If details of Earth’s climate could be resolved down to 4-km spatial resolution and hourly time scales for long time periods, it could revolutionize climate change studies around the world. The increased computer power, better supporting data, and improvements in atmospheric modeling now available make this possible for the conterminous United States (CONUS). This paper presents such a 4-km, 40-plus-year dataset (called the CONUS404) that was produced through a collaborative initiative between the U.S. Geological Survey (USGS) and the National Science Foundation–sponsored National Center for Atmospheric Research (NCAR). This dataset was created by NCAR staff on the USGS High Performance Computer (HPC) Denali, and took over 11 months of continuous simulation time to complete. The CONUS404 data are archived and accessible to the public at https://doi.org/10.5066/P9PHPK4F (Rasmussen et al. 2023).
In the past, in situ and remote sensing datasets have been assimilated into atmospheric models. Kalnay et al. (1996) created the first long-term global reanalysis dataset using surface, weather balloon, and remotely sensed data. This dataset and others like it have been a valuable tool to understand the physics and dynamics of Earth at a variety of time and space scales (e.g., Pascolini-Campbell et al. 2021; Hari et al. 2022; Ballinger et al. 2022). A “global reanalysis type” dataset has been created ∼ every 5 years using new and improved ground bases and remote sensing datasets and assimilation methods (e.g., Uppala et al. 2005; Kobayashi et al. 2014; Hersbach et al. 2020). A relatively long period (>10 years) of climate forcing data are generally needed to calibrate hydrological models against important observed hydrologic fluxes such as streamflow.
Ideally, these global datasets used to force and calibrate hydrological models should have physically consistent estimates of temperature, radiation, and precipitation that are at fine time and space scales consistent with the desired applications. Current reanalysis products, such as the fifth-generation European Centre for Medium-Range Weather Forecasts atmospheric reanalysis of the global climate dataset (ERA5; Hersbach et al. 2020), provide hourly forcing data, but only at 30-km or coarser grid spacing. While these data satisfy the criteria of fine temporal resolution and physical consistency across the provided atmospheric variables, they are generally too spatially coarse to resolve mesoscale atmospheric phenomena, such as convective systems, and to accurately place them over/onto the hydrologic modeling basins (such as a 12-digit hydrologic units). As a result, hydrologists have typically resorted to using collections of variables extracted from independently produced source data (see Table 1) and that still might not represent local-scale weather. While such an approach has proven effective for many applications, calibration methods can end up compensating for omissions or physical inconsistencies and ultimately provide “the right answer for the wrong reasons.” There is a clear need to create a spatiotemporal dataset of sufficient duration, resolution, and physical consistency that supports accurate simulation of hydrologic states and fluxes at scales (i.e., mesoscales) that are relevant for water-resources managers and policy-makers.
A summary of gridded and station observation datasets that contain at least 42 years of data. Boldface text indicates gridded reanalysis type data, including the dynamically downscaled CONUS-404 dataset.
Creating such a fine-scale and self-consistent dataset is no easy task but now possible thanks to three notable science and technology advancements in the last decade: 1) available high-quality global reanalysis (e.g., ERA5) with reasonable spatial resolution as boundary conditions for fine-scale dynamic downscaling, 2) convection-permitting modeling with improved physics parameterizations that is able to resolve mesoscale and land–atmosphere interaction features critical for watershed hydrologic modeling (Rasmussen et al. 2011; Barlage et al. 2021), and 3) significantly increased computational resources. Therefore, NCAR and USGS scientists initiated a collaborative computationally intensive effort to create a new 40-plus-year dataset with 4-km grid spacing (i.e., at convection permitting spatial scale) over the conterminous United States from 1979 to 2021 using a mesoscale model through dynamical downscaling.
Model simulations with horizontal grid spacing equal to or less than 4 km without a convective parameterization are able to well capture precipitation and other important input needed to drive hydrological models (Ikeda et al. 2010; Rasmussen et al. 2011; Prein et al. 2015; Barlage et al. 2021; Ikeda et al. 2021). Here, the model used was the state-of-the-art Weather Research and Forecasting (WRF) mesoscale model (Skamarock et al. 2008) over CONUS with its outer domain (Fig. 1) constrained by the hourly 0.25° ERA5 global dataset to produce a 42-plus-year dynamically consistent dataset with hourly 4-km output.
(a) Domain of the 4-km CONUS404 simulation with land surface elevation height (m) (www.usgs.gov/coastal-changes-and-impacts/gmted2010); (b) land-use and land-cover type used in the model simulation from MODIS satellite data (https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=968).
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
The NCAR Water System Program and other research have shown that a resolution of 4 km is minimally sufficient to 1) resolve the orographically forced precipitation that forms snowpack in complex terrain (Ikeda et al. 2010; Rasmussen et al. 2011; He et al. 2019; Ikeda et al. 2021), 2) reproduce the observed dominant warm-season precipitation systems [i.e., mesoscale convective systems (MCSs)] in the central and eastern United States (Prein et al. 2015, 2017), 3) represent fine-scale groundwater–atmosphere interactions that drive MCSs in the summer season (reduces temperature warm bias and water vapor dry bias; Barlage et al. 2021), 4) capture the intensity, frequency, and amount of precipitation including light to moderate convective rainfall (this paper; Scaff et al. 2020), and 5) represent gravity wave drag without a gravity wave parameterization (Stephan et al. 2019). In fact, using a 4-km horizontal grid spacing removes the need for both a convective and wave drag parameterization in our simulations and is the focus of current research and workshops (see www.cima.fcen.uba.ar/cpcmw2022/ website). The ability to appropriately capture the water cycle in climate models represents the culmination of two decades of research mentioned above. This goal was described by Trenberth et al. (2003), and has largely been achieved by decreasing climate model grid spacing down to 4 km or less with the results discussed above.
