Abstract
Assessing changes in future aridity requires an understanding of variations in the atmospheric demand for water. Such assessments are often driven by estimations of potential evapotranspiration (ETP) and/or reference evapotranspiration (ET0), yet no comprehensive and validated estimate of these climate metrics exists to date from the Coupled Model Intercomparison Project 6 (CMIP6). Here we describe the development and validation of a published dataset of global monthly estimates of the Penman-Monteith derived ET0, its advective and radiation components, Priestley-Taylor derived ETP, and vapor pressure deficit from 16 CMIP6 projections and four emissions scenarios. Historical validation of the ensemble of CMIP6 evaporative demand shows general agreement with observationally-derived baselines of ET0 and ETP from the Climate Research Unit (CRU) and ERA5-Land reanalysis products, with GCM biases driven primarily by regional differences in modeled humidity and advective contributions to ET0. Overall, evaporative demand is projected to increase across all emissions scenarios, with the largest increases over polar regions, and with a larger contribution from advection particularly for regions with higher baseline ET0.
Similar content being viewed by others
Background & Summary
The evaporative demand of surface water to the atmosphere has been shown to be an essential driver and explanatory variable for numerous evaluations of changes in surface freshwater availability, agricultural water demand and drought risk1,2,3,4,5,6,7,8,9. Despite its ubiquitous use in analyses of changes in regional and global hydroclimates, there is a notable absence of an organized dataset of estimates of evaporative demand from the latest generation of Coupled Model Intercomparison Project 6 (CMIP6) General Circulation Models (GCMs). To address this need, we have developed monthly estimates of evaporative demand across a multi-model ensemble of CMIP6 GCM simulations These 16 GCMS are hereafter referred to as the CMIP6-MME.
Debate over the utility of the differing methods available for estimating evaporative demand5,10,11,12 and limitations in the availability of certain hydroclimate variables required for those estimations can lead to a large range of differing methodologies applied for the estimations of ETP13, ET014,15,16, and VPD16,17. The purpose of this dataset18 is to serve as a tool for future hydroclimate analyses and to promote methodological consistency across studies that utilize future estimations of evaporative demand. To this end, we provide monthly estimations using two of the most commonly used methods: the Priestley-Taylor potential evapotranspiration (ETP)13 method, and the FAO-56 Penman-Monteith reference evapotranspiration (ET0)16 method for both short grass and alfalfa vegetation land-cover. We also provide a monthly estimate of a key driving variable, vapor pressure deficit (VPD) (Sec. 2.1), that is commonly used to assess ecological impacts. All quantities are provided for the historical simulation period (1850–2014) and four future (2015–2100) CMIP6 ScenarioMIP shared socioeconomic pathways (SSPs), 1–2.6, 2–4.5, 3–7.0, and 5–8.519,20. These estimations are generated for the CMIP6-MME, of which ensemble members are selected based on the availability of all climate variables needed to complete the multiple estimations evaporative demand described below.
Following the development of the CMIP6-MME of ETP, ET0, and VPD, we perform a validation using observation-based historical baselines of both the monthly estimations of ET0 from the Climate Research Unit (CRU) dataset21 and estimations of ETP and ET0 derived from the ERA5-Land reanalysis22. Along with the expected variation in performance due to differences in GCM simulated historical hydroclimates23,24, we note a general coherence in the bias that exists across all GCMs within both ETP and ET0 that likely arise due the assumptions embedded within each methodology. Specifically, we evaluate differences the two major components evaporative demand, the advection and radiation components, from each GCM using the same components from ERA5-Land as the baseline to gain more insight into the sources of bias.
In sum, this data descriptor addresses the need for a comprehensive, methodologically consistent archive of future changes in multiple measures of evaporative demand from the latest suite of climate projections. We generate CMIP6-derived estimates of ETP, ET0, and VPD, validate the dataset over the historical period with multiple observational datasets, and conduct a brief evaluation of projected future changes in ETP and ET0. This dataset18 is built using methods developed for multi-disciplinary hydroclimate and agricultural analyses13,15, with the goal of providing the user community a consistent and comprehensive product useful for understanding future changes in water availability. Section 2 describes the methods used to estimate ETP, ET0, VPD, and other intermediate step variables along with a description of the CMIP6-MME. Section 3 contains a description of where to access and contents of the data repository. Section 4 contains the technical validation of the multiple estimates of evaporative demand and section 5 directs users to the Python codes used to calculate all of the variables described in section 2.
