the EC flux footprint within each orchard. Footprint analyses for daytime fluxes were performed based on Kljun et al. (2015); contours represent the 90% fluxes’ source/sink footprint isoline from a Gaussian kernel distribution estimation. Box-and-whisker plots summarize the ET variability when considering the EC flux footprint and the entire orchard areas. Total ET from EC for the same period is represented by a horizontal black line in each box-and-whisker plot.
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
Soil management and carbon fluxes.
As part of the T-REX study, we have monitored carbon fluxes at the ecosystem scale over almond orchards that cover the span of soil management from bare soil, naturally occurring winter cover crops (NOWCC), winter seeded and spring mowed cover crop (SMCC), and winter seeded and graze cover crop (SGCC). Our results suggest that cover crops have a positive impact on carbon sequestration, especially during the fall and spring seasons, but winter and summer seasons’ respiration seems to partially offset such benefits (Fig. 13). The sites with the largest cumulative net ecosystem exchange (NEE) are associated with larger uptake during the growing season, which is also associated with larger ET fluxes. Our results also suggest that climate areas such as the San Joaquin Valley can sustain longer growing seasons leading to greater carbon uptake. In addition, varieties with a longer period from bloom to leaf senescence seem to positively affect carbon sequestration. These results illustrate the need to further investigate the different facets of managing perennial woody orchards to optimize greenhouse gas fluxes and carbon sequestration in the soil. We argue that it is critical to advance the understanding of cover crop impacts considering the different varieties in a cover crop mix, the management throughout the season (e.g., mowing, grazing, irrigation type), the soil properties, and the hydroclimatological context. Furthermore, we observe that orchard management focusing on maximizing yield showed much larger seasonal carbon uptake, even in the absence of cover crops. Thus, in the context of almond orchards, the management of by-products from almond production (e.g., hulls and shells) seems critical when assessing the sustainability of these agroecosystems. These results highlight the importance of further investigating the relationship between water and carbon in Mediterranean perennial woody crops to inform NbCS management and policies.
(a) Net ecosystem exchange (NEE) and (b) cumulative NEE from four almond orchards under different soil management practices: bare soil, naturally occurring winter cover crops (NOWCC), winter-seeded and spring-mowed cover crop (SMCC), and winter-seeded and graze cover crop (SGCC).
Citation: Bulletin of the American Meteorological Society 105, 1; 10.1175/BAMS-D-22-0118.1
T-REX research road map
The goals of the T-REX project are to develop the science and technology necessary to create the next generation of remote sensing–based decision support tools [e.g., OpenET; described in more detail by Melton et al. (2022)] for perennial woody crops, while also partnering with cooperators to ensure that the technologies and insights developed are integrated into industry and grower operations.
Satellite-based remote crop sensing, modeling, and novel analysis present great potential to support the transition toward climate-smart sustainable agroecosystems. However, there are still several key unknowns that need to be addressed to balance the goals of economic productivity and ecosystem sustainability. One critical area that requires further understanding is how individual trees/vines respond spatially and temporally to biotic and abiotic stress. This includes determining how management practices and environmental stressors affect crop physiological responses at various scales. We argue that more research is needed to address how leaf-level studies using gas exchange and water potential measurements can be scaled and interpreted at the canopy and whole-field levels. Another challenge is linking leaf, soil, and canopy-level spectroscopy and thermal information to satellite-collected data. This requires the development of novel machine learning and artificial intelligence algorithms that can integrate and analyze diverse datasets from multiple sources. Additionally, it will be critical to develop protocols for integrating these emerging technologies into farm management practices to ensure that the information generated is actionable and can be used to make informed decisions for precision crop management. While satellite-based remote crop sensing and modeling offer exciting opportunities for increasing the resiliency of woody perennial crops in semiarid regions, there is still much to be learned about how these technologies can be leveraged to achieve the goals of economic productivity and ecosystem sustainability goals. Our ongoing research and collaborations between scientists, farmers, and industry leaders will be critical in advancing our understanding and application of these emerging technologies.