This paper uses these new results to downscale the global ERA5 dataset from a 30-km horizontal grid spacing to 4-km horizontal grid spacing using the WRF Model over a 40-plus-year period over the conterminous United States (Fig. 1). The second section provides an overview of the modeling approach, and the third section shows evaluations of the CONUS404 dataset quality, including biases, that demonstrates the use of CONUS404 to improve simulation of the some of the critical hydrometeorological variables listed in the appendix. The fourth section gives a summary of evaluation and the fifth section some final remarks.
Modeling approach used to create the CONUS404 dataset
The WRF Model version 3.9.1 (Skamarock et al. 2008) is used to create a 4-km gridded model output over the CONUS (also covering a significant portion of southern Canada and northern Mexico, Fig. 1) for 40-plus years from 1 October 1979 to 30 September 2021. The model domain has 1,368 grid points in the east–west direction and 1,016 grid points in the north–south direction yielding a domain size of 5,472 km × 4,064 km. The model is configured with 51 stretched vertical levels, with the top at 50 hPa. The finest vertical grid spacing is ∼50 m near the surface, stretching to ∼700 m near the model top. The model is initialized at 0000 universal coordinated time (UTC) 1 October 1979 and updated every 3 h at the lateral and lower boundaries using the ERA5 reanalysis data (Hersbach et al. 2020). The model time step is 20 s throughout the simulation period.
The convection-permitting WRF configuration including the choice of physics schemes used in the CONUS404 simulations was built upon our experiences from the Colorado headwaters simulation (Rasmussen et al. 2011, 2014) and multiyear CONUS simulations (Liu et al. 2017), as well as our previous comprehensive testing of various physics parameterizations (Liu et al. 2011). The microphysics package was shown to be the critical scheme for convection-permitting simulations (Liu et al. 2011).
In summary, the key WRF physics packages used were Thompson and Eidhammer (2014) microphysics, Yonsei University (YSU) planetary boundary layer scheme (Hong et al. 2006), the Rapid Radiative Transfer Model (RRTMG) radiation scheme (Iacono et al. 2008), and the Noah multiparameterization land surface model (Noah-MP; Niu et al. 2011) coupled to the Miguez-Macho–Fan groundwater scheme (MMF; Miguez-Macho et al. 2007; Barlage et al. 2021). The Noah-MP multiple-physics-based land surface model used for CONUS404 employed a 10-yr climatology of Moderate Resolution Imaging Spectroradiometer (MODIS) monthly leaf area index (https://ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/products/MOD15A2/), and the partitioning between rain and snow was determined by the WRF microphysics and not Noah-MP. In the CONUS404 simulation, the updated version of the Noah-MP code differs from the WRF/Noah-MP standard release in WRF v.3.9.1 in the following ways: 1) snow cover fraction formulation was improved to better match MODIS observed snow cover and albedo and to reduce winter/spring temperature cold bias in midwestern U.S. regions covered by snow, 2) snow emissivity values for fresh snow were changed from 1.0 to 0.95; and 3) wind–canopy absorption parameters modified as function of land-cover and land-use types based on Goudriaan (1977) and Niu and Yang (2004). These modifications and tests are described in He et al. (2021). As a result of this NCAR–USGS collaborative project, the modifications were released in the official Noah-MP community model GitHub (https://github.com/NCAR/noahmp) as well as in WRF V4.3 (www2.mmm.ucar.edu/wrf/users/docs/technote/contents.html).
The WRF Model simulates climate warming due to greenhouse gases (GHG) by varying the concentration of CO2 and CH4 according to observations during the 42-yr simulation period. Observed time-varying GHGs were used in WRF. The aerosol–radiation interactions were considered by using a monthly 5° × 4° aerosol climatological dataset in the RRTMG scheme (Tegen et al. 1997), which has six types of aerosols: organic carbon, black carbon, sulfate, sea salt, dust, and stratospheric aerosol. In addition, the ocean and lake temperature were prescribed to be the same as the ERA5 data. It is important to minimize the model drift and maintaining land surface–boundary layer–atmosphere interactions (Lo et al. 2008; Xu and Yang 2015). Therefore, spectral nudging of temperature, winds, and geopotential height was applied every 2 min above the boundary layer for wavenumbers 1–3 in the zonal (east–west) direction and wavenumbers 1–2 in the meridional (north–south) direction (>1,800-km scale; Miguez-Macho et al. 2004) to reduce the drift of simulated synoptic storm locations as they progressed in the interior of the large geographic domain and to more accurately capture the vertical distribution of temperature, moisture, and wind over the conterminous United States. This technique keeps the high-resolution CONUS404 simulations steering flow consistent with the coarser input reanalysis data and allows WRF to resolve mesoscale processes at 4-km horizontal scales. Spectral nudging improved the previous high-resolution 13-yr conterminous U.S. simulations (called CONUS-I; Liu et al. 2017; Prein et al. 2017; Ikeda et al. 2021) and significantly improved the current 42-yr simulation as shown in Fig. 2. Such spectral nudging helps to create a realistic vertical distribution of temperature, humidity, and wind that is nearly indistinguishable from radiosonde data. Thus, it allows the high-resolution local and regional forcing (topography, low-level convergence, land surface heterogeneity) to operate under the correct synoptic weather conditions over the CONUS404 simulation period.
