Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Scientific Data volume 12 , Article number: 1879 ( 2025 ) Cite this article
Potential evapotranspiration plays a crucial role in hydrological analysis. Therefore, reliable projection of potential evapotranspiration and/or reference evapotranspiration (ETo) at the local scale under various climate change scenario is essential. To achieve this, climate datasets from various General Circulation Models must be downscaled and bias corrected to enhance their accuracy. In this study, we developed daily bias-corrected datasets of ETo at 0.25° spatial resolution. We used outputs from the Coupled Model Intercomparison Project Phase 6 (CMIP6) for the historical and four Shared Socioeconomic Pathways. The datasets include temperature, solar radiation, wind speed, and relative humidity from 12 different models at various spatial scales for South Asia. The bias correction was performed using the Quantile-Mapping approach. The resulting bias-corrected climate variables (dataset available in repository) were subsequently used to estimate historical and future ETo over South Asia. A significant difference was observed between the original and bias-corrected datasets when evaluated against the ERA-5 reanalysis observation data. Penman-Monteith method was utilized for ETo estimation to reduce the uncertainties of temperature-based methods.
Evapotranspiration is a crucial component of the hydrological cycle, with its causal powers in the area of agronomy, hydrology, ecology and biogeography 1 , 2 , 3 , 4 , 5 , 6 , 7 . Evapotranspiration is influenced by climatic factors, namely: solar radiation, relative humidity, wind speed and temperature; along with vegetative and soil characteristics 8 . Actual evapotranspiration (AET) is a water balance variable and is reflective of energy and water regime 9 . Potential evapotranspiration (PET) serves as a climatic control on AET 10 , 11 . Though PET may be estimated using energy balance approach, its estimation using aerodynamic approach shows that PET is also reflective of water regime: through surface wetness which influences vapor pressure deficit 12 . PET may have negligible influence on AET in water-limited regions, but limits AET in energy-limited regions 13 , 14 , 15 . Assuming precipitation as the only source of water, we see that there has been a negligible increase in the area of arid regions (i.e., water-limited regions) over South Asia under various climatic scenarios, compared to the historical period 16 (see Table S1 ). Intensive irrigation in South Asia’s arid zones 17 can artificially create energy-limited conditions, thereby masking the underlying water-limited nature of these regions. Increasing PET during drought conditions over energy-limited regions shall increase AET, thereby depleting water resources and causing flash drought 18 . This shows the importance of reliable PET estimation over future climatic scenarios, for effective simulation of AET and thereby discharge over South Asia. In spite of the increasing temperature, recent studies have shown a declining trend of PET over South Asia for the historical period. Solar radiation, wind speed and vapor pressure were observed as the major climatic drivers to the decline of PET 19 , 20 , 21 , 22 , 23 . The availability of bias-corrected temperature dataset for future scenarios over South Asia 24 influences the use of temperature-based method for PET estimation. These methods fail to attribute the contribution of the other climatic drivers to the trend of PET 14 and may lead to overestimation of drought frequency under a warming climate 25 , 26 , 27 . Study by Fisher et al . 28 shows that PET over tropics is overestimated using Pristley Taylor in comparison to Penman Monteith method, as the former does not include vapor pressure deficits. Increasing humidity over the tropics decreases the vapor pressure deficit, which reduces the atmospheric demand. Thus, we use the FAO-56 Penman Monteith (PM) method, recommended by the Food and Agriculture Organisation of the United Nations, for estimation of PET 29 . FAO-56 defines PET as AET of a hypothetical reference crop(ETo), which is actively growing and adequately watered with an assumed crop height of 0.12 m, a fixed surface resistance of 70 sm −1 and an albedo of 0.23 29 , 30 . Although prior efforts have estimated global PM ETo monthly, they either come at 0.5 (downscaled using bilinear interpolation) 31 or at original GCM resolution 32 . Bias corrected monthly datasets of future projection of ETo has been prepared for South Asia 33 , 34 , 35 , but is not available in open source platforms and were used for the research objective of their respective study only. Thus, a bias-corrected daily high-resolution ETo dataset remains unavailable. This limitation is particularly evident in South Asia, a region that is both densely populated and highly vulnerable to climate-induced water stress 36 , 37 , 38 . While Global Climate Models (GCMs) are invaluable for understanding future climate trajectories, GCM outputs are typically available at coarse spatial resolutions (often >100 km), which are inadequate for regional- and local-scale applications. To address this limitation, downscaling and bias correction of GCM outputs are necessary to refine their utility for impact modelling. This study addresses the above gaps by developing a bias-corrected, high-resolution (0.25°) daily PM-ETo dataset over South Asia using the FAO-56 Penman–Monteith method. We use simulations from 12 CMIP6 39 GCMs covering the historical period (1960–2014) and future SSPs (Shared Socioeconomic Pathways) 40 scenarios (2015–2100), including SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5. Bias correction and spatial downscaling of each required sub-variable (temperature, wind speed, solar radiation, and relative humidity) were conducted using the widely adopted Quantile Mapping approach 24 , 41 , 42 , with ERA5 reanalysis data serving as the observational reference. Since raw GCM outputs are available at coarse resolutions, this study improves spatial accuracy through bias correction and downscaling. The raw daily datasets for temperature, wind speed, solar radiation, and relative humidity from various CMIP6 models, were first re-gridded to a common 1° spatial resolution using bilinear interpolation 24 . To facilitate bias-correction, the ERA-5 reanalysis observational data – originally available at a 0.25° spatial resolution – were upscaled to match the 1° resolution of model datasets. This allowed for consistent comparison and evaluation of the differences in the probability distribution functions between the observed and modelled datasets. A multi-model ensemble was prepared for each climatic scenario for each sub-variable, which was then used to prepare the ETo dataset. In other words, bias-corrected ETo is available for the five climatic scenarios over South Asia, which is prepared using a multi-model median ensemble of each sub-variable for each climatic scenario. In addition to ETo, the individual downscaled and bias-corrected climate variables are also provided as standalone datasets, enhancing their utility for a wide range of climate-related applications. For example, the wind speed dataset can support high-resolution wind energy assessments in regions where ground-based observations are limited 43 , 44 . Similarly, solar radiation datasets can aid in solar power generation modeling, desalination technology design, and vegetation growth analysis 45 , 46 , 47 , 48 , 49 . Relative humidity datasets, which influence plant physiology, hydrology, and even public health, also benefit from correction and refinement to support climate resilience planning 50 , 51 , 52 .