Methods
The first evaporative demand methodology described and applied herein is the FA0-56 Penman-Monteith formulation (Sec. 2.1) for estimating reference evapotranspiration (ET0)14,16 which has the most inputs relative to other methods. Within this section we describe the calculation of monthly net surface-radiation (Sec. 2.1.1), 2-meter wind-speed (Sec. 2.1.2), vapor pressure (Sec. 2.1.3), and saturated vapor pressure (Sec. 2.1.4) from both the GCMs and the ERA5-Land reanalysis. Further, we describe the iterations of ET0 estimations using two different assumptions of vegetation cover: short grass and alfalfa (Sec. 2.1.5). In Sec. 2.2, we describe the method used to determine the advective and radiation components of ET0 and how those components should be interpreted. The subsequent section contains the description of the Priestley-Taylor potential evapotranspiration formulation (ETP) (Sec. 2.3) and the terms used in this dataset to generate monthly estimations of ETP. The ETP calculated is not split into advective and radiation components due to the choice of coefficient that represents the relative contribution of each component (described further in Sec. 2.3). The final section of the methods (Sec. 2.4) describes the land masking that we apply as a last step for preparation of the data to align with the intended domain of application of the Penman-Monteith and Priestley-Taylor methodologies. Finally, in Sec. 2.5, we describe the set of GCMs were selected based on availability of data and used to generate the CMIP-MME estimates of potential and reference evapotranspiration.
FAO 56 Penman-Monteith Reference Evapotranspiration
The FAO-56 Penman-Monteith ET0 method was developed to estimate the evaporative demand of a well-irrigated crop land, with assumptions about the land-surface vegetation allowing for ET0 to be applied in the approximation of required water usage in agricultural operation (R. G. Allen et al.25; R. G. Allen & Pruitt16). It has been applied widely (e.g. Ashofteh Parisa-Sadat et al.26; Beven27; Kay & Davies28; Kingston et al.29; McKenney & Rosenberg30; Tabari & Hosseinzadeh Talaee31) and is generally considered among the most accurate methods, despite it’s relatively large data requirement32,33. Included in the following subsections are descriptions of the ET0 method that we applied in the generation of the subject dataset18 and all of the intermediate steps required to convert the input variables derived from CMIP6-MME output to represent their form as described in the FA0-56 guidelines. To provide an estimate of the two most commonly applied crop types to the formulation, we generate estimations for both short-grass and alfalfa crops in this dataset18 (described further in Sec. 2.1.5).
Where ET0 is reference evapotranspiration (mm day−1), Δ is the slope of the saturated vapor pressure curve (kPa C−1), Rn is the net radiation (MJ day−1), γ is the psychrometric constant (kPa K−1), Cn and Cd are empirically derived values for a given reference crop from Allen et al.25 (s m−1), T is monthly mean air temperature (°C), (es-ea) is the saturation vapor pressure deficit (kPa), and u2 is the monthly average wind speed at 2 meters (m s−1).
Net Surface-Radiation
Net radiation (Rn) is calculated here using sensible (hfss) and latent (hfls) heat fluxes from the GCM simulations to implicitly account for the contribution of the ground heat flux (G) to the net radiation balance as is described in Allen et al.25. This approach is consistent with methods applied in previous evaluations of the surface-energy balance2 as follows:
Though the ground heat flux has been shown in previous studies to be nearly negligible at the monthly time scale2, this approach allows us to avoid assumption of zero G in our estimation of ET0. We followed this method of using the latent and sensible heat flux here as our analysis using the same category of GCM output data2.
2-Meter Wind-speed
FAO-56 guidelines call for the use of the surface windspeeds at a height of 2-meter above the land surface25, but all CMIP6-MME monthly windspeed output is documented as the 10-meter surface windspeeds. Furthermore, the monthly average 10-meter surface windspeeds we use here are directly computed outputs for each model within the CMIP6 archive (variable name: SfcWind). This approach allows for a more effective capturing of the monthly turbulent flux contribution to ET0 as opposed to computing average wind speed from north-south and east-west velocities which yields much smaller magnitude monthly wind speed values. To convert windspeeds to match the specifications of FAO-56, we apply the equation from the FAO-56 guidelines25:
Where u2 is the 2-meter monthly averaged surface windspeed, uz is the monthly averaged surface windspeed at height z, and z is the height above the land surface expressed in meters of the surface windspeed of the input data, in this case 10 meters.
Vapor Pressure
Following the guidance of the FAO-56 manual25, we estimate the saturated vapor pressure (es) using the monthly mean air temperature from each of the GCMs. Then we utilize es and the monthly mean relative humidity (hr) to calculate a monthly vapor pressure deficit (VPD) and the slope of the saturated vapor pressure curve25. Though this variable is an intermediate step in the estimation of the evaporative demand, VPD has utility in a variety of agricultural34,35 and ecological36,37,38 applications due to its modulation of stomatal transpiration and relationship with vegetation production, so we have included this variable in the published dataset18.