Comparison of CONUS404 simulated vertical profile of temperature to radiosonde data at (left) 1200 and (right) 0000 UTC over the regions indicated by the radiosonde dots on the U.S. map above each column (e.g., the Northeast region is defined by the four radiosonde points shown in red on the map). Black dots represent all radiosonde sites in the United States. (top ) The difference in temperature between CONUS404 and radiosonde), (middle) the difference in relative humidity, and (bottom) the difference in wind speed averaged over the 42 years. The y axis in the plots is elevation. The hatched regions of the plots indicate differences that are statistically significant (less than 5% on the Mann–Whitney U test) than the background. Black contours are observed average values over the 40 years of the specified variable. The key for each row is given on the right.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Evaluation of the CONUS404 output of precipitation, temperature, and downward solar radiation
The following provides an evaluation of precipitation, temperature, and downward solar radiation, which are three of the key variables used to force hydrologic models and to understand the evolution of regional hydroclimate. Table 1 presents station datasets that used to evaluate CONUS404 output of precipitation and temperature, as well as widely used gridded datasets and reanalysis dataset used to drive hydrologic models, which the CONUS404 data will be evaluated against (Table 2).
Qualitative comparison of hydrological model forcing datasets. Good performance means that datasets have higher skill compared to others while suboptimal performance is interpreted relatively to other datasets.
Precipitation.
It is difficult to estimate the mean annual observed precipitation over CONUS by interpolating across observations due to the wide variety of precipitation gauge types, wind undercatch (Rasmussen et al. 2012), uncertainties in radar–rain gauge relationship in complex terrain, and other factors. We compared the CONUS404 data to the PRISM (PRISM Climate Group 2014) data for both precipitation and temperature. Although we do not view the PRISM dataset as a “reference” dataset, it has been widely used and has the benefit of some human intervention in its quality control (Strachan and Daly 2017). The PRISM developers also chose sites with long-term records (Rupp et al. 2022), which provides better temporal homogeneity across our the 42-yr model simulation.
In the following, we evaluate CONUS404 precipitation by first examining large-scale precipitation patterns, then its diurnal variations, followed by summer MCSs and the probability distribution of hourly precipitation, and finally precipitation characteristics in high-elevation sites.
Annual precipitation amount and trend.
In general, the mean of the annual total CONUS404 precipitation (Fig. 3a) agrees well with the PRISM estimate (Fig. 3b), and both datasets show precipitation maxima over the mountains of the Pacific Northwest and the southeastern United States. The precipitation trend in both CONUS404 and PRISM (Figs. 3c,d) indicates statistically significant decreased precipitation in the southwestern United States and significantly increasing precipitation in the Northeast over the period of the CONUS404 simulation.
Averaged annual total precipitation (mm) for 1986–2020 water year from (a) CONUS404 and (b) PRISM; 1986–2020 annual total precipitation linear trends (mm yr−1) from (c) CONUS404 and (d) PRISM using the least squares fit. The data from 1979 to 1985 were not included in this estimate to account for the model spinup periods. The stippled region in (c) and (d) indicate statistical significance at the 5% level based on the Mann–Kendall test. A water year is defined as the 12-month period from 1 Oct for any given year through 30 Sep of the following year. A water year is designated by the calendar year in which it ends.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Monthly precipitation amount.
As shown in Fig. 4a, winter and early spring precipitation are major contributors to the annual precipitation over the Pacific Northwest, while summer precipitation substantially contributes to the annual precipitation in the eastern and southeastern United States. More importantly, monthly precipitation difference between PRISM and CONUS404 over most of the CONUS, shown in Fig. 4b, is close to zero. There are apparent winter differences along the coastal mountains of the Pacific Northwest and in the lower Mississippi River basin, and summer differences along Florida and southeastern coasts up to about Maryland. A quantitative statistical evaluation of CONUS404 monthly precipitation is documented in Table S1 in the online supplemental material (https://doi.org/10.1175/BAMS-D-21-0326.2).
(a) CONUS404 monthly precipitation climatology from 1986 to 2020 and (b) the mean precipitation bias (i.e., CONUS404 − PRISM) during the 1986–2020 water years.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Diurnal cycle of precipitation.