Micrometeorological measurements are critical to evaluate the performance of remote sensing products. The T-REX project builds on several years of research focused on creating and advancing tools to inform irrigation decisions and water stress management in vineyards through the GRAPEX project. Beyond using the eddy covariance technique for remotely sensed ET product evaluation, our research aims to
- 1)increase the accuracy and quantify the uncertainty of micrometeorological-based ET estimations,
- 2)understand and quantify the role and interactions of canopy structures with turbulence within and above the canopy,
- 3)analyze the relationship between agricultural management and climate on water and carbon relationships at the orchard scale,
- 4)develop novel and low-cost alternatives for ET ground-truthing, and
- 5)link leaf-to-satellite observations for precision agriculture.
Thus, we have identified research priorities based on preliminary findings and the overall objectives underlying the T-REX network:
- 1)We will investigate, quantify, and model the impact of advection on ET.
- 2)We will investigate and model the role of canopy structures, density, and interrow surface characteristics on radiation, water, and carbon partitioning.
- 3)We will analyze the influence of climate and agricultural management practices on water–carbon relationships to identify how and when climate-smart and regenerative practices influence the resilience, water dynamics, and carbon-uptake capacity of orchards.
- 4)We will develop low-cost micrometeorological sensing techniques to increase the number of ET ground-truthing sites without compromising the quality of our evaluation standards.
- 5)We will scale leaf-level gas exchange, hyperspectral, and thermal observations to the canopy level. Then, we will identify the advantages and limitations of similar observations from UAS and satellites while recognizing canopy complexities.
Basic and applied research using UAS aims to evaluate and further advance the technology, protocols, algorithms, and results to quantify tree-scale water use and stress. As described in the case of satellite applications, several factors (e.g., advection) may also be affecting ET estimates based on UAS data retrievals. Moreover, additional factors only visible at the UAS scale need further study and improvement; hence, providing a robust option for almond producers in California and other regions where UAS are accessible would be possible.
For the research using UAS imagery, we aim to
- 1)develop protocols and refine algorithms that allow for an accurate description of the tree-interrow conditions, necessary for the application of highest possible resolution energy balance models;
- 2)understand and describe potential restrictions on the use of the technology (e.g., optimal flight time window/shadows occurrence) that ensures adequate estimation of ET;
- 3)integrate ground and micrometeorological measurements across the sites to describe almond water stress at tree scale using UAS information;
- 4)use the UAS technology in creative ways to support T-REX studies (e.g., collect vertical weather measurements for advection studies, provide the necessary information for leaf to satellite integration); and
- 5)evaluate usability of new or noncommon UAS imaging sensors (e.g., 10-band and SWIR spectral cameras) toward analysis supporting ET and water stress.
Our main priority is to reach a level of maturity in the estimation of ET and water stress using UAS and satellite-based remote sensing technology that allows for the transfer of knowledge and science to almond producers and similar interest organizations toward precision irrigation and management of the orchard block expanding the capabilities of the installed irrigation systems.
Concluding remarks
The preliminary results presented in this article illustrate the feasibility of developing the next generation of agriculture decision support tools based on remote sensing–based technologies, particularly for California almond growers. The DisALEXI modeling framework performs well in WWF and OLA almond orchards, with modeled daily ET estimates tracking closely with observed values (average R2 = 0.81) and errors (average MAE = 0.89 mm day−1) aligning with those presented for GRAPEX (MAE =1.09 mm day−1). However, each site offers a unique perspective for model evaluation efforts in almond orchards. Specifically, results to date support the hypothesis that advective conditions at site VAC cause negative bias in DisALEXI ET estimates as modeled values are unable to match observed magnitudes produced by the influx of additional energy. Efforts are underway to both better quantify the effects of advection on EC tower estimates and to provide more spatially representative estimates of ET for DisALEXI over thermally heterogeneous landscapes commonly associated with advective conditions. Furthermore, rapid declines in observed ET during hull split present a challenge for satellite-based ET modeling efforts, in large part due to the gap between Landsat (i.e., TIR imagery) overpasses. Many studies have evaluated supplementing additional satellite-based TIR platforms within the DisALEXI workflow, improving the temporal sampling of satellite-based ET (Gao et al. 2006; Anderson et al. 2021; Xue et al. 2021). Results are promising and offer a potential solution to the rapid decline in ET found in August at these sites. Specifically, we are working on implementing a data fusion process that was developed by Xue et al. (2022), where the additional platforms of Sentinel-2 and VIIRS are included to provide higher-temporal-resolution ET estimates capable of deciphering the rapid changes in ET found during hull split and between Landsat overpasses.