Saturated Vapor Pressure
The saturated vapor pressure is calculated directly from the monthly mean air temperature (Tavg) and represents the theoretical pressure of an air mass at saturation for the given Tavg. It is calculated using formulas from25,39:
The FAO56 guidelines suggest that the use of mean of maximum and minimum air temperatures (Tmax and Tmin respectively) for use in the saturated vapor pressure on daily timescales is preferable to using daily mean air temperature due to the non-linearity of the diurnal temperature variations, particularly as it can bias the estimation of relative humidity (Allen et al.25). However, authors tested the use of monthly Tmax and Tmin in the estimation of es and compared with the use of monthly Tavg and found marginal differences in the resulting ET0 (not shown). This paired with the lack of availability of Tmin and Tmax from the CMIP6 archive for all models within the CMIP6-MME used here, it is preferable to use the monthly Tavg values in this step.
Vapor Pressure Deficit (VPD)
VPD is calculated using the relative humidity (hr) and es to estimate the difference between the actual vapor pressure of the air and the saturated vapor pressure of the air on a monthly time scale. VPD is calculated as:
Saturated Vapor Pressure Curve
The final intermediate variable required for both the ET0 and ETP (described below) is the slope of the saturated vapor pressure curve (∆). ∆ is the instantaneous rate at which the saturation vapor pressure changes at a given temperature and is calculated as:
Vegetation Parameters
We apply two formulations of the vegetation parameters that we calibrated for use in the Allen et al.16 formulation. Guidance from agriculturally focused evaporative demand analyses40 suggests that short-grass and alfalfa are the most widely applied crop types for estimation of ET0, which is the justification for our selection. Parameter values for Cn and Cd (Eq. 1) are 900 and 0.34 for respectively for short grass and 1600 and 0.38 for alfalfa16. These values represent the vegetation resistance terms within the advective portion of the ET0 formulation and were empirically derived in25.
Radiation and Advection components of ET0
To evaluate the relative contributions of evaporative demand components to the historical ET0 and future changes to ET0, we split the numerator of Eq. 1 to serve as close approximations of the radiation (the 0.408ΔRn numerator term) and advection (the \(\gamma \left(\frac{{C}_{n}}{T+273}\right)\left({e}_{s}-{e}_{a}\right)\) numerator term) components. One can observe that both terms, ET0,Rad (Eq. 7) and ET0,Adv (Eq. 8) have the same denominator that contains a windspeed term, so it is not to be considered the pure separation of radiation and advection, simply a rough approximation of the relative contribution of each term. When one evaluates the ratio of the two terms, as is done in the technical validation of this manuscript, the advective component that exists in the denominators of the separated components (Eqs. 7, 8) cancel out it then becomes a comparison the relative magnitudes of the radiation and advection terms in the numerator. To be explicit about the equations used, equations of the radiation and advection components of ET0 are shown as Eqs. (7, 8) respectively.
Priestley-Taylor
The Priestley-Taylor method is applied for the quantification of monthly radiation-based ETP13, both for comparison to the ET0 method of estimating evaporative demand and to allow for evaluation of projected changes to evaporative demand driven primarily by increased net radiation at the surface. The Priestley-Taylor methodology assumes a constant ratio of the advective and radiative components of ETP, therefore its inclusion in this dataset18 allows for evaluation of how the increases in surface energy will drive changes to evaporative demand without the inclusion of more uncertain climate parameters such as wind speed or estimates of vegetation contributions. The equation for the Priestley-Taylor formulation of ETP is as follows:
Where Δ, Rn, and γ were previously defined in section 2.1 and \(\alpha \) is a constant equal to 1.26 empirically derived in13. The \(\alpha \) coefficient of 1.26, which is most commonly applied to empirically account for the non-radiative components of PET for humid regions5,13, is held constant to avoid introducing uncertainty associated with regionalized calibration of the parameter. Other studies have shown that the coefficient can be varied spatially with a higher value generally corresponding to PET estimation in more arid regions41,42. If users wish to estimate ETP using distributed coefficients, then we recommend a simple transformation by dividing the entire dataset by 1.26 to which the desired distributed coefficients could be applied.
Land Masking
The methodologies presented here were explicitly developed for application to land-areas, as opposed to open water areas25. Because of this, we mask out all oceans and large water body grid cells using land-sea masks obtained for each GCM, where grid cells containing less than 5% of the total area labelled as land were screened and replaced with null values. We include an archive of the land-sea masks in the published dataset18 for the convenience of data users.