The bias correction framework was implemented using open-source tools such as Climate Data Operators (CDO) 53 and R 54 , ensuring transparency and reproducibility. The generated datasets are provided in NetCDF format and are publicly available for use in further research. Evaluation against ERA5 observations confirmed significant improvements in the accuracy of the bias-corrected variables and derived ETo over the calibration period (1960–2014). While rising mean temperatures are projected under all SSPs, the corresponding ETo trends show more nuanced responses due to varying contributions from the four sub-variables, highlighting the importance of physically based ETo estimation under changing climate scenarios. Overall, a projected increase in ETo is observed with respect to historical data over South Asia under various climate scenarios, except for SSP3-7.0, which shows a decrease over most sub-continent areas.
In summary, this study delivers the first regional-scale, bias-corrected, daily ETo dataset for South Asia based on the Penman–Monteith method using CMIP6 projections. These datasets offer a valuable foundation for improving regional water resource assessments, agricultural planning, and climate impact studies, particularly under future warming scenarios.
The spatial downscaling approach is used to enhance the spatial resolution of CMIP6 GCMs data, making it suitable for local-scale climate impact assessments. The bias correction method of quantile mapping first adjusts the biases and further downscale the data, ensuring it accurately represents observed climate patterns across South Asia. The South Asian countries include India, Bangladesh, Nepal, Bhutan, Pakistan, Afghanistan and Sri Lanka (Fig. 1 ). The raw data comprises outputs from 12 CMIP6 models: average temperature (tas), relative humidity (hurs), wind speed (sfcWind) and solar radiation (rsds). The variables were selected on a daily timescale for the historical period (1960–2014), and four main future SSP scenarios: SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, spanning the years 2015 to 2100. Once these variables are bias-corrected, they are used to estimate historical and future reference evapotranspiration (ETo) at 0.25° × 0.25° resolution using the Penman-Monteith (PM) method. We applied the commonly used statistical downscaling method based on detrended Quantile Mapping approach for bias-correction and downscaling 55 , 56 .
South Asia domain of the dataset, which includes 7 countries: India, Sri Lanka, Nepal, Bhutan, Bangladesh, Pakistan and Afghanistan.
The dataset generated to estimate ETo in this study includes four climate variables: temperature (tas), relative humidity (hurs), wind speed (sfcWind) and solar radiation (rsds) of 12 CMIP6 GCM over South Asia ( https://esgf-node.ipsl.upmc.fr/search/cmip6-ipsl/ ) 57 . The dataset has a common variant label of r1i1p1f1 and is available at various spatial resolutions, as shown in Table 1 . We obtained the ERA-5 reanalysis daily dataset of temperature, solar radiation, wind speed, relative humidity and the computed reference evapotranspiration (ETo) at 0.25° spatial resolution for the base period of 1960–2014 ( https://cds.climate.copernicus.eu ) 58 . The Digital Elevation Model (DEM) has been acquired from the Shuttle Radar Topography Mission (SRTM), NASA’s Earth data mission ( https://www.earthdata.nasa.gov/data/instruments/srtm ) 59 . It is available at a spatial resolution of 30 meters (1 arc-second), which is upscaled to the resolution of the climate dataset (0.25°). Carbon dioxide (CO 2 ) concentrations under both the historical (1960–2013) and future (2015–2100) scenarios on a monthly temporal scale and 1° spatial resolution for South Asia are obtained from the data repository ( https://doi.org/10.5281/zenodo.5021361 ) 60 . The CO 2 concentration data is further processed to downscale to 0.25° spatial resolution and incorporated into the modified PM-ETo computation. The availability of CO 2 data till 2013 for the historical period limits the availability of ETo historical data from 1960 to 2013 (rather than 2014).
The method applied to downscale the CMIP6 GCM outputs consists of three steps: Re-gridding and data pre-processing, bias correction (BC), and spatial disaggregation (SD). The main purpose of the first step is to re-grid the dataset from CMIP6 models and observation data at the same spatial resolution for the ease of intercomparison 61 , 62 and to detrend their climatic trends for further computational steps. For the consistency of GCMs and observation data, we interpolated the variables to an intermediate resolution of 1° × 1° using bilinear interpolation. We remove the trends from the dataset of each grid point in the GCM historical and future projections, only to be added back after bias correction. This step is crucial to preserve the daily climatic trends of temperature, wind speed, solar radiation, and relative humidity. Climatological averages were computed for each calendar day using a 9-year running mean of the daily time series. These averages are then used to calculate the anomalies, which are performed for both observation (ERA5) and CMIP6 data (historical and future scenarios).
The quantile-based mapping method is used for this study as it not only improves the first order but also the higher moments of variables. It attempts to derive a transfer function based on the cumulative distribution function (CDFs) of historical GCM outputs and the observational data, to map the model outputs onto a new distribution that aligns with observations. However, this method may change the variable trend, leading to the deterioration of the climate change signal 42 . Therefore, removing climatic trends is a crucial pre-processing step before applying bias-adjustment 41 , 42 . To address this, we developed an anomaly-based dataset, which will be utilized in the subsequent step.
As atmospheric variables from GCMs are known to exhibit systematic biases, they are corrected in the bias correction (BC) step to ensure reliable and accurate estimates at regional and local scales 24 . A univariate detrended quantile mapping method is used, which acts independently on trends and anomalies 55 , 56 , 62 . The bias correction is performed at each grid point and is applied iteratively to each day of the year. In this step, the correction of biases in GCM atmospheric variables is performed using ERA-5 reanalysis data as observation data. Candidate days are selected for each grid point on each day of the year using a moving window of ±15 days 63 . This means that for any given calendar date, the cumulative distribution function (CDF) is generated based on the values from that specific day, along with the 15 preceding and 15 succeeding days, resulting in a 31-day window per year. These 31 daily values are collected over a 55-year base period (1960–2014), providing a total of 31 × 55 = 1,705 values to construct individual CDFs for both the GCM historical and the reference (observational) data. For example, to generate the CDF for the 16 th day of the year, i.e., 16 January, we applied the moving window approach, taking the values from 16 January itself along with 1 to 15 January (15 days before) and 17 to 31 January (15 days after), covering 31 days. This process is repeated for each day of the year.
The computed averages (trends) and anomalies in the detrending step further require the adjustment of residuals by computation of the adjustment factor function (A-factor). The A-factor is the difference between the corresponding CDF values of historical data from the observation and is a function of Quantile (q), which is mapped by sampling q from values of 0.01 to 0.99 by steps of 0.01.
where, \({F}_{{Y}^{^{\prime} }}\) and \({F}_{{X}^{^{\prime} }}\) are the cumulative distribution function of reference data anomalies \({Y}^{^{\prime} }\) and historical data anomalies \({X}^{^{\prime} }\) respectively, and q is the Quantile which ranges from 0 to 1.