Many studies show that kilometer-scale climate model simulations better reproduce observed frequency and intensity of hourly precipitation [see Prein et al. (2015) for a review] even though the total amount is well represented in coarser-resolution simulations such as the 36-km simulations with a convective parameterization reported by Mooney et al. (2017) or ERA5. Figure 5 compares these precipitation characteristics in 4-km CONUS404 simulations (without using a convection parameterization scheme) to the WRF 36-km simulations with three different convection parameterization schemes. CONUS404 simulations provide sufficient details to capture the intensity, frequency, duration, and amount of hourly precipitation. Coarse-resolution climate models and reanalysis data (e.g., ERA5) usually focused on daily precipitation amount. CONUS404, by contract, essentially captured the hourly behavior of all precipitation by the right reason (e.g., correctly simulating the phases of mesoscale systems).
Comparison of CONUS404-simulated mean diurnal cycles of hourly precipitation (a),(d) amount, (b),(e) frequency, and (c),(f) intensity averaged over the western United States [stations map inset in (a)] with (top) 36- and (bottom) 4-km horizontal grid spacing during June–August. The 36-km results are from Mooney et al. (2017) and the 4-km estimates are from CONUS404. The 36-km simulations used three different convective parameterizations: Kain–Fritsch (Kain 2004), Tiedtke (1989), and NSAS (Han and Pan 2011), and the 4-km simulation did not use convective parameterization. The bottom panels also show ERA5 precipitation statistics (in blue). The shading in the top panel shows the sensitivity to microphysics, radiation, and planetary boundary layer schemes [see Mooney et al. (2017) for details]. The shading in the bottom panel shows the spread in the interannual variability.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
MCSs.
The chief reason for the CONUS404 to reproduce observed-precipitation patterns is largely the WRF’s explicit simulation of convection that allows the simulation to capture MCSs, the dominant summer precipitation producing storms east of the Rocky Mountains (Prein et al. 2017). Simulated MCSs is evaluated through a Lagrangian framework to identify and track them as objects over time according to cloud temperature characteristics in both the CONUS404 and in GOES Gridded Satellite (GridSat) observational data (Fig. 6). GridSat data are 3-hourly calibrated infrared (near 11 μm) brightness temperature observations (available from 1981 onward; Knapp et al. 2011). The CONUS404 and ERA5 data were regridded to the GridSat grid before applying the Lagrangian framework to track the MCSs. Cloud temperature was derived from longwave outgoing radiation at the top of the atmosphere following Wu and Yan (2011). The MCS identification and tracking algorithm is further described in Maddox (1980), Prein et al. (2021), and Feng et al. (2021).
(a)–(c) An example of MCS extraction using the Lagrangian method over the Midwest on 5 Jun 2020 based on (a) the GridSat satellite channel 4, (b) the CONUS404 simulation, and (c) ERA5. Red polygons outline the identified MCS and the red line with red circles shows the MCS track. (d)–(g) The mean seasonal cycle of MCS frequencies is shown for the (d) Southeast, (e) Midwest, (f) Southern Plains, and (g) Northeast for GridSat (black), CONUS404 (red), and ERA5 (blue) data. The shading shows the interquartile spread of year-to-year variations in MCS frequencies. (h)–(k) Annual (thin lines) and 5-yr running-averaged (thick lines) MCS frequency time series from GridSat (black), CONUS404 (red), and ERA5 (blue) over the (h) Southwest, (i) Midwest, (j) Southern Plains, and (k) Northeast. The correlation coefficients between GridSat and CONUS404 (rCONUS404) and GridSat and ERA5 (rERA5) are shown in (h)–(k).
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
An example of an MCS that passed over the U.S. Midwest on 5 June 2020 is shown based on GridSat, CONUS404, and ERA5 brightness temperature (Figs. 6a–c). ERA5 simulates a blurry looking, small system that is displaced northward and has a shorter lifespan relative to GridSat observations. By contrast, CONUS404 MCSs matched those of GridSat much more closely and produce a more realistic looking storm in terms of its structure, initiation, and east–west movement.
Further, Figs. 6d–g show a closer match of MCS statistics east of the Continental Divide between CONUS404 and GridSat than between ERA5 and GridSat. The seasonal cycle of MCSs frequencies is well simulated in CONUS404, while ERA5 features a 50% low bias in frequency, which is particularly pronounced in June, July, and August. The CONUS404 simulation shows the largest deviations from observations in the Midwest where more frequent MCSs occur in spring and slightly fewer MCSs in late summer.
Figures 6h–k show that CONUS404 improves the frequency, long-term variability, trends, and annual cycle of MCSs compared to ERA5. This is a substantial improvement compared to earlier convection-permitting model simulations over the CONUS (Prein et al. 2020).
Liu et al. (2017) reported that the central United States had a substantial warm and dry bias (5°C warm bias and 50% too-low relative humidity) in the previous CONUS-1 simulation, which was associated with a low number of MCSs by Prein et al. (2020). Barlage et al. (2021) demonstrated that incorporating the MMF groundwater scheme in WRF substantially reduced warm/dry biases in the central United States. This conclusion also supports the interpretation that appropriate groundwater representation in the downscaling process is an important factor in the ability of CONUS404 to improve depiction of MCSs over the input ERA5 data (Figs. 7a,d). As MCSs are the main source of summer precipitation east of the Rockies, this is considered a major improvement to the CONUS404 simulation of precipitation.