The successful expansion from wine grapes (GRAPEX) to almonds (T-REX) has allowed the development of projects in other perennial woody cropping systems in California. Specifically, the T-REX project team and its collaborators are currently expanding efforts into table grapes and olive and pistachio orchards (projected start date of spring 2023), with the intent to promote sustainable water use and improve soil carbon management in these systems. We hope that collaborative projects such as these, where stakeholder-driven applied research and basic research merge, can work as a template in promoting sustainable agricultural water use and carbon management in the state of California.
Acknowledgments.
We thank Olam Group, Westwind Farms Inc., and Sumit Sharma for access to almond orchard sites and support for deployment of meteorological flux towers. We thank Zac Ellis and Jim Eckberg for their valuable insights from the perspective of stakeholders (farm managers and the almond industry). We thank all members from the USDA-UC Davis McElrone Lab for their assistance on IOPs data collection. We thank Dr. Bonnie McGill for the illustrations created for this article (Figs. 1 and 2). Financial support for this research was provided by the California Department of Food and Agriculture (2020 CA Specialty Crop Block Grant 20-0001-031-SF), the Almond Board of California (WATER16 and WATER18), and USDA Agricultural Research Service. USDA is an equal opportunity provider and employer. On behalf of all authors, the corresponding authors state that there is no conflict of interest.
Data availability statement.
The data used in this research article are available upon request. Interested parties can contact the corresponding author to obtain access to the data. Data files include information about the location and format of the data, any necessary permissions required to access the data, and any restrictions on its use. The data will be made available in a timely fashion in compliance with any applicable data-sharing policies or regulations. We encourage others to use and build upon our findings, and we look forward to collaborating with researchers who wish to utilize our data.
References
Allen, R. G., L. S. Pereira, D. Raes, and M. Smith, 1998: Crop evapotranspiration: Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, 300 pp., www.fao.org/docrep/X0490E/X0490E00.htm.
Allen, R. G., M. Tasumi, and R. Trezza, 2007: Satellite-based energy balance for mapping evapotranspiration with internalized calibration (METRIC)—Model. J. Irrig. Drain. Eng., 133, 380–394, https://doi.org/10.1061/(ASCE)0733-9437(2007)133:4(380).
Anderson, M. C., J. Norman, G. Diak, W. Kustas, and J. Mecikalski, 1997: A two-source time-integrated model for estimating surface fluxes using thermal infrared remote sensing. Remote Sens. Environ., 60, 195–216, https://doi.org/10.1016/S0034-4257(96)00215-5.
Anderson, M. C., J. Norman, J. R. Mecikalski, R. D. Torn, W. P. Kustas, and J. B. Basara, 2004: A multiscale remote sensing model for disaggregating regional fluxes to micrometeorological scales. J. Hydrometeor., 5, 343–363, https://doi.org/10.1175/1525-7541(2004)005<0343:AMRSMF>2.0.CO;2.
Anderson, M. C., J. M. Norman, W. P. Kustas, F. Li, J. H. Prueger, and J. R. Mecikalski, 2005: Effects of vegetation clumping on two-source model estimates of surface energy fluxes from an agricultural landscape during SMACEX. J. Hydrometeor., 6, 892–909, https://doi.org/10.1175/JHM465.1.
Anderson, M. C., J. M. Norman, J. R. Mecikalski, J. A. Otkin, and W. P. Kustas, 2007a: A climatological study of evapotranspiration and moisture stress across the continental United States based on thermal remote sensing: 1. Model formulation. J. Geophys. Res., 112, D10117, https://doi.org/10.1029/2006JD007506.