CMIP6-MME
The CMIP6-MME used to generate this dataset18 was selected based on the availability of data needed to generate monthly estimations of ETP and ET0. Input variables were downloaded from multiple ESGF data nodes at monthly resolution for each of the CMIP6-MME described in Table 1. All data were checked for quality control to screen missing data or data corruption for all input variables. Only one realization from each GCM is used in this dataset18, primarily due to the lack of consistency in the number of individual GCM realizations from each modelling organization. The Python code used to generate this dataset18 is provided for those who wish to extend this analysis to multiple realizations from individual models.
For each of the 16 GCMs in Table 1, monthly estimations of ETP, ET0, components of ET0, and VPD were generated for the historical CMIP6 simulation period (1850–2014) and for the future period of 2015–2099 using four future emission scenarios (SSPs 1–2.6, 2–4.5, 3–7.0. and 5–8.5). The only exception is the lack of a SSP 3–7.0 scenario simulation for the HadGEM3-GC31-LL due to a lack of publication of that specific simulation being published on the ESGF data node. All GCM derived variables contained in the dataset are produced in the native resolution of the parent GCM simulation.
Data Records
This dataset18 contains NetCDF files of estimated ETP, multiple estimations of ET0 using vegetation parameters for both short-grass and alfalfa, the advective and radiative components of ET0, and estimations of VPD that were calculated in the process of estimating ET0. All NetCDF files contain information on the date of generation of the dataset, authors of the dataset, variable IDs, input variable information, and attributes of the parent GCM simulations used to generate the estimations of all variables described above. The link to the archived dataset can be found here: https://doi.org/10.5281/zenodo.778975918.
Each NetCDF file contains one variable from one GCM for either the historical simulation or one of the four future emission scenarios that were selected for use in this dataset18. The naming convention used for each NetCDF file is consistent with the CMIP6 ESGF naming conventions including variable, frequency, GCM name, scenario, variant ID, grid ID, and date range. Each file for the derived variables within this dataset contains data on the native grid of the parent GCM that was used to generate the estimations described above. Additionally, we include a single land-mask for each individual GCM to clarify which grid cells were selected as land-surface grid cells. Further description of the file structure and organization of data within the dataset18 can found in the “dataset_README.txt” file.
Technical Validation
For the validation of the ETP and ET0 estimations across the ensemble of GCMs, we use 30-year means of monthly estimated evaporative demand rates and 30-year mean annual total evaporative demand depth on a grid cell basis for both the GCMs and the historical baseline datasets. The use of long-term means on a grid cell basis permits a validation of the climatic evaporative demand and an evaluation of the spatial coherence of the simulated climate within the ensemble of GCMs when compared to the multiple historical baselines. For this analysis, we independently generate monthly estimates of ETP and ET0 from ERA5-Land in order to ensure consistency with our methods, though published hourly estimates of ET043 could also serve as a baseline for observationally based historical ET0.
The validation period of 1951–1980 coincides with the earliest available data from the ERA5-Land reanalysis and an overlapping period within the historical simulations of the GCMs with minimal acceleration of greenhouse gas emission impacts. We acknowledge that differences in the sensitivities of GCMs within CMIP623,24 could result in the inclusion of GCMs with significant warming even within the period of 1951–1980, but the impact of any warming trends in this early historical period is likely to be minimal because we are considering the long-term mean of ETP and ET0 rates.
To compute the spatial correlation and percent bias between estimated fluxes within the GCMs and the observational datasets we align the GCM-simulated evaporative demand fluxes to the 0.5° × 0.5° grid cell resolution of the CRU21 and ERA5-Land reanalysis fluxes using a conservative remapping algorithm44. The aligning of the GCM derived ETP and ET0 is used only for the validation of the dataset18 during the historical period and is not a feature of the dataset as published; all GCM estimated evaporative demand terms in the published dataset18 remain at the native resolution of the respective GCM simulation.
Historical Simulations
Comparisons of the 30-year mean annual rate of ETP and ET0 (mm/day) reveal a strong one-to-one relationship between the GCM simulations and historical baselines during the period 1951–1980 (Fig. 1). The global performance of the GCM derived estimations of ET0 and ETP, is evaluated based on the global average of the 30-year mean annual percent bias (Table 2) and the spatial Pearson correlation45 between the 30-year mean annual evaporative demand rates of all land-surface grid cells (Table 3). Validation shown for ET0 is based on the short-grass vegetation parameterization within the ET0 formulation for brevity, but results for alfalfa vegetation are not qualitatively different from those shown for the short-grass vegetation.
Comparison of GCM ETP and ET0 to observational datasets. The CMIP6-MME estimated 30-year mean annual ETP and ET0 rates (mm/day) are compared to historical validation datasets (1951–1980) on a grid by grid basis. The resolution of all datasets are spatially adjusted to match the CRU dataset 0.5° × 0.5° resolution for validation, whereas the published datasets retain native GCM resolutions. ET0 from GCMs is compared to CRU ET0 (magenta) and ERA5-Land ET0 (green), whereas ETP estimated from GCMs is compared to ERA5-Land ETP (blue).