Difference in May–September (a)–(c) 2-m air mean temperature (T2) and (d)–(f) precipitation between CONUS1, CONUS404, and PRISM over CONUS averaged for the 13 common years simulated by CONUS-1.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Another interesting aspect regarding the scale dependency of the MMF groundwater scheme was that the warm and dry bias still remained for WRF simulations with a grid spacing larger than 9 km, but largely disappeared for 4-km and finer horizontal grid spacings. This is ascribed to the finer-resolution applications being able to resolve the river valleys (e.g., riparian zones), which contributes to recharging the depleted summer root-zone soil moisture (Barlage et al. 2021). Without this recharge mechanism, the roots are not able to sustain evapotranspiration. Downward solar radiation is converted to sensible heat flux, which results in the air near the surface becoming overheated and dried out.
Impacts of groundwater can be seen in the temperature and precipitation difference plots from the CONUS-1 simulation and the CONUS404 for the months of May–September (Fig. 7). Note that the only difference in summer-weather physics parameterizations between CONUS-1 and CONUS404 is that the latter incorporated the MMF groundwater scheme. Compared to CONUS-1, CONUS404 reduced the warm temperature bias by 1°–2°C during this period by adding the groundwater scheme, and increased precipitation by ∼100 mm yr−1. The CONUS404 bias of monthly temperature (not shown) is slightly above 1°C over the central United States from February through September, confirming the important role of groundwater–atmosphere interactions in regulating regional surface temperature and precipitation over the central United States.
Probability density distribution.
Regarding the probability density function of hourly precipitation rate, CONUS404 agrees better than ERA5 and AORC, compared with the COOP observations cross the CONUS (serving as “measured” in this case), as well as in the South Atlantic Gulf, mid-Atlantic, and Ohio regions (Fig. 8). CONUS404 provides improved estimates of heavy and extreme hourly precipitation compared to ERA5 and AORC datasets (Fig. 8, Table 1). One possible reason for lack of high precipitation rate for the lower Colorado River basin is that the 4-km CONSUS404 does not resolve the steep mountain peaks in that basin, which typically drives the heavy precipitation.
Probability density function of hourly precipitation rates for the (a) CONUS, (b) South Atlantic–Gulf, (c) mid-Atlantic, (d) Ohio, and (e) lower Colorado regions. The black curve is from the COOP observations, red from CONUS-CONUS404, blue from AORC, and green from ERA5.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Precipitation and snow over complex terrains.
The CONUS404 simulation performs well in capturing orographic-driven precipitation over the complex terrain of the western United States based on comparison to Snow Telemetry (SNOTEL) network (www.nrcs.usda.gov/wps/portal/wcc/home/snowClimateMonitoring/snowpack/snowpackMaps, accessed 2020) data (Figs. 9 and 10). CONUS404 estimation of annual total precipitation (Fig. 10a) and its spatial variability across SNOTEL sites is excellent over the 35-yr period, but the simulated snow water equivalent (SWE) is lower than estimated by SNOTEL stations by about 15%. This is consistent with prior WRF simulation results for the Colorado River headwaters and for western mountains (Ikeda et al. 2010; Rasmussen et al. 2011; Liu et al. 2017; Ikeda et al. 2021), which also show that resolutions coarser than 6 km are inadequate to estimate winter precipitation in complex terrain. He et al. (2021) emphasized that the high bias in WRF-simulated downward solar radiation, a high downward sensible heat flux to snowpack, and enhanced ground solar radiation absorption in Noah-MP are possible reasons for low bias in SWE.
(left) SNOTEL sites (dots) and their elevation (color coding) used in this analysis. (right) Photographs were taken by Roy Rasmussen of NCAR from the Glacial Lake Ecosystem Experiments Site in Wyoming, showing a (top right) typical SNOTEL site with a precipitation gauge and (bottom right) snow pillow used at the SNOTEL sites.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Comparison of CONUS404 with SNOTEL data averaged over water year 1986–2020: (a) accumulated precipitation from CONUS404 (red solid) and SNOTEL (blue solid), and SWE from CONUS404 (red dashed) and SNOTEL (blue dashed); (b) scatterplot of annual precipitation accumulation from CONUS404 and SNOTEL. The data were averaged over all the SNOTEL sites shown in Fig. 9a. Each dot represents one year of data at one site. The first 5 years (1979–85) are not included in the average due to spinup issues.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Temperature.
CONUS404 mean annual temperature (Fig. 11a) agrees well with PRISM (Fig. 11b), and so does the 35-yr trend (Figs. 11c,d). Both the CONUS404 and PRISM temperature trends reveal the same cooling over northern Montana, and a clear warming signal over a large portion of the CONUS region, especially over the southwestern United States. These CONUS404 simulated trends are consistent with IPCC estimates (IPCC 2014).