Anderson, M. C., J. M. Norman, J. R. Mecikalski, J. A. Otkin, and W. P. Kustas, 2007b: A climatological study of evapotranspiration and moisture stress across the continental United States based on thermal remote sensing: 2. Surface moisture climatology. J. Geophys. Res., 112, D11112, https://doi.org/10.1029/2006JD007507.
Anderson, M. C., and Coauthors, 2021: Interoperability of ECOSTRESS and Landsat for mapping evapotranspiration time series at sub-field scales. Remote Sens. Environ., 252, 112189, https://doi.org/10.1016/j.rse.2020.112189.
Aubinet, M., T. Vesala, and D. Papale, 2012: Eddy Covariance: A Practical Guide to Measurement and Data Analysis. Springer, 438 pp., https://doi.org/10.1007/978-94-007-2351-1.
Bambach, N. E., and Coauthors, 2022a: Evapotranspiration uncertainty at micrometeorological scales: The impact of the eddy covariance energy imbalance and correction methods. Irrig. Sci., 40, 445–461, https://doi.org/10.1007/s00271-022-00783-1.
Bambach, N. E., M. E. Gilbert, and K. T. Paw U, 2022b: Introducing a dynamic photosynthetic model of photoinhibition, heat, and water stress in the next-generation land surface model ACASA. Agric. For. Meteor., 312, 108702, https://doi.org/10.1016/j.agrformet.2021.108702.
Bastiaanssen, W. G., M. Menenti, R. Feddes, and A. Holtslag, 1998: A remote sensing surface energy balance algorithm for land (SEBAL). 1. Formulation. J. Hydrol., 212–213, 198–212, https://doi.org/10.1016/S0022-1694(98)00253-4.
Bellvert, J., K. Adeline, S. Baram, L. Pierce, B. L. Sanden, and D. R. Smart, 2018: Monitoring crop evapotranspiration and crop coefficients over an almond and pistachio orchard throughout remote sensing. Remote Sens., 10, 2001, https://doi.org/10.3390/rs10122001.
Burchard-Levine, V., and Coauthors, 2022a: Application of a remote-sensing three-source energy balance model to improve evapotranspiration partitioning in vineyards. Irrig. Sci., 40, 593–608, https://doi.org/10.1007/s00271-022-00787-x.
Burchard-Levine, V., and Coauthors, 2022b: A remote sensing-based three-source energy balance model to improve global estimations of evapotranspiration in semi-arid tree-grass ecosystems. Global Change Biol., 28, 1493–1515, https://doi.org/10.1111/gcb.16002.
Cammalleri, C., M. Anderson, F. Gao, C. Hain, and W. Kustas, 2014a: Mapping daily evapotranspiration at field scales over rainfed and irrigated agricultural areas using remote sensing data fusion. Agric. For. Meteor., 186, 1–11, https://doi.org/10.1016/j.agrformet.2013.11.001.
Cammalleri, C., M. Anderson, and W. Kustas, 2014b: Upscaling of evapotranspiration fluxes from instantaneous to daytime scales for thermal remote sensing applications. Hydrol. Earth Syst. Sci., 18, 1885–1894, https://doi.org/10.5194/hess-18-1885-2014.
CDFA, 2022: California agriculture exports 2019–2020. California Department of Food and Agriculture, 15 pp., https://www.cdfa.ca.gov/Statistics/PDFs/2020_Exports_Publication.pdf.
Dee, D., M. Balmaseda, G. Balsamo, R. Engelen, A. Simmons, and J.-N. Thépaut, 2014: Toward a consistent reanalysis of the climate system. Bull. Amer. Meteor. Soc., 95, 1235–1248, https://doi.org/10.1175/BAMS-D-13-00043.1.
Deng, Z., H. K. Vice, M. E. Gilbert, M. A. Adams, and T. N. Buckley, 2021: A double-ratio method to measure fast, slow and reverse sap flows. Tree Physiol., 41, 2438–2453, https://doi.org/10.1093/treephys/tpab081.
Douglas, E., A. Beltrán-Przekurat, D. Niyogi, R. Pielke Sr., and C. Vörösmarty, 2009: The impact of agricultural intensification and irrigation on land–atmosphere interactions and Indian monsoon precipitation-a mesoscale modeling perspective. Global Planet. Change, 67, 117–128, https://doi.org/10.1016/j.gloplacha.2008.12.007.