Similarly, the adjustment of the trend requires the computation of the offset factor (C-factor) from the averages of the model and observation.
where, \(\bar{Y}\) and \(\bar{X}\) are the computed averages from reference data and historical data, respectively.
The bias-adjusted daily GCM value is obtained by identifying the A-factor at the same Quantile as the model anomaly and then adding it to the anomaly at that Quantile. After the BC step, the trends are adjusted as the sum of the C-factor and the original trend, to be added back to the bias-adjusted anomalies of model simulations, for each grid point and day of the year.
where, \({{X}^{* }}^{^{\prime} }\) is the bias-adjusted anomaly and \({X}^{* }\) is the bias-adjusted time series for a given grid point on a given day of the year.
Since the quantile mapping method remains stable across entire simulations (both historical and future periods), the same technique is applied to bias-correct future projections (2015–2100). Lavoie et al . 62 describes a detailed step-by-step process for bias-adjusted process.
In the spatial downscaling (SD) step, the bias-corrected model data were interpolated to the higher resolution of 0.25° × 0.25°. First, we compute the daily climatology of observation data at the initial raw CMIP6 model resolution of 1° × 1°. Second, a “scale factor” is generated for each grid point on each day, based on the difference between bias-corrected data and daily observational climatology at the initial model resolution. Third, the interpolation of “scale factor” to the resolution of 0.25° × 0.25°. Finally, to obtain the statistically downscaled model data at the desired resolution, we add the daily observational climatology of a variable at its original resolution to the interpolated “scale factor” at the same resolution.
The bias-corrected downscaled dataset of climate variables developed at 0.25° spatial resolution to estimate evapotranspiration under historical and future scenarios. The daily ETo data for each climatic scenario is computed using the widely preferred FAO Penman-Monteith equation 29 .
where, \({ETo}\) is the daily PM (mm/day), \({R}_{n}\) is the net surface radiation (MJ m −2 day −1 ), \({T}_{{av}}\) is daily average air temperature (°C), \({u}_{2}\) is the wind speed at the 2 m height (m/s), \({e}_{s}\) is the saturation vapor pressure (kPa), \({e}_{a}\) is the actual vapor pressure (kPa), \(\Delta \) is the slope of the saturation vapor pressure versus temperature curve, and \(\gamma \) is the psychrometric constant (kPa °C −1 ). The impact of CO 2 concentration on evapotranspiration is considered in the present study by incorporating it in the surface resistance (r s ) component of the PM method. Recent studies suggest that increasing concentration of CO 2 under global warming shall increase surface resistance, thereby reducing ETo 64 , 65 , 66 . Since, \({r}_{s}\) is fixed at 70 s.m −1 , the value corresponds to a particular CO 2 concentration and hence \({r}_{s}\) value should be estimated at different atmospheric CO 2 levels 67 . Thus, the influence of CO 2 to ETo trends is added. The empirical function of CO 2 concentration is used in the computation 68 :
\({r}_{s}\) is the surface resistance in (s.m −1 ), and \(\left[{{CO}}_{2}\right]\) presents the atmospheric \({{CO}}_{2}\) concentration in (ppm).
The wind speed acquired at 10 m height is converted to a height of 2 m for the ease of computation using Eq. ( 9 ):
where, \({u}_{2}\) is the wind speed at 2 m above the ground surface (ms −1 ), \({u}_{z}\) is measured wind speed at z m above the ground surface (ms −1 ), and z is the height of measurement above the ground surface (m).
Net radiation ( \({R}_{n}\) ) was computed using the following formulas:
where, \({R}_{{ns}}\) =net solar or short-wave radiation (MJ m −2 day −1 ), \(\alpha \) = albedo (0.23) for grass reference crop, \({R}_{s}\) = incoming solar radiation (MJ m −2 day −1 ).
where, \({R}_{{nl}}\) is the net outgoing longwave radiation (MJ m −2 day −1 ), \(\sigma \) is the Stefan Boltzmann constant (4.903 × 10-9 MJK −4 m −2 day −1 ), \({T}_{{av},K}\) is mean air temperature in Kelvin (K), \({e}_{a}\) is the actual vapor pressure (kPa), \({R}_{s}\) is the solar or shortwave radiation (MJ m −2 day −1 ), \({R}_{{so}}\) is the clear sky radiation (MJ m −2 day −1 ).
As we have the relative humidity (RH) variable for ETo estimation, it is involved in the computation of actual vapor pressure ( \({e}_{a}\) ) using the following formulas:
where, \({e}_{s}\) is the saturation vapor pressure (kPa).
A potential limitation of empirical quantile mapping method is the assumption of stationarity which may alter climatic signals during the bias correction process 69 . This limitation is addressed in our study by employing trend-preserving strategies in the quantile mapping method. Secondly, univariate quantile mapping is used in this study, i.e., they operate on a single variable and location at a time to retain the temporal structure of the climatic model 70 , 71 . As the variables are corrected one at a time, they may become mutually inconsistent and generate physically implausible situations 72 , 73 . Checks have been made in this study to ensure that the intervariable dependencies are conserved post bias correction (see technical validation section). Despite the limitations, quantile mapping continues to remain the most widely used bias correction method due to its flexibility and non-parametric adaptability across climatic variables.
Recent studies show that PET(ETo) is overestimated when calculated using Penman equations (and its variants) 12 , 74 , 75 . The assumption of well-watered surface is not valid over most land surfaces, thus the atmospheric conditions corresponding to hypothetical saturated surface is unknown. Zhou et al . 76 uses energy based equation to show the overestimation of PET, when calculated using Penman equation for future climate scenarios. However, land surface and crop models use Penman Monteith equation for estimation of PET, and the bias corrected dataset of the four climatic drivers shared in the repository, shall be useful for such modelling frameworks in studies on future projections.
The median ensemble of daily bias-corrected CMIP6 ETo dataset at 0.25° over South Asia for emission scenarios: historical, SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, is available on Zenedo ( https://doi.org/10.5281/zenodo.15670655 ) 77 . The ensemble is prepared using the climate models listed in Table 1 . The median multi-model ensemble of daily bias-corrected datasets of the sub-variables, namely temperature ( tas ), solar radiation( rsds ), wind speed( sfcWind ) and relative humidity( hurs ) are also shared along with ETo (eto). The units are K, W/m 2 , m/sec, % and mm, respectively. The dataset is shared as NETCDF files through a CC-BY 4.0 license. The daily bias-corrected datasets are cropped for the 7 South Asian countries: India, Bangladesh, Sri Lanka, Pakistan, Afghanistan, Nepal and Bhutan. There are 37.zip folders in the repository.