The 35-yr (1986–2020) mean annual temperatures from (a) CONUS404 and (b) PRISM; 35-yr mean annual temperature trend from (c) CONUS404 and (d) PRISM using the least squares fit. The data from 1979 to 1985 were not included in this estimate to avoid model spinup periods that were found to exist for these years. The stippled region in (c) and (d) indicate statistical significance at the 5% level based on the Mann–Kendall test.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Differences in the annual mean temperature between CONUS404 and PRISM are relatively small over most of CONUS (Fig. 11). However, a bias assessment using the surface temperature at ASOS sites shows a slight CONUS404 cold bias in the east (<1°C in most cases) and a slight warm bias in the west (also <1°C in most cases; not shown). This may be attributed to the inadequacy of the WRF land surface and planetary boundary layer parameterization schemes to handle winter boundary layer development. The sounding data suggest a morning stable boundary layer in the western United States, and CONUS404 also forms a stable boundary layer, but not as cold or quickly as the sonde observations (Fig. 2), leading to a winter warm bias in the early morning hours at most western sites (1200 UTC). In addition, the temperature in October is warmer than observed, and since October is the transition month from rain to snow, a substantial amount of the precipitation in October is simulated in CONUS404 as rain instead of snow. This is evident in time series of precipitation amount from both the CONUS404 and the SNOTEL (figure not shown).
The cold bias in the eastern United States seems to be planetary boundary layer related as well, but not due to an inversion. The sounding data do not show the development of an inversion of cold air near the surface as in the western United States (Fig. 2). The CONUS404 cools too much and creates a layer of cold air extending from 1,000 m to the surface. While 4 km seems to create a reasonable hydroclimate for precipitation, there does seem to be some regionally specific temperature biases near the surface. As shown in the 2-m air temperature probability–density functions comparison between CONUS404 and ASOS station observations (Fig. S1), CONUS404 well reproduced the observed features (especially the extreme high temperature), but has a longer tail for the extreme low temperature due to cold bias.
Temperature at high elevations is of particular interest for snow hydrologic modeling, because it is often used to partition the precipitation into snowfall and rainfall in snow hydrologic models. Comparison of the SNOTEL data with independent NCAR temperature measurements at five Wyoming SNOTEL sites revealed a cold bias in the SNOTEL measured temperature at elevations above 2,000 m. The CONUS404 temperature agreed well with the independent NCAR data. The NCAR sensors are calibrated before deployment and have been used extensively in the evaluation of WRF simulation results in this region (Rasmussen et al. 2018).
Data from one of the NCAR sites are shown from 2008 to 2009 in Fig. 12 (other years and sites show the same behavior). Note the nearly 1:1 agreement with the RAL HY temperature and the CONUS404 temperature in the lower-right scatterplot, and the significant bias (nearly 2°C), when the RAL HY temperature is compared to the SNOTEL temperature.
Temperature at a research site, Highway 47 in SE Wyoming, maintained by the Research Applications Laboratory (RAL) and located in SE Wyoming. The original data from this site (hereafter called RAL HY) were used by Rasmussen et al. (2018) to evaluate snow enhancement studies in the region. The RAL HY temperature collocated with the SNOTEL temperature (°C) are compared in (d) a time series and (e) a scatterplot. (a) The location and photos of the (b) RAL HY site (taken by Roy Rasmussen of NCAR) and (c) SNOTEL site (https://wcc.sc.egov.usda.gov/nwcc/site?sitenum=852).
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
The cold bias in SNOTEL temperature observations was also observed by Oyler et al. (2015) and was attributed to the deployment of a new temperature sensor by the Natural Resources Conservation Service (NRCS) in which the conversion from voltage to temperature was nonlinear. Our analysis in Fig. 12 used the old linear equation converting the voltage to temperature. If the new voltage to temperature nonlinear equation of Oyler et al. (2015) is used to estimate SNOTEL temperature, a good agreement with the independent NCAR temperature sensor data results (figure not shown). The use of the SNOTEL temperature data from the new equation revealed a relatively small bias of the CONUS404 to SNOTEL data (analysis not shown).
Radiation.
The satellite-based National Solar Radiation Database (NSRDB) provides 4-km downward shortwave estimates with about +5% bias against surface site observations (Sengupta et al. 2018), and has been used in previous evaluations of WRF predictions of downward solar radiation (Kim et al. 2022). It is also used here to evaluate downward solar radiation in the CONUS404 data for one year (Fig. 13). The CONUS404 spatial pattern of downward solar radiation shows a remarkable similarity with the observed downward solar radiation for 2018, with high values in the desert southwest and low values in the northeastern United States.
Spatial distributions of 2018 mean surface downward solar radiation (W m−2) from (a) CONUS404 and (b) NSRDB and their (c) absolute bias and (d) percent bias.
Citation: Bulletin of the American Meteorological Society 104, 8; 10.1175/BAMS-D-21-0326.1
Nevertheless, CONUS404 shows a general positive bias in the downward shortwave radiation, especially in the mid-Atlantic region and over Montana and Idaho (Figs. 13c,d), possibly due to uncertainties in treating aerosol properties and water vapor in WRF. However, we suspect that the main cause may be that nonprecipitation clouds simulated in the CONUS404 are underestimated. This is currently under investigation.
Summary evaluation
Table 2 presents a qualitative evaluation of a variety of potential climate datasets that could be used to force hydrologic models, and compares CONUS404 characteristics to many of the available gridded datasets over CONUS (Table 1). We note that the CONUS404 dataset presented in this paper provides good performance in all categories listed on the left except for the bias that has been discussed and noted in previous sections of this paper. Here, good performance should be interpreted relative to the skill of other datasets. The quantitative bias evaluation presented in the paper provides WRF Model developers a road map on needed improvements to the model as well as an indication of the strengths and weaknesses of the current dataset. One of the strengths is the ability to resolve hydroclimate phenomenon at very fine scales.