The percent bias of the 30-year global mean annual GCM estimates of ET0 range from −19.8 to −5.2% when compared to the CRU dataset and from −7.0 to 18.2% compared the ERA5-Land derived estimate (Table 2). ETP from the GCMs demonstrate a range of percent bias from −12.7 to 3.5% for the global mean when compared to the ERA5-Land ETP (Table 2). Visual inspection of the spread of points in Fig. 1. show the largest variations about the one-to-one line are in regions of with the highest historical evaporative demand.
Pearson correlations (R) of the 2-d spatial pattern of 30-year mean annual ET0 range from 0.89 to 0.95 when evaluating against the CRU dataset and 0.91 to 0.95 against the ERA5-Land estimated ET0 (Table 3). The long-term estimated ETP from the CMIP6-MME and ERA5-Land are even more strongly correlated in the historical period, with R values ranging from 0.94–0.98 (Table 3). The strength of these correlations suggests that, despite the biases in evaporative demand estimations, the GCM simulations all generally reproduce the observed spatial pattern of ETP and ET0 across the global land surface.
FAO-56 Penman-Monteith
The sign and magnitude of biases of 30-year mean annual ET0 reveal some noticeable regional patterns across GCMs. The northern hemisphere appears to have a consistent negative bias for all GCMs (Fig. 2), which is likely the source of the global mean negative ET0 bias for the historical period across all 16 GCMs (Table 2) given the majority (~68%) of the land-surface on the Earth is in the northern hemisphere.
Differences between GCM and CRU ET0. CMIP6-MME derived estimations of historical 30-year mean annual short-grass ET0 (mm/day) are compared to the CRU historical baseline of ET0 over the period (1951-1980) on a grid cell by grid cell basis over the land area. Differences here are in units of (mm/day).
Biases relative to the CRU ET0 are larger than those relative to the ERA5-Land derived estimates (Table 2) and the R of GCMs estimated ET0 is consistently lower when comparing to the CRU ET0 versus ERA5-Land derived estimates (Table 3). These differences in the relative agreement of the CMIP6-MME derived estimates are likely due to relatively minor differences in the methods applied to estimate the monthly time series of vapor pressure in CRU’s estimates and the methods used here for both the ERA-Land derived historical baselines and model simulations. Differences in the estimation of vapor pressure deficit between the two datasets arise from differing quantification of relative humidity and different accounting of mean monthly temperatures which are required for estimation of the saturation vapor pressure. When compared the ERA5-Land derived estimates of ET0 all GCMs have high spatial correlation (R > 0.91) (Table 3) and only two of the models produce large annual biases: ACCESS-ESM1-5 and MPI-ESM1-2-HR (both 18.2% bias during the historical simulation) (Table 2).
Priestley-Taylor
Scatter about the one-to-one line exists within all ensembled members of the CMIP6-MME compared to the observational datasets, but the magnitude of the global bias tends to be smaller within the ETP estimations from GCMs compared to the biases of ET0. The global multi-model ensemble mean percent bias is −2.3% for ETP from ERA5-Land, the smallest of the ensemble biases compared to the ensemble mean ET0 bias of −11.0% for the CRU dataset and 5.3% for the ERA5-Land ET0 (Table 2). The smaller range of evaporative demand rates within the ETP estimations compared to ET0 (Fig. 1) are consistent with the assumption of a uniform α coefficient of 1.26 within the ETP formulation (see Sec. 2.3 description of the coefficient). Others have shown that arid regions should generally have a larger magnitude α coefficient applied in the ETP formula41,42, which in this case could lead to a more effective representation of the advective component of evaporative demand within those regions. While it is beyond the scope of this dataset18 to calibrate spatially distributed α coefficients, one could further explore this avenue by performing a transformation of the ETP described in Section 2.3.
Historical ETP from all members of the CMIP6-MME demonstrate positive biases in humid parts of northern South America due simulated net radiation that is much larger in magnitude than in the ERA5-Land reanalysis (Fig. 3). However, this positive bias does not appear to consistently impact regions with high relative humidity climates, as only 8 out 16 GCMs demonstrate the same positive bias within humid regions of Central Africa. Bands of small-magnitude negative ETP bias in the highest northern latitudes are present for 11 out of 16 GCMs, consistent with known differences in the radiative balance of northern-hemisphere ice covered land-areas within the CMIP6-MME compared to observation datasets46.
Differences between GCM and ERA5-Land ETP. CMIP6-MME derived estimations of historical 30-year mean annual short-grass ETP (mm/day) are compared to the ERA5-Land historical baseline of ETP over the period (1951-1980) on a grid cell by grid cell basis over the land area. Differences here are in units of (mm/day).