35 of them is named as { country }_{ variable }.zip. Each of these folders contains daily bias corrected data for five climatic scenarios of a variable, for a South Asian country, and is named as { variable }_ day _{ country } _{ scenario } _ bcds.nc
SouthAsia_monthly_eto.zip: This folder contains monthly ETo data over the entire South Asia for the five climatic scenarios and are named as eto_month_SouthAsia _{scenario} _bcds.nc . Monthly data is shared for easy application of bias-corrected ETo for drought applications.
era5_eto.zip: Climatic sub-variables from ERA5 have been used for bias correction. This folder contains the ETo estimated using the ERA5 climatic sub-variables for the reference period. Daily and monthly ETo data for the reference period is shared which are named as era5_day_eto_{country}_1960_2013.n c and era5_month_eto_southAsia_1960_2013.n c
Physically unrealistic values may be generated after bias adjustments 78 , 79 , and thus, health checks were performed to confirm if the bias-corrected datasets produced any physically unrealistic values. Bounding values from Thrasher et al . 80 were used to evaluate each sub-variable, and no discrepancies were found in the ensemble of the bias-adjusted simulations of the climatic sub-variables. However, data points exceeding the upper bounds of relative humidity and lower bounds of wind speed were noted. Overall, less than 0.002% of the outlying values were noted in the entire archive. Outliers were seen only in 1.9% of the total grid points for relative humidity, which were seen only in Nepal. Outliers for wind speed lie in the extreme north of India and Pakistan, amounting to 1.2% of the grid points. These outlier data points were set as NA during the estimation of ETo, which was later gap-filled using the distance-weighted average method.
The datasets from 12 CMIP6 GCMs were bias-corrected and downscaled for the historical (1960–2014) and future (2015–2100) periods over South Asia. The original (raw) and their respective bias-corrected projections from all models were combined to form a multi-model ensemble median. Subsequently, the mean annual bias in the raw and bias-corrected historical datasets was estimated against the ERA5 observational data, as shown in Fig. 2 . A high warm bias in raw data was found in the Upper Himalayas and extreme north regions of India, Pakistan and Afghanistan, whereas a high cold bias was observed in the Lower Himalayan region, north-eastern states of India and some northern parts of Pakistan and Afghanistan (Fig. 2a ). Rest of South Asia was covered with slight warm bias. A high positive bias in the solar radiation variable was observed in Bhutan and the north-eastern state of India, and a positive bias in rest of the sub-continent (Fig. 2c ); all substantially reduced after bias-adjustment (Fig. 2d ). The bias found in relative humidity is highly moist in northern region of India and highly dry in central and western states of India and central Pakistan (Fig. 2e ). Mild moist bias is observed in south and eastern India. Positive bias in wind speed is seen over the Himalayan region, the coastline of India and Sri Lanka, while high negative bias seen over western Afghanistan (Fig. 2g ). The bias-adjustment by quantile mapping approach successfully removed the biases in all the variables over the South Asian region (Fig. 2b,d,f,h ). Positive biases in the derived ETo were observed across most of the South Asian region, except for some negative biases over western Afghanistan and Pakistan, as well as in certain north-eastern and southern areas and the Ladakh state of India (Fig. 3a ). Notably, an extreme positive bias is observed in the north of Afghanistan. These biases in ETo, as computed from the original climate variables (Fig. 3a ), are substantially reduced following bias adjustment (Fig. 3b ).
Multi-model ensemble median bias of temperature, solar radiation, relative humidity and wind speed in 12 CMIP6 GCMs. ( a ) Bias (%) in annual mean daily temperature for the historical period (1960-2014), ( b ) Bias (%) in annual mean daily temperature after the bias correction, ( c,d ) Bias (%) in annual mean daily solar radiation before and after bias correction, ( e,f ) and ( g,h ) same as ( c,d ) but for relative humidity (%) and wind speed (ms −1 ) respectively.
Multi-model ensemble median bias (%) in annual mean reference evapotranspiration derived from four climate variables before ( a ) and after ( b ) bias correction for historical period (1960–2013).
Availability of freshwater is not only dependent on precipitation or evaporative demand, but also on their interaction at the seasonal scale 81 . Thus, it is important that the climatology of the reference dataset is properly constructed 82 . The monthly climatology of each sub-variable for the selected 12 CMIP6 models over the calibration period (1960–2014) is illustrated in Fig. S1 . We see the presence of systematic biases within the raw data, which collapses to zero post-bias correction. The climatology of ETo, derived from the ensemble of all GCMs (before and after bias-correction), and ERA-5 data, is presented in Fig. 4 . The dashed line representing the ETo climatology from the bias-corrected data closely aligns with the ERA5 observations and shows a significant departure from the climatology estimated using the original GCM data. While the overall climatological trend or pattern remains consistent across raw, bias-corrected, and observed data, the monthly values are shifted post-correction—highlighting the importance of bias correction for climate variables.
ETo monthly climatology derived from multi-model ensemble median of each climate variable obtained from 12 GCMs over the historical period (1960-2013): before (blue line) and after (dashed red line) bias-correction. The ETo climatology (1960-2013) derived from ERA-5 climate data represented in black line.