Final remarks
To address the lack of mesoscale structures in precipitation and temperature in global reanalysis datasets, the USGS and NCAR undertook a major collaborative initiative to create an unprecedented, 4-km, 40-plus-year, self-consistent, “reanalysis type” regional hydroclimate dataset, named CONUS404, over the conterminous United States useful for hydrological and meteorological analysis and modeling. This dataset was generated by the WRF Model dynamically downscaling the latest ERA5 dataset. Consistent with recent scientific findings in the area of convection-permitting modeling, the CONUS404 approach also provides high fidelity of spatiotemporal distribution of precipitation and temperature by better resolving fine-scale weather events such as summer MCSs and winter orographic precipitation. Such success was largely attributed to the improved WRF physics parameterization schemes (e.g., microphysics and groundwater) in the last decade and the use of the convection-permitting modeling approach. CONUS404 provides dynamically and spatiotemporally consistent hydrologic forcing conditions. In addition, it provides the researchers and stakeholders with high-resolution, long-term datasets at scales relevant to address climate, hydrology, and environmental issues.
An evaluation of precipitation, temperature, and downward solar radiation reveals some of the strengths and weaknesses of this dataset. The strengths are the depiction of precipitation and temperature at the mesoscale (hourly and longer and scales of 4 km and larger). Weaknesses are that the downward solar radiation seems to be overestimated at the surface on the order of 5% and that snowpack is underestimated by ∼15%. The radiation bias seems to be related to the inadequate representation of nonprecipitating clouds in WRF. Both biases may be the subject of future research. Despite these caveats, this dataset provides an estimate of the three-dimensional structure of the atmosphere for the past 40-plus years at the mesoscale that is dynamically balanced and available to the community.
Acknowledgments.
This research was supported by the U.S. Geological Survey (USGS) Water Mission Area’s Integrated Water Prediction science program (Grant 140G0121F0357) and the NCAR Water System Program. NCAR is a major facility sponsored by the National Science Foundation (NSF) under Cooperative Agreement 1852977. Thanks are given to the three anonymous reviewers and Steven Markstrom and John Hammond at USGS, whose comments greatly improved the paper. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Data availability statement.
The CONUS404 data are archived and accessible to the public at https://doi.org/10.5066/P9PHPK4F (Rasmussen et al. 2023).
Appendix: Tables A1–A5: List of CONUS404 data output variables
Appendix Tables A1–A5 provide detailed information about the CONUS404 output variables.
List of hourly CONUS404 variables archived on USGS Black Pearl tape drive system. Bucket variables are accumulated over the full model time period (42 years). Instantaneous means at the model output time. Highlighted variables (in red font) in the first column are common forcing variables to drive land surface and hydrologic models (Chen et al. 2007).
Postprocessed hourly 2D variables derived from the model output.
Additional hourly postprocessed 3D data in the CONUS404 dataset.
Daily climatology data files: daily maximum, minimum, and mean values.
Fifteen-minute increment output data in the CONUS404 dataset.
References
Abatzoglou, J. T., 2013: Development of gridded surface meteorological data for ecological applications and modelling. Int. J. Climatol., 33, 121–131, https://doi.org/10.1002/joc.3413.
Ballinger, T. J., G. W. K. Moore, Y. Garcia-Quintana, P. G. Myers, A. A. Imrit, D. Topál, and W. N. Meier, 2022: Abrupt northern Baffin Bay autumn warming and sea‐ice loss since the turn of the twenty‐first century. Geophys. Res. Lett., 49, 1944–8007, https://doi.org/10.1029/2022GL101472.
Barlage, M., F. Chen, R. Rasmussen, Z. Zhang, and G. Miguez‐Macho, 2021: The importance of scale‐dependent groundwater processes in land‐atmosphere interactions over the central United States. Geophys. Res. Lett., 48, e2020GL092171, https://doi.org/10.1029/2020GL092171.
Chen, F., and Coauthors, 2007: Description and evaluation of the characteristics of the NCAR High-Resolution Land Data Assimilation System. J. Appl. Meteor. Climatol., 46, 694–713, https://doi.org/10.1175/JAM2463.1.
Daly, C., G. H. Taylor, and W. P. Gibson, 1997: The PRISM approach to mapping precipitation and temperature. Proc. 10th Conf. on Applied Climatology, Reno, NV, Amer. Meteor Soc., 20–23.
Feng, Z., and Coauthors, 2021: A global high‐resolution mesoscale convective system database using satellite‐derived cloud tops, surface precipitation, and tracking. J. Geophys. Res. Atmos., 126, e2020JD034202, https://doi.org/10.1029/2020JD034202.
Goudriaan, J., 1977: Crop micrometeorology: A simulation study. Ph.D. dissertation, Wageningen University, 262 pp.
Han, J., and H.-L. Pan, 2011: Revision of convection and vertical diffusion schemes in the NCEP Global Forecast System. Wea. Forecasting, 26, 520–533, https://doi.org/10.1175/WAF-D-10-05038.1.