Advection and Radiation Components of ET0
The variations in biases of ET0 estimations between GCMs are primarily driven by differences in the biases of the advective component of the evaporative demand. Nine of the 16 GCMs produce both (1) too much advection-driven ET0 (ET0,ADV) in regions where ERA5-Land has lower magnitude (<3 mm/day) ET0,ADV and (2) too little ET0,ADV in regions of higher magnitude ERA5-Land ET0,ADV (Fig. 4). Only one of the GCMs, MRI-ESM2-0, consistently produces more ET0,ADV than ERA5-Land across all regions (Fig. 4). A potential driver of this bias that exists across multiple GCMs is that the simulated VPD is too large (small) within the regions that demonstrate low (high) levels of advection within ERA5-Land. This hypothesis is supported by examination of the tendency for the nine GCMs that display this bimodal advection bias to underpredict total ET0 in more arid regions such as Northern Africa where VPD is known to be high47 (Fig. 2). The radiative component of ET0 appears to be well-captured by the historical simulations across all 16 GCMs (Fig. 4), which is consistent with the performance of the ETP estimations that are primarily driven by the net radiation (Fig. 1).
Comparison of advective and radiative components GCM ET0 to ERA5-Land. The advective and radiative components of short-grass ET0 for each GCM are compared to the estimated components of ET0 from ERA5-Land. Each point represents the 30-year mean monthly ET0 (mm/day) from 1951–1980 in the historical period for each grid cell across the global land-surface. The advection component (light green) of ET0 is primarily modulated by the wind speed and vapor pressure deficit, whereas the radiation component is dependent on the net radiation.
Code availability
The Python scripts used to calculate ETP, ET0, ET0 components, and VPD can be found within the repository alongside the data described herein. Accompanying the code is a small subset of GCM data that can be used to test run the script. Python scripts utilize a small subset of libraries in the Python3 base and the xarray (v2022.11.0) library to handle calculations of the gridded datasets. The python code used to generate the dataset18 described above can be found in the GitHub repository using the following link: https://github.com/nelsbjarke/PET.
References
Berghuijs, W. R., Larsen, J. R., Emmerik, T. H. Mvan & Woods, R. A. A Global Assessment of Runoff Sensitivity to Changes in Precipitation, Potential Evaporation, and Other Factors. Water Resources Research 53, 8475–8486 (2017).
Cook, B. I., Smerdon, J. E., Seager, R. & Coats, S. Global warming and 21st century drying. Clim Dyn 43, 2607–2627 (2014).
Cook, B. I. et al. Twenty‐First Century Drought Projections in the CMIP6 Forcing Scenarios. Earth’s Future 8, (2020).
Hobbins, M. T. et al. The Evaporative Demand Drought Index. Part I: Linking Drought Evolution to Variations in Evaporative Demand. Journal of Hydrometeorology 17, 1745–1761 (2016).
Milly, P. C. D. & Dunne, K. A. On the Hydrologic Adjustment of Climate-Model Projections: The Potential Pitfall of Potential Evapotranspiration. Earth Interactions 15, 1–14 (2011).
Mukherjee, S., Mishra, A. & Trenberth, K. E. Climate change and drought: a perspective on drought indices. Current Climate Change Reports 4, 145–163 (2018).
Scheff, J. & Frierson, D. M. Terrestrial aridity and its response to greenhouse warming across CMIP5 climate models. Journal of Climate 28, 5583–5600 (2015).
Scheff, J. & Frierson, D. M. W. Scaling potential evapotranspiration with greenhouse warming. Journal of Climate 27, 1539–1558 (2013).
Thornthwaite, C. W. An Approach toward a Rational Classification of Climate. Geographical Review 38, 55–94 (1948).
Hargreaves, G. H. & Samani, Z. A. Estimating potential evapotranspiration. Journal of the irrigation and Drainage Division 108, 225–230 (1982).
Milly, P. C. D. & Dunne, K. A. Colorado River flow dwindles as warming-driven loss of reflective snow energizes evaporation. Science https://doi.org/10.1126/science.aay9187 (2020).
Xiang, K., Li, Y., Horton, R. & Feng, H. Similarity and difference of potential evapotranspiration and reference crop evapotranspiration – a review. Agricultural Water Management 232, 106043 (2020).
Priestley, C. H. B. & Taylor, R. J. On the Assessment of Surface Heat Flux and Evaporation Using Large-Scale Parameters. Mon. Wea. Rev. 100, 81–92 (1972).
Allen, R. G., Jensen, M. E., Wright, J. L. & Burman, R. D. Operational Estimates of Reference Evapotranspiration. Agronomy Journal 81, 650–662 (1989).