Given the observed improvements in climatology following bias adjustment, it is equally essential to validate the seasonal performance of the bias-corrected data at the spatial scale across South Asia. Accordingly, the changes in raw and bias-corrected data compared to observations for each climate variable on a seasonal basis are presented in Fig. 5 . The four seasons considered include winter (DJF), summer (MAM), monsoon (JJAS) and post-monsoon (ON). The seasons are selected based on the climate of India and may not match those of other countries. In the original CMIP data, temperature exhibits a warm bias during DJF, MAM and ON seasons over the upper Himalayan region of India and the northern and central parts of Afghanistan. In contrast, the Middle Himalayas and north-eastern parts of India, the remainder of Afghanistan, and some parts of Pakistan display a significant cold bias. The JJAS season exhibits a similar spatial pattern, except for parts of Pakistan and other areas in India, which show slight warm biases. For solar radiation, the JJAS season is characterized by strong positive biases across much of the region, while during the other seasons, this bias is mainly confined to Nepal, Bhutan, and parts of north-eastern India. Slight positive biases are observed in DJF, MAM and ON seasons over most regions of the subcontinent, while slight negative biases are evident during MAM over Jammu & Kashmir, and Ladakh states of India and northern Afghanistan. For relative humidity, DJF and MAM seasons show positive biases over Jammu & Kashmir and Ladakh states of India and the far northern regions of Afghanistan, whereas negative biases dominate western and central India and most of Pakistan. During JJAS season, positive biases occur over Sri Lanka and parts of Ladakh and Tamil Nadu states of India, while Afghanistan, Pakistan, and western and northwestern India experience negative biases. In the ON season, Ladakh shows a positive bias, whereas Pakistan and western to northwestern India display negative biases. For wind speed, all seasons exhibit positive biases over the upper Himalayas of the Indian region, parts of Afghanistan, and Sri Lanka. A notable exception is the JJAS season, where a negative bias is observed over the extreme western part of Afghanistan. Following bias correction, these seasonal biases are substantially reduced across all variables. Consequently, the ETo estimated from the bias-adjusted climate data shows a marked reduction in bias across all seasons, as illustrated in Fig. 6 . The original data displays widespread positive ETo biases in all seasons—particularly extreme during JJAS season across northern and central Afghanistan and Pakistan, and in western, central, southwestern, and northern India (excluding Ladakh). Negative ETo biases for all seasons are seen across western Pakistan, Afghanistan, central north-eastern India, and Ladakh state. These biases are effectively mitigated in the seasonal ETo estimates derived from the bias-corrected ensemble median of all GCM models.
Multimodel ensemble median bias of temperature(%), solar radiation(%), relative humidity(%) and wind speed(ms −1 ) in raw (i) and bias-corrected (ii) data; for ( a ) winter (DJF), ( b ) summer (MAM), ( c ) monsoon (JJAS) and ( d ) post-monsoon (ON) seasons in historical period of 1960-2014.
Seasonal bias (%) of multi-model ensemble median of ETo estimated from raw and bias-corrected climate data for ( a,b ) winter, DJF; ( c,d ) summer, MAM; ( e,f ) monsoon, JJAS and ( g,h ) post-monsoon, ON seasons in historical period of 1960-2014.
We further estimated and compared the multi-model ensemble median projected changes of mean annual values of variables, using original(raw) and bias-corrected data of all far-future scenarios (2066–2100) against the historical for 1960–2014. Overall, increased projected temperatures were observed in all the SSP scenarios against historical, with the highest increase in Afghanistan, Pakistan and northern India covering the states of J&K and Ladakh (Fig. 7 ). The magnitude of projected increment is highest in SSP5-8.5 and lowest in SSP1-2.6. For solar radiation, all the scenarios except SSP1-2.6 showed projected decrease (Fig. 8 ). Higher magnitudes of projected decrease of solar radiation during SSP3-7.0 is observed over the Gangetic plains. Projections of relative humidity show a similar pattern on spatial scale for all the scenarios, with bipolar changes observed over South Asia (Fig. 9 ). A decrease of relative humidity is seen over Afghanistan and north India, while an increase is observed over the rest of South Asia. The highest magnitude of projected increase and decrease was found during SSP3-7.0 and SSP5-8.5, respectively. Wind speed shows a projected increase over the Thar desert (for all scenarios) and eastern coastline of India (for SSP5-8.5), as shown in Fig. 10 . A projected decrease is seen mostly at the border of Pakistan and Afghanistan, along with Nepal, Bhutan, Sri Lanka, Northern and North-eastern India. The projected change in ETo is seen to increase for all scenarios except SSP3-7.0, where a decrease is observed over the Indian mainland and Bangladesh (Fig. 11 ). However, for all scenarios, a higher magnitude of the projected increase is seen over Afghanistan, and the northern regions of Pakistan and India. This validates the preserved trends in the bias-adjusted simulations and shows that the projected changes in the multi-model median of ETo and the four sub-variables remain conserved post-bias correction.
Multi-model ensemble median projected change in annual mean temperature (%) for the far-future (2066-2100) with respect to the historical period (1960-2014) in original simulations ( a,c,e & g ) and bias-adjusted simulations ( b,d,f & h ).
Same as Fig. 7 , but for solar radiation.
Same as Fig. 7 , but for relative humidity.
Same as Fig. 7 , but for wind speed.
Projected changes in seasonal reference evapotranspiration are also preserved post-bias adjustment (Figs. S2 , 3 , 4 , 5 ). Projected increase in ETo observed during the DJF season across central and north Afghanistan, north Pakistan, Jammu & Kashmir, Ladakh states of India and the Upper Himalayan part of India and Nepal in all the SSP scenarios, while it shows neutral or slight increase in the remaining parts of South Asia. The most pronounced projected increase of ETo is found in SSP5-8.5. During the JJAS season, a slight projected increase in ETo is seen across the region under SSP1-2.6. Under SSP2-4.5 and SSP5-8.5, projection increases over Afghanistan, Pakistan, Jammu & Kashmir, Ladakh, and the north-eastern states of India, while it decreases over Sri Lanka and the western parts of India. SSP3-7.0, in contrast, shows a projected increase of ETo in Afghanistan, Pakistan, Jammu & Kashmir, and Ladakh states of India, with neutral or decreased ETo in the rest of the region. During MAM season, ETo is projected to increase over central and northern Afghanistan, north Pakistan, and J&K and Ladakh states of India, with the magnitude of increase following the order: SSP1-2.6 < SSP2-4.5 < SSP3-7.0 < SSP5-8.5. Projection in remaining parts of South Asia shows neutral or slightly increased ETo across all SSPs, except in SSP3-7.0, which shows a projected decrease over Bangladesh, West Bengal, Bihar states, and parts of north-eastern India. From SSP1-2.6 to SSP5-8.5, a low to extreme increase in projected ETo is observed over J&K and Ladakh states of India, while the rest of South Asia experiences mostly neutral or slightly increased projections. SSP3-7.0 again stands out by showing neutral or decreased projections in ETo over several areas across the region. Country specific multi-model projections for the four scenarios along with the historical mean can be seen in supplemental Tables S2 – S6 .
This study uses a univariate quantile mapping approach, where each climatic sub-variable is adjusted independently, without considering the influence of one variable on another. The interaction of the four climatic sub-variables may influence ETo trend 83 , and thus it is important to verify that their inter-variable dependence structure is not impacted post-bias-adjustment. The correlation matrices of the four variables for each scenario, before and after bias correction are shown in Fig. 12 . It shows that the univariate bias correction did not affect variable interdependency. Thus, the sub-variables retained their mutual dependencies, even though they were not corrected jointly. This shows effective bias-correction of the climatic sub-variables, before being used for estimation of ETo.