Drechsler, K., A. Fulton, and I. Kisekka, 2022: Crop coefficients and water use of young almond orchards. Irrig. Sci., 40, 379–395, https://doi.org/10.1007/s00271-022-00786-y.
Ebert, L. A., A. Talib, S. C. Zipper, A. R. Desai, K. T. Paw U, A. J. Chisholm, J. Prater, and M. A. Nocco, 2022: How high to fly? Mapping evapotranspiration from remotely piloted aircrafts at different elevations. Remote Sens., 14, 1660, https://doi.org/10.3390/rs14071660.
Foken, T., 2017: Experimental methods for estimating the fluxes of energy and matter. Micrometeorology, Springer, 143–205, https://doi.org/10.1007/978-3-642-25440-6_4.
Foken, T., and C. J. Napo, 2008: Micrometeorology. Vol. 2. Springer, 306 pp.
Fry, J. A., and Coauthors, 2011: Completion of the 2006 national land cover database for the conterminous United States. Photogramm. Eng. Remote Sens., 77, 858–864.
Gao, F., J. Masek, M. Schwaller, and F. Hall, 2006: On the blending of the Landsat and MODIS surface reflectance: Predicting daily Landsat surface reflectance. IEEE Trans. Geosci. Remote Sens., 44, 2207–2218, https://doi.org/10.1109/TGRS.2006.872081.
Gao, F., W. P. Kustas, and M. C. Anderson, 2012: A data mining approach for sharpening thermal satellite imagery over land. Remote Sens., 4, 3287–3319, https://doi.org/10.3390/rs4113287.
Gao, F., M. Anderson, C. Daughtry, A. Karnieli, D. Hively, and W. Kustas, 2020: A within-season approach for detecting early growth stages in corn and soybean using high temporal and spatial resolution imagery. Remote Sens. Environ., 242, 111752, https://doi.org/10.1016/j.rse.2020.111752.
Gao, R., and Coauthors, 2023: ET partitioning assessment using the TSEB model and sUAS information across California Central Valley vineyards. Remote Sens., 15, 756, https://doi.org/10.3390/rs15030756.
Gilson, E., and S. Kenehan, 2018: Food, Environment, and Climate Change: Justice at the Intersections. Rowman and Littlefield, 342 pp.
Girona, J., M. Mata, and J. Marsal, 2005: Regulated deficit irrigation during the kernel-filling period and optimal irrigation rates in almond. Agric. Water Manage., 75, 152–167, https://doi.org/10.1016/j.agwat.2004.12.008.
Hansen, M. C., R. S. DeFries, J. R. Townshend, and R. Sohlberg, 2000: Global land cover classification at 1 km spatial resolution using a classification tree approach. Int. J. Remote Sens., 21, 1331–1364, https://doi.org/10.1080/014311600210209.
Huang, X., and D. L. Swain, 2022: Climate change is increasing the risk of a California megaflood. Sci. Adv., 8, eabq0995, https://doi.org/10.1126/sciadv.abq0995.
Jha, G., F. Nicolas, R. Schmidt, K. Suvočarev, D. Diaz, I. Kisekka, K. Scow, and M. A. Nocco, 2022: Irrigation decision support systems (IDSS) for California’s water-nutrient-energy nexus. Agronomy, 12, 1962, https://doi.org/10.3390/agronomy12081962.
Kang, Y., and Coauthors, 2022: Evaluation of satellite leaf area index in California vineyards for improving water use estimation. Irrig. Sci., 40, 531–551, https://doi.org/10.1007/s00271-022-00798-8.
Kljun, N., P. Calanca, M. Rotach, and H. P. Schmid, 2015: A simple two-dimensional parameterisation for Flux Footprint Prediction (FFP). Geosci. Model Dev., 8, 3695–3713, https://doi.org/10.5194/gmd-8-3695-2015.