Crop evapotranspiration: guidelines for computing crop water requirements. (Food and Agriculture Organization of the United Nations, 1998).
Allen, R. G. & Pruitt, W. O. FAO‐24 Reference Evapotranspiration Factors. Journal of Irrigation and Drainage Engineering 117, 758–773 (1991).
Anderson, D. B. Relative humidity or vapor pressure deficit. Ecology 17, 277–282 (1936).
Bjarke, N., Barsugli, J. & Livneh, B. CMIP6 derived ensemble of global vapor pressure deficit, potential evapotranspiration, and reference evapotranspiration. Zenodo https://doi.org/10.5281/ZENODO.7789759 (2023).
O’Neill, B. C. et al. The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci. Model Dev. 9, 3461–3482 (2016).
Tebaldi, C. et al. Climate model projections from the Scenario Model Intercomparison Project (ScenarioMIP) of CMIP6. Earth Syst. Dynam. 12, 253–293 (2021).
Harris, I., Osborn, T. J., Jones, P. & Lister, D. Version 4 of the CRU TS monthly high-resolution gridded multivariate climate dataset. Sci Data 7, 109 (2020).
Copernicus Climate Change Service. ERA5-Land monthly averaged data from 2001 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS). https://doi.org/10.24381/CDS.68D2BB30 (2019).
Marvel, K., Schmidt, G. A., Miller, R. L. & Nazarenko, L. S. Implications for climate sensitivity from the response to individual forcings. Nature Climate Change 6, 386–389 (2016).
Sherwood, S. C. et al. An Assessment of Earth’s Climate Sensitivity Using Multiple Lines of Evidence. Reviews of Geophysics 58, e2019RG000678 (2020).
Allen, R. G., Pereira, L. S., Raes, D. & Smith, M. Crop evapotranspiration-Guidelines for computing crop water requirements-FAO Irrigation and drainage paper 56. Fao, Rome 300, D05109 (1998).
Parisa-Sadat, A., Bozorg, H. O. & Mariño Miguel, A. Risk Analysis of Water Demand for Agricultural Crops under Climate Change. Journal of Hydrologic Engineering 20, 04014060 (2015).
Beven, K. A sensitivity analysis of the Penman-Monteith actual evapotranspiration estimates. Journal of Hydrology 44, 169–190 (1979).
Kay, A. L. & Davies, H. N. Calculating potential evaporation from climate model data: A source of uncertainty for hydrological climate change impacts. Journal of Hydrology 358, 221–239 (2008).
Kingston, D. G., Todd, M. C., Taylor, R. G., Thompson, J. R. & Arnell, N. W. Uncertainty in the estimation of potential evapotranspiration under climate change. Geophysical Research Letters 36 (2009).
McKenney, M. S. & Rosenberg, N. J. Sensitivity of some potential evapotranspiration estimation methods to climate change. Agricultural and Forest Meteorology 64, 81–110 (1993).
Tabari, H. & Hosseinzadeh Talaee, P. Sensitivity of evapotranspiration to climatic change in different climates. Global and Planetary Change 115, 16–23 (2014).
López-Urrea, R., Martín de Santa Olalla, F., Fabeiro, C. & Moratalla, A. Testing evapotranspiration equations using lysimeter observations in a semiarid climate. Agricultural Water Management 85, 15–26 (2006).
Ortega-Farias, S., Olioso, A., Antonioletti, R. & Brisson, N. Evaluation of the Penman-Monteith model for estimating soybean evapotranspiration. Irrigation Science 23, 1–9 (2004).
Sinclair, T. R. et al. Limited-transpiration response to high vapor pressure deficit in crop species. Plant Science 260, 109–118 (2017).
Will, R. E., Wilson, S. M., Zou, C. B. & Hennessey, T. C. Increased vapor pressure deficit due to higher temperature leads to greater transpiration and faster mortality during drought for tree seedlings common to the forest–grassland ecotone. New Phytologist 200, 366–374 (2013).
Grossiord, C. et al. Plant responses to rising vapor pressure deficit. New Phytologist 226, 1550–1566 (2020).
López, J., Way, D. A. & Sadok, W. Systemic effects of rising atmospheric vapor pressure deficit on plant physiology and productivity. Global Change Biology 27, 1704–1720 (2021).
Yuan, W. et al. Increased atmospheric vapor pressure deficit reduces global vegetation growth. Science advances 5, eaax1396 (2019).
Chow, V. T., Maidment, D. & Mays, L. Applied Hydrology. (McGraw-Hill Book Company, 1988).