Correlation matrices of mean annual values averaged over South Asia for the four sub-variables (tas: temperature, rsds: solar radiation; sfcWind: wind speed; hurs: relative humidity) before and after bias correction for the historical ( a,b ), SSP1-2.6 ( c,d ), SSP2-4.5 ( e,f ), SSP3-7.0 ( g,h ), SSP5-8.5 ( i,j ).
The dataset for the daily bias-corrected CMIP6 climatic drivers ( tas, hurs, sfcWind and rsds ) and ETo over South Asia for the future climatic scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5) is available on Zenedo 77 using the following link: https://doi.org/10.5281/zenodo.15670655 . The repository contains the bias corrected data cropped for each country of South Asia (India, Bangladesh, Sri Lanka, Pakistan, Afghanistan, Nepal and Bhutan) as NETCDF files. The dataset also contains the ERA5 dataset used as reference for bias correction. The dataset is freely available for reuse or modification with attribution (CC-BY 4.0).
Code used in this study is accessible on GitHub ( https://github.com/mrani307/BiasCorrection_CMIP6_PET ).
Miao, Q. et al . Modeling water use, transpiration and soil evaporation of spring wheat-maize and spring wheat-sunflower relay intercropping using the dual crop coefficient approach. Agric Water Manag 165 , 211–229 (2016).
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. Earths Future 8 (2020).
Thornthwaite, C. W. An Approach toward a Rational Classification of Climate . Source: Geographical Review vol. 38 (1948).
Scheff, J. & Frierson, D. M. W. Terrestrial aridity and its response to greenhouse warming across CMIP5 climate models. J Clim 28 , 5583–5600 (2015).
Hobbins, M. et al . A global long-term daily reanalysis of reference evapotranspiration for drought and food-security monitoring. Sci Data 10 (2023).
Medina, A. I., Martí, D. A. & Bidau, C. J. Subterranean rodents of the genus Ctenomys (Caviomorpha, Ctenomyidae) follow the converse to Bergmann’s rule. J Biogeogr 34 , 1439–1454 (2007).
Monteith, J. L. Evaporation and Surface Temperature . Quart. J. R. Met. SOC vol. 107 (1981).
Budyko, M. I. Climate and Life . (Hydrometeoizdat, Leningrad, 1971).
Lhomme, J.-P. Towards a rational definition of potential evaporation. Hydrol Earth Syst Sci 1 , 257–264 (1997).
Tu, Z., Yang, Y. & Roderick, M. L. Testing a maximum evaporation theory over saturated land: implications for potential evaporation estimation. Hydrol Earth Syst Sci 26 , 1745–1754 (2022).
Zhou, S. & Yu, B. Physical basis of the potential evapotranspiration and its estimation over land. J Hydrol (Amst) 641 (2024).
McVicar, T. R. et al . Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Implications for evaporation. Journal of Hydrology 416–417 , 182–205, https://doi.org/10.1016/j.jhydrol.2011.10.024 (2012).
Roderick, M. L., Hobbins, M. T. & Farquhar, G. D. Pan evaporation trends and the terrestrial water balance. II. Energy balance and interpretation. Geogr Compass 3 , 761–780 (2009).
Budyko, M. I. Climate and Life . (Academic Press, New York, 1974).
Beck, H. E. et al . High-resolution (1 km) Köppen-Geiger maps for 1901–2099 based on constrained CMIP6 projections. Sci Data 10 (2023).
Zhou, Y. et al . Satellite-informed simulation of irrigation in South Asia: opportunities and uncertainties. J Hydrol (Amst) 641 (2024).
Zhao, M., A, G., Liu, Y. & Konings, A. G. Evapotranspiration frequently increases during droughts. Nat Clim Chang 12 , 1024–1030 (2022).
Dai, W., Subedi, R., Jin, K. & Hao, L. Spatiotemporal variation of potential evapotranspiration and meteorological drought based on multi-source data in Nepal. Natural Hazards Research 3 , 271–279 (2023).
Ahmed, K. et al . Divergence of potential evapotranspiration trends over Pakistan during 1967–2016. Theor Appl Climatol 141 , 215–227 (2020).
Jerin, J. N. et al . Spatiotemporal trends in reference evapotranspiration and its driving factors in Bangladesh, https://doi.org/10.1007/s00704-021-03566-4/Published .
Bandyopadhyay, A., Bhadra,; A, Raghuwanshi, N. S & Singh, R. Temporal Trends in Estimates of Reference Evapotranspiration over India, https://doi.org/10.1061/ASCEHE.1943-5584.0000006 .
Saha, A. & Jain, M. K. Improved Framework for Attribution Analysis of Contributing Climatic Factors Influencing Reference Evapotranspiration. J Hydrol Eng 30 (2025).
Mishra, V., Bhatia, U. & Tiwari, A. D. Bias-corrected climate projections for South Asia from Coupled Model Intercomparison Project-6. Sci Data 7 (2020).
Dai, A. Drought under global warming: A review. Wiley Interdisciplinary Reviews: Climate Change vol. 2 45–65, https://doi.org/10.1002/wcc.81 (2011).
Sheffield, J., Wood, E. F. & Roderick, M. L. Little change in global drought over the past 60 years. Nature 491 , 435–438 (2012).
Article ADS PubMed CAS Google Scholar
Gupta, V., Kumar Jain, M. & Singh, V. P. Multivariate Modeling of Projected Drought Frequency and Hazard over India. J Hydrol Eng 25 (2020).
Fisher, J. B., Whittaker, R. J. & Malhi, Y. ET come home: Potential evapotranspiration in geographical ecology. Global Ecology and Biogeography vol. 20 1–18, https://doi.org/10.1111/j.1466-8238.2010.00578.x (2011).
Allen, R. G. Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements-FAO Irrigation and Drainage Paper . (F.A.O., 1998).
Zomer, R. J., Xu, J. & Trabucco, A. Version 3 of the Global Aridity Index and Potential Evapotranspiration Database. Sci Data 9 , 409 (2022).
Article PubMed PubMed Central Google Scholar
Song, Y. H., Chung, E. S., Shahid, S., Kim, Y. & Kim, D. Development of global monthly dataset of CMIP6 climate variables for estimating evapotranspiration. Sci Data 10 (2023).