Knipper, K. R., and Coauthors, 2019: Evapotranspiration estimates derived using thermal-based satellite remote sensing and data fusion for irrigation management in California vineyards. Irrig. Sci., 37, 431–449, https://doi.org/10.1007/s00271-018-0591-y.
Knipper, K. R., and Coauthors, 2020: Using high-spatiotemporal thermal satellite ET retrievals to monitor water use over California vineyards of different climate, vine variety and trellis design. Agric. Water Manage., 241, 106361, https://doi.org/10.1016/j.agwat.2020.106361.
Krishnamurthy R, P. K., J. B. Fisher, R. J. Choularton, and P. M. Kareiva, 2022: Anticipating drought-related food security changes. Nat. Sustainability, 5, 956–964, https://doi.org/10.1038/s41893-022-00962-0.
Kustas, W. P., and J. M. Norman, 1999: Evaluation of soil and vegetation heat flux predictions using a simple two-source model with radiometric temperatures for partial canopy cover. Agric. For. Meteor., 94, 13–29, https://doi.org/10.1016/S0168-1923(99)00005-2.
Kustas, W. P., and Coauthors, 2018: The Grape Remote Sensing Atmospheric Profile and Evapotranspiration Experiment. Bull. Amer. Meteor. Soc., 99, 1791–1812, https://doi.org/10.1175/BAMS-D-16-0244.1.
Lehmann, J., D. A. Bossio, I. Kögel-Knabner, and M. C. Rillig, 2020: The concept and future prospects of soil health. Nat. Rev. Earth Environ., 1, 544–553, https://doi.org/10.1038/s43017-020-0080-8.
Leuning, R., E. van Gorsel, W. J. Massman, and P. R. Isaac, 2012: Reflections on the surface energy imbalance problem. Agric. For. Meteor., 156, 65–74, https://doi.org/10.1016/j.agrformet.2011.12.002.
Li, S., S. Kang, L. Zhang, J. Zhang, T. Du, L. Tong, and R. Ding, 2016: Evaluation of six potential evapotranspiration models for estimating crop potential and actual evapotranspiration in arid regions. J. Hydrol., 543, 450–461, https://doi.org/10.1016/j.jhydrol.2016.10.022.
Liu, X., and Coauthors, 2020: Changes in nut consumption and subsequent cardiovascular disease risk among us men and women: 3 large prospective cohort studies. J. Amer. Heart Assoc., 9, e013877, https://doi.org/10.1161/JAHA.119.013877.
Massman, W., 2000: A simple method for estimating frequency response corrections for eddy covariance systems. Agric. For. Meteor., 104, 185–198, https://doi.org/10.1016/S0168-1923(00)00164-7.
Mauder, M., M. Cuntz, C. Drüe, A. Graf, C. Rebmann, H. P. Schmid, M. Schmidt, and R. Steinbrecher, 2013: A strategy for quality and uncertainty assessment of long-term eddy-covariance measurements. Agric. For. Meteor., 169, 122–135, https://doi.org/10.1016/j.agrformet.2012.09.006.
Mauder, M., T. Foken, and J. Cuxart, 2020: Surface-energy-balance closure over land: A review. Bound.-Layer Meteor., 177, 395–426, https://doi.org/10.1007/s10546-020-00529-6.
Melton, F. S., and Coauthors, 2022: OpenET: Filling a critical data gap in water management for the western United States. J. Amer. Water Resour. Assoc., 58, 971–994, https://doi.org/10.1111/1752-1688.12956.
Moore, C., 1986: Frequency response corrections for eddy correlation systems. Bound.-Layer Meteor., 37, 17–35, https://doi.org/10.1007/BF00122754.
Moran, M. S., 2004: Thermal infrared measurement as an indicator of plant ecosystem health. Thermal Remote Sensing in Land Surface Processes, CRC Press, 256–282.
Nassar, A., and Coauthors, 2020: Influence of model grid size on the estimation of surface fluxes using the two source energy balance model and sUAS imagery in vineyards. Remote Sens., 12, 342, https://doi.org/10.3390/rs12030342.
Nassar, A., and Coauthors, 2021: Assessing daily evapotranspiration methodologies from one-time-of-day sUAS and EC information in the GRAPEX project. Remote Sens., 13, 2887, https://doi.org/10.3390/rs13152887.