Irmak, S., Allen, R. G. & Whitty, E. B. Daily Grass and Alfalfa-Reference Evapotranspiration Estimates and Alfalfa-to-Grass Evapotranspiration Ratios in Florida. Journal of Irrigation and Drainage Engineering 129, 360–370 (2003).
Aschonitis, V. G. et al. High-resolution global grids of revised Priestley–Taylor and Hargreaves–Samani coefficients for assessing ASCE-standardized reference crop evapotranspiration and solar radiation. Earth Syst. Sci. Data 9, 615–638 (2017).
Cristea, N. C., Kampf, S. K. & Burges, S. J. Revised Coefficients for Priestley-Taylor and Makkink-Hansen Equations for Estimating Daily Reference Evapotranspiration. J. Hydrol. Eng. 18, 1289–1300 (2013).
Singer, M. B. et al. Hourly potential evapotranspiration at 0.1° resolution for the global land surface from 1981-present. Sci Data 8, 224 (2021).
Jones, P. W. First- and Second-Order Conservative Remapping Schemes for Grids in Spherical Coordinates. Mon. Wea. Rev. 127, 2204–2210 (1999).
Holický, M. Introduction to Probability and Statistics for Engineers. https://doi.org/10.1007/978-3-642-38300-7 (Springer Berlin Heidelberg, 2013).
Abe, M., Nozawa, T., Ogura, T. & Takata, K. Effect of retreating sea ice on Arctic cloud cover in simulated recent global warming. Atmospheric Chemistry and Physics 16, 14343–14356 (2016).
Yang, Z., Zhang, Q., Hao, X. & Yue, P. Changes in Evapotranspiration Over Global Semiarid Regions 1984–2013. J. Geophys. Res. Atmos. 124, 2946–2963 (2019).
Dix, M. et al. CSIRO-ARCCSS ACCESS-CM2 model output prepared for CMIP6 CMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.2281 (2019).
Ziehn, T. et al. CSIRO ACCESS-ESM1.5 model output prepared for CMIP6 CMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.2288 (2019).
Danabasoglu, G. NCAR CESM2 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.7627 (2019).
Lovato, T. & Peano, D. CMCC CMCC-CM2-SR5 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.3825 (2020).
Lovato, T., Peano, D. & Butenschön, M. CMCC CMCC-ESM2 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.13195 (2021).
EC-Earth Consortium (EC-Earth). EC-Earth-Consortium EC-Earth3 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.4700 (2019).
Krasting, J. P. et al. NOAA-GFDL GFDL-ESM4 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.8597 (2018).
Volodin, E. et al. INM INM-CM4-8 model output prepared for CMIP6 CMIP piControl. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.5080 (2019).
Volodin, E. et al. INM INM-CM5-0 model output prepared for CMIP6 CMIP piControl. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.5081 (2019).
Boucher, O. et al. IPSL IPSL-CM6A-LR model output prepared for CMIP6 CMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.1534 (2018).
Shiogama, H., Abe, M. & Tatebe, H. MIROC MIROC6 model output prepared for CMIP6 ScenarioMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.898 (2019).
von Storch, J.-S. et al. MPI-M MPIESM1.2-HR model output prepared for CMIP6 HighResMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.762 (2017).
Wieners, K.-H. et al. MPI-M MPI-ESM1.2-LR model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.6595 (2019).
Yukimoto, S. et al. MRI MRI-ESM2.0 model output prepared for CMIP6 CMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.621 (2019).
Tang, Y. et al. MOHC UKESM1.0-LL model output prepared for CMIP6 CMIP historical. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.6113 (2019).
Roberts, M. MOHC HadGEM3-GC31-LL model output prepared for CMIP6 HighResMIP. Earth System Grid Federation https://doi.org/10.22033/ESGF/CMIP6.1901 (2017).
Acknowledgements
We acknowledge the financial support provided by from the NOAA grant titled “Physically-based evaluation of CMIP6 hydrologic projections for the western United States (NA19OAR4310284) and the NOAA grant titled: “Western Water Assessment: Building resilience to compound hazards in the Intermountain West (NA21OAR4310309) that made this research possible.
Author information
Authors and Affiliations
Contributions
N.B., B.L., and J.B. wrote, edited, and reviewed the manuscript; N.B. and J.B. developed scripts to perform calculations and generation of dataset; N.B. was responsible for data access, storage, and management; J.B. and N.B. performed quality control on dataset and methodologies applied; B.L., N.B., and J.B. all coordinated the publication of the dataset in the repository and generated ideas for the description and validation of the dataset.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Bjarke, N., Barsugli, J. & Livneh, B. Ensemble of CMIP6 derived reference and potential evapotranspiration with radiative and advective components. Sci Data 10, 417 (2023). https://doi.org/10.1038/s41597-023-02290-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41597-023-02290-0
- Springer Nature Limited