Bjarke, N., Barsugli, J. & Livneh, B. Ensemble of CMIP6 derived reference and potential evapotranspiration with radiative and advective components. Sci Data 10 (2023).
Mondal, S. K. et al . Projected changes in temperature, precipitation and potential evapotranspiration across Indus River Basin at 1.5–3.0 °C warming levels using CMIP6-GCMs. Science of The Total Environment 789 , 147867 (2021).
Zhai, J. et al . Future drought characteristics through a multi-model ensemble from CMIP6 over South Asia. Atmos Res 246 , 105111 (2020).
Sarkar, S., Mukherjee, A., Senapati, B. & Duttagupta, S. Predicting Potential Climate Change Impacts on Groundwater Nitrate Pollution and Risk in an Intensely Cultivated Area of South Asia. ACS Environmental Au 2 , 556–576 (2022).
Article PubMed PubMed Central CAS Google Scholar
Vinke, K. et al . Climatic risks and impacts in South Asia: extremes of water scarcity and excess. Reg Environ Change 17 , 1569–1583 (2017).
Gao, X., Schlosser, C. A., Fant, C. & Strzepek, K. The impact of climate change policy on the risk of water stress in southern and eastern Asia. Environmental Research Letters 13 (2018).
Rehman, A., Batool, Z., Ma, H., Alvarado, R. & Oláh, J. Climate change and food security in South Asia: the importance of renewable energy and agricultural credit. Humanit Soc Sci Commun 11 , 342 (2024).
Eyring, V. et al . Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci Model Dev 9 , 1937–1958 (2016).
O’Neill, B. C. et al . The Scenario Model Intercomparison Project (ScenarioMIP) for CMIP6. Geosci Model Dev 9 , 3461–3482 (2016).
Thrasher, B., Khajehei, S., Kim, J. B. & Brosnan, I. NASA Earth Exchange Downscaled Climate Projections 30 Arcseconds CMIP6. Scientific Data 11 (2024).
Xu, L. & Wang, A. Application of the Bias Correction and Spatial Downscaling Algorithm on the Temperature Extremes From CMIP5 Multimodel Ensembles in China. Earth and Space Science 6 , 2508–2524 (2019).
Millstein, D., Jeong, S., Ancell, A. & Wiser, R. A database of hourly wind speed and modeled generation for US wind plants based on three meteorological models. Sci Data 10 (2023).
An, K., Cai, W., Lu, X. & Wang, C. High-resolution gridded dataset of China’s offshore wind potential and costs under technical change. Sci Data 12 , 69 (2025).
Fang, J. et al . A long-term reconstruction of a global photosynthesis proxy over 1982–2023. Scientific Data 12 (2025).
Fthenakis, V. et al . A solar energy desalination analysis tool, sedat, with data and models for selecting technologies and regions. Sci Data 9 (2022).
Deng, Y. et al . Harmonized and Open Energy Dataset for Modeling a Highly Renewable Brazilian Power System. Sci Data 10 (2023).
Abraha, M. G. & Savage, M. J. Comparison of estimates of daily solar radiation from air temperature range for application in crop simulations. Agric For Meteorol 148 , 401–416 (2008).
Buster, G., Rossol, M., Maclaurin, G., Xie, Y. & Sengupta, M. A physical downscaling algorithm for the generation of high-resolution spatiotemporal solar irradiance data. Solar Energy 216 , 508–517 (2021).
Sentelhas, P. C. et al . Suitability of relative humidity as an estimator of leaf wetness duration. Agric For Meteorol 148 , 392–400 (2008).
Gunawardhana, L. N., Al-Rawas, G. A. & Kazama, S. An alternative method for predicting relative humidity for climate change studies. Meteorological Applications 24 , 551–559 (2017).
Laurence, H., Fabry, F., Dutilleul, P., Bourgeois, G. & Zawadzki, I. Estimation of the Spatial Pattern of Surface Relative Humidity Using Ground Based Radar Measurements and Its Application to Disease Risk Assessment . Agricultural and Forest Meteorology vol. 111 (2002).
Schulzweida, U. CDO User Guide 2.3.0 , https://doi.org/10.5281/zenodo.10020800 (2023).
R Core Team. R: A Language and Environment for Statistical Computing. https://www.R-project.org/ (2024).
Cannon, A. J., Sobie, S. R. & Murdock, T. Q. Bias correction of GCM precipitation by quantile mapping: How well do methods preserve changes in quantiles and extremes? J Clim 28 , 6938–6959 (2015).
Gennaretti, F., Sangelantoni, L. & Grenier, P. Toward daily climate scenarios for Canadian arctic coastal zones with more realistic temperatureprecipitation interdependence. J Geophys Res 120 , 11,862–11,877 (2015).
Petrie, R. et al . Coordinating an operational data distribution network for CMIP6 data. Geosci Model Dev 14 , 629–644 (2021).
Copernicus Climate Change Service. ERA5 hourly data on single levels from 1940 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS) , https://doi.org/10.24381/cds.adbb2d47 (2023).
NASA JPL. NASA Shuttle Radar Topography Mission Global 3 arc second. NASA Land Processes Distributed Active Archive Center , https://doi.org/10.5067/MEASURES/SRTM/SRTMGL3.003 (2013).
Cheng, W. et al . Global monthly gridded atmospheric carbon dioxide concentrations under the historical and future scenarios. Sci Data 9 (2022).
Thrasher, B., Maurer, E. P., McKellar, C. & Duffy, P. B. Technical Note: Bias correcting climate model simulated daily temperature extremes with quantile mapping. Hydrol Earth Syst Sci 16 , 3309–3314 (2012).
Lavoie, J. et al . An ensemble of bias-adjusted CMIP6 climate simulations based on a high-resolution North American reanalysis. Sci Data 11 (2024).
Smitha, P. S., Narasimhan, B., Sudheer, K. P. & Annamalai, H. An improved bias correction method of daily rainfall data using a sliding window technique for climate change impact assessment. J Hydrol (Amst) 556 , 100–118 (2018).
Kim, D., Choi, M. & Kang, M. Testing new potential evaporation formulations for identifying soil moisture deficiency in agricultural areas under global warming. J Hydrol (Amst) 661 (2025).
Roderick, M. L., Greve, P. & Farquhar, G. D. On the assessment of aridity with changes in atmospheric CO2. Water Resour Res 51 , 5450–5463 (2015).
Liu, Z., Wang, T., Li, C., Yang, W. & Yang, H. A physically-based potential evapotranspiration model for global water availability projections. J Hydrol (Amst) 622 (2023).