Newton, P., N. Civita, L. Frankel-Goldwater, K. Bartel, and C. Johns, 2020: What is regenerative agriculture? A review of scholar and practitioner definitions based on processes and outcomes. Front. Sustainable Food Syst., 4, 577723, https://doi.org/10.3389/fsufs.2020.577723.
Nieto, H., and Coauthors, 2019: Evaluation of TSEB turbulent fluxes using different methods for the retrieval of soil and canopy component temperatures from UAV thermal and multispectral imagery. Irrig. Sci., 37, 389–406, https://doi.org/10.1007/s00271-018-0585-9.
Nocco, M. A., N. Weeth Feinstein, M. N. Stock, B. M. McGill, and C. J. Kucharik, 2020: Knowledge co-production with agricultural trade associations. Water, 12, 3236, https://doi.org/10.3390/w12113236.
Norman, J. M., W. P. Kustas, and K. S. Humes, 1995: Source approach for estimating soil and vegetation energy fluxes in observations of directional radiometric surface temperature. Agric. For. Meteor., 77, 263–293, https://doi.org/10.1016/0168-1923(95)02265-Y.
Norman, J. M., and Coauthors, 2003: Remote sensing of surface energy fluxes at 101-m pixel resolutions. Water Resour. Res., 39, 1221, https://doi.org/10.1029/2002WR001775.
Novick, K. A., and Coauthors, 2022: Informing nature-based climate solutions for the United States with the best-available science. Global Change Biol., 28, 3778–3794, https://doi.org/10.1111/gcb.16156.
Oreskes, N., K. Shrader-Frechette, and K. Belitz, 1994: Verification, validation, and confirmation of numerical models in the Earth sciences. Science, 263, 641–646, https://doi.org/10.1126/science.263.5147.641.
Ortiz-Bobea, A., T. R. Ault, C. M. Carrillo, R. G. Chambers, and D. B. Lobell, 2021: Anthropogenic climate change has slowed global agricultural productivity growth. Nat. Climate Change, 11, 306–312, https://doi.org/10.1038/s41558-021-01000-1.
Paw U, K. T., D. D. Baldocchi, T. P. Meyers, and K. B. Wilson, 2000: Correction of eddy-covariance measurements incorporating both advective effects and density fluxes. Bound.-Layer Meteor., 97, 487–511, https://doi.org/10.1023/A:1002786702909.
Paw U, K. T., S. Wharton, and J. Kochendorfer, 2003: Evapotranspiration: Measuring and modeling. IV Int. Symp. on Irrigation of Horticultural Crops 664, Davis, CA, International Society for Horticultural Science, 537–554, https://www.ishs.org/ishs-article/664_68.
Potchter, O., D. Goldman, D. Kadish, and D. Iluz, 2008: The oasis effect in an extremely hot and arid climate: The case of southern Israel. J. Arid Environ., 72, 1721–1733, https://doi.org/10.1016/j.jaridenv.2008.03.004.
Pozníková, G., M. Fischer, B. van Kesteren, M. Orság, P. Hlavinka, Z. Žalud, and M. Trnka, 2018: Quantifying turbulent energy fluxes and evapotranspiration in agricultural field conditions: A comparison of micrometeorological methods. Agric. Water Manage., 209, 249–263, https://doi.org/10.1016/j.agwat.2018.07.041.
Priestley, C. H. B., and R. J. Taylor, 1972: On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Wea. Rev., 100, 81–92, https://doi.org/10.1175/1520-0493(1972)100<0081:OTAOSH>2.3.CO;2.
Prueger, J. H., and Coauthors, 2019: Crop water stress index of an irrigated vineyard in the central valley of California. Irrig. Sci., 37, 297–313, https://doi.org/10.1007/s00271-018-0598-4.
Rhoades, A. M., and Coauthors, 2022: Asymmetric emergence of low-to-no snow in the midlatitudes of the American Cordillera. Nat. Climate Change, 12, 1151–1159, https://doi.org/10.1038/s41558-022-01518-y.
Roberts, M., A. Milman, and W. Blomquist, 2021: The Sustainable Groundwater Management Act (SGMA): California’s prescription for common challenges of groundwater governance. Water Resilience: Management and Governance in Times of Change, Springer, 41–63, https://doi.org/10.1007/978-3-030-48110-0_3.
Ruehr, S., X. Lee, R. Smith, X. Li, Z. Xu, S. Liu, X. Yang, and Y. Zhou, 2020: A mechanistic investigation of the oasis effect in the Zhangye cropland in semiarid western China. J. Arid Environ., 176, 104120, https://doi.org/10.1016/j.jaridenv.2020.104120.
Schotanus, P., F. Nieuwstadt, and H. De Bruin, 1983: Temperature measurement with a sonic anemometer and its application to heat and moisture fluxes. Bound.-Layer Meteor., 26, 81–93, https://doi.org/10.1007/BF00164332.
Senay, G. B., M. Budde, J. P. Verdin, and A. M. Melesse, 2007: A coupled remote sensing and simplified surface energy balance approach to estimate actual evapotranspiration from irrigated fields. Sensors, 7, 979–1000, https://doi.org/10.3390/s7060979.
Siirila-Woodburn, E. R., and Coauthors, 2021: A low-to-no snow future and its impacts on water resources in the western United States. Nat. Rev. Earth Environ., 2, 800–819, https://doi.org/10.1038/s43017-021-00219-y.
Swain, D. L., B. Langenbrunner, J. D. Neelin, and A. Hall, 2018: Increasing precipitation volatility in twenty-first-century California. Nat. Climate Change, 8, 427–433, https://doi.org/10.1038/s41558-018-0140-y.
Volk, J. M., J. Huntington, R. Allen, F. Melton, M. Anderson, and A. Kilic, 2021: flux-data-qaqa: A Python package for energy balance closure and post-processing of eddy flux data. J. Open Source Software, 6, 3418, https://doi.org/10.21105/joss.03418.
Volk, J. M., and Coauthors, 2023: Development of a benchmark eddy flux evapotranspiration dataset for evaluation of satellite-driven evapotranspiration models over the conus. Agric. For. Meteor., 331, 109307, https://doi.org/10.1016/j.agrformet.2023.109307.
Wahl, E. R., E. Zorita, H. F. Diaz, and A. Hoell, 2022: Southwestern United States drought of the 21st century presages drier conditions into the future. Commun. Earth Environ., 3, 202, https://doi.org/10.1038/s43247-022-00532-4.
Wang, T., J. Verfaillie, D. Szutu, and D. Baldocchi, 2023: Handily measuring sensible and latent heat exchanges at a bargain: A test of the variance-Bowen ratio approach. Agric. For. Meteor., 333, 109399, https://doi.org/10.1016/j.agrformet.2023.109399.
Webb, E. K., G. I. Pearman, and R. Leuning, 1980: Correction of flux measurements for density effects due to heat and water vapour transfer. Quart. J. Roy. Meteor. Soc., 106, 85–100, https://doi.org/10.1002/qj.49710644707.
Wheeler, T., and J. Von Braun, 2013: Climate change impacts on global food security. Science, 341, 508–513, https://doi.org/10.1126/science.1239402.
Wilson, S. M., 2010: Environmental justice movement: A review of history, research, and public health issues. J. Public Manage. Soc. Policy, 16, 19–50.
Xue, J., M. C. Anderson, F. Gao, C. Hain, Y. Yang, K. R. Knipper, W. P. Kustas, and Y. Yang, 2021: Mapping daily evapotranspiration at field scale using the harmonized Landsat and Sentinel-2 dataset, with sharpened VIIRS as a Sentinel-2 thermal proxy. Remote Sens., 13, 3420, https://doi.org/10.3390/rs13173420.
Xue, J., and Coauthors, 2022: Improving the spatiotemporal resolution of remotely sensed ET information for water management through Landsat, Sentinel-2, ECOSTRESS and VIIRS data fusion. Irrig. Sci., 40, 609–634, https://doi.org/10.1007/s00271-022-00799-7.