Gui, Y. et al . Attribution analyses of reference evapotranspiration changes in China incorporating surface resistance change response to elevated CO2. J Hydrol (Amst) 599 (2021).
Yang, Y., Roderick, M. L., Zhang, S., McVicar, T. R. & Donohue, R. J. Hydrologic implications of vegetation response to elevated CO2 in climate projections. Nature Climate Change vol. 9 44–48, https://doi.org/10.1038/s41558-018-0361-0 (2019).
Casanueva, A. et al . Testing bias adjustment methods for regional climate change applications under observational uncertainty and resolution mismatch. Atmospheric Science Letters 21 , e978 (2020).
Vrac, M. Multivariate-Intervariable, Spatial, and Temporal-Bias Correction*, https://doi.org/10.1175/JCLI-D-14 .
Maraun, D. et al . Towards process-informed bias correction of climate change simulations. in Nature Climate Change vol. 7 764–773 (Nature Publishing Group, 2017).
Chen, C., Haerter, J. O., Hagemann, S. & Piani, C. On the contribution of statistical bias correction to the uncertainty in the projected hydrological cycle. Geophys Res Lett 38 (2011).
Muerth, M. J. et al . On the need for bias correction in regional climate scenarios to assess climate change impacts on river runoff. Hydrol Earth Syst Sci 17 , 1189–1204 (2013).
Tu, Z. & Yang, Y. On the Estimation of Potential Evaporation Under Wet and Dry Conditions. Water Resour Res 58 , e2021WR031486 (2022).
Kim, D., Cho, E. & Chun, J. A. Constraining the power-function complementary relationship with a steady-state Budyko equation for predicting terrestrial evapotranspiration globally. Agric For Meteorol 344 (2024).
Zhou, S. & Yu, B. Reconciling the Discrepancy in Projected Global Dryland Expansion in a Warming World. Glob Chang Biol 31 , e70102 (2025).
Article ADS PubMed PubMed Central CAS Google Scholar
Saha, A., Jain, M. K., Joshi, P., Das, S. & Rawat, N. S. Development of Daily Downscaled, Bias-corrected CMIP6 climate Datasets for estimating PET in South Asia. Zenodo , https://doi.org/10.5281/zenodo.15670655 (2025).
Ballarin, A. S. et al . CLIMBra - Climate Change Dataset for Brazil. Sci Data 10 (2023).
Switanek, M. B. et al . Scaled distribution mapping: a bias correction method that preserves raw climate model projected changes. Hydrol Earth Syst Sci 21 , 2649–2666 (2017).
Thrasher, B. et al . NASA Global Daily Downscaled Projections, CMIP6. Sci Data 9 (2022).
Feng, X., Thompson, S. E., Woods, R. & Porporato, A. Quantifying Asynchronicity of Precipitation and Potential Evapotranspiration in Mediterranean Climates. Geophys Res Lett 46 , 14692–14701 (2019).
Fedele, G., Reder, A. & Mercogliano, P. Statistical Downscaling over Italy using EQM: CMIP6 Climate Projections for the 1985–2100 Period. Scientific Data 12 (2025).
She, D., Xia, J. & Zhang, Y. Changes in reference evapotranspiration and its driving factors in the middle reaches of Yellow River Basin, China. Science of the Total Environment 607–608 , 1151–1162 (2017).
Dix, M. et al . CSIRO-ARCCSS ACCESS-CM2 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.2285 (2019).
Ziehn, T. et al . CSIRO ACCESS-ESM1.5 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.2291 (2019).
Swart, N. C. et al . CCCma CanESM5 model output prepared for CMIP6 CMIP https://doi.org/10.22033/ESGF/CMIP6.1303 (2019).
Danabasoglu, G. NCAR CESM2-WACCM model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.10026 (2019).
Lovato, T. & Peano, D. CMCC CMCC-CM2-SR5 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.1365 (2020).
Lovato, T., Peano, D. & Butenschön, M. CMCC CMCC-ESM2 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.13168 (2021).
Li, L. CAS FGOALS-g3 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.2056 (2019).
John, J. G. et al . NOAA-GFDL GFDL-ESM4 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.1414 (2018).
Volodin, E. et al . INM INM-CM4-8 model output prepared for CMIP6 CMIP https://doi.org/10.22033/ESGF/CMIP6.1422 (2019).
Volodin, E. et al . INM INM-CM5-0 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.12322 (2019).
Shiogama, H., Abe, M. & Tatebe, H. MIROC MIROC6 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.898 (2019).
Yukimoto, S. et al . MRI MRI-ESM2.0 model output prepared for CMIP6 ScenarioMIP https://doi.org/10.22033/ESGF/CMIP6.638 (2019).
A.S. wishes to sincerely acknowledge the financial support provided by the Prime Minister’s Research Fellowship (PM-31-22-630-414) under the Government of India. The authors acknowledge the National Supercomputing Mission (NSM) for providing computing resources of ‘PARAM Ganga’ at the Indian Institute of Technology Roorkee, which is implemented by C-DAC and supported by the Ministry of Electronics and Information Technology (MeitY) and Department of Science and Technology (DST), Government of India.
Department of Hydrology, Indian Institute of Technology Roorkee, Roorkee, India
Aniruddha Saha, Manoj Kumar Jain, Pranita Joshi & Naimesh Singh Rawat
Department of Soil and Water Systems, University of Idaho, Moscow, ID, USA
Moscow Forestry Sciences Laboratory, USDA Forest Service-Rocky Mountain Research Station, Moscow, ID, USA
Search author on: PubMed Google Scholar
A.S., P.J. and M.K.J. conceived and designed the analysis. A.S. was responsible for data wrangling, processing and applying the downscaling algorithm. P.J. was responsible for downscaling algorithm and drafted the initial manuscript with A.S.; N.S.R. and P.J. were responsible for data collection and data storage. S.D. and A.S. were responsible for technical validation. All authors contributed to the development of the manuscript.
Correspondence to Manoj Kumar Jain .
The authors declare no competing interests.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
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 licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence 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 licence, visit http://creativecommons.org/licenses/by/4.0/ .
Saha, A., Jain, M.K., Joshi, P. et al. Development of daily downscaled, bias-corrected CMIP6 climate datasets for estimating reference evapotranspiration (ETo) in South Asia. Sci Data 12 , 1879 (2025). https://doi.org/10.1038/s41597-025-06149-4
Version of record : 28 November 2025
DOI : https://doi.org/10.1038/s41597-025-06149-4
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
© 2026 Springer Nature Limited
Loading comments…