Hari, V., S. Ghosh, W. Zhang, and R. Kumar, 2022: Strong influence of North Pacific Ocean variability on Indian summer heatwaves. Nat. Commun., 13, 5349, https://doi.org/10.1038/s41467-022-32942-5.
He, C., F. Chen, M. Barlage, C. Liu, A. Newman, W. Tang, K. Ikeda, and R. Rasmussen, 2019: Can convection‐permitting modeling provide decent precipitation for offline high‐resolution snowpack simulations over mountains? J. Geophys. Res. Atmos., 124, 12 631–12 654, https://doi.org/10.1029/2019JD030823.
He, C., F. Chen, R. Abolafia-Rosenzweig, K. Ikeda, C. Liu, and R. Rasmussen, 2021: What causes the unobserved early-spring snowpack ablation in convection-permitting WRF modeling over Utah mountains? J. Geophys. Res. Atmos., 126, e2021JD035284, https://doi.org/10.1029/2021JD035284.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hong, S.-Y., Y. Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Ikeda, K., and Coauthors, 2010: Simulation of seasonal snowfall over Colorado. Atmos. Res., 97, 462–477, https://doi.org/10.1016/j.atmosres.2010.04.010.
Ikeda, K., and Coauthors, 2021: Snowfall and snowpack in the western U.S. as captured by convection permitting climate simulations: Current climate and pseudo global warming future climate. Climate Dyn., 57, 2191–2215, https://doi.org/10.1007/s00382-021-05805-w.
IPCC, 2014: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, https://doi.org/10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–472, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kim, J.-H., P. A. Jimenez, M. Sengupta, J. Yang, J. Dudhia, S. Alessandrini, and Y. Xie, 2022: The WRF-solar ensemble prediction system to provide solar irradiance probabilistic forecasts. IEEE J. Photovoltaics, 12, 141–144, https://doi.org/10.1109/JPHOTOV.2021.3117904.
Knapp, K. R., and Coauthors, 2011: Globally gridded satellite observations for climate studies. Bull. Amer. Meteor. Soc., 92, 893–907, https://doi.org/10.1175/2011BAMS3039.1.
Kobayashi, S., and Coauthors, 2015: The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5–48, https://doi.org/10.2151/jmsj.2015-001.
Liu, C., K. Ikeda, G. Thompson, R. Rasmussen, and J. Dudhia, 2011: High-resolution simulations of wintertime precipitation in the Colorado headwaters region: Sensitivity to physics parameterizations. Mon. Wea. Rev., 139, 3533–3553, https://doi.org/10.1175/MWR-D-11-00009.1.
Liu, C., and Coauthors, 2017: Continental-scale convection-permitting modeling of the current and future climate of North America. Climate Dyn., 49, 71–95, https://doi.org/10.1007/s00382-016-3327-9.
Livneh, B., E. A. Rosenberg, C. Lin, B. Nijssen, V. Mishra, K. M. Andreadis, E. P. Maurer, and D. P. Lettenmaier, 2013: A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States: Update and extensions. J. Climate, 26, 9384–9392, https://doi.org/10.1175/JCLI-D-12-00508.1.
Lo, J. C.-F., Z.-L. Yang, and R. A. Pielke Sr., 2008: Assessment of three dynamical climate downscaling methods using the Weather Research and Forecasting (WRF) Model. J. Geophys. Res., 113, D09112, https://doi.org/10.1029/2007JD009216.
Maddox, R. A., 1980: Mesoscale convective complexes. Bull. Amer. Meteor. Soc., 61, 1374–1387, https://doi.org/10.1175/1520-0477(1980)061<1374:MCC>2.0.CO;2.
Miguez-Macho, G., G. L. Stenchikov, and A. Robock, 2004: Spectral nudging to eliminate the effects of domain position and geometry in regional climate model simulations. J. Geophys. Res., 109, D13104, https://doi.org/10.1029/2003JD004495.
Miguez-Macho, G., Y. Fan, C. P. Weaver, R. Walko, and A. Robock, 2007: Incorporating water table dynamics in climate modeling: 2. Formulation, validation, and soil moisture simulation. J. Geophys. Res., 112, D13108, https://doi.org/10.1029/2006JD008112.
Mooney, P. A., C. Broderick, C. L. Bruyère, F. J. Mulligan, and A. F. Prein, 2017: Clustering of observed diurnal cycles of precipitation over the United States for evaluation of a WRF multiphysics regional climate ensemble. J. Climate, 30, 9267–9286, https://doi.org/10.1175/JCLI-D-16-0851.1.
Newman, A. J., and Coauthors, 2015: Gridded ensemble precipitation and temperature estimates for the contiguous United States. J. Hydrometeor., 16, 2481–2500, https://doi.org/10.1175/JHM-D-15-0026.1.
Niu, G.-Y., and Z.-L. Yang, 2004: The effects of vegetation canopy processes on snow surface energy and mass balances. J. Geophys. Res., 109, D23111, https://doi.org/10.1029/2004JD004884.
Niu, G.-Y., and Coauthors, 2011: The community Noah land surface model with multi-parameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. J. Geophys. Res., 116, D12109, https://doi.org/10.1029/2010JD015139.
Oyler, J. W., S. Z. Dobrowski, A. P. Ballantyne, A. E. Klene, and S. W. Running, 2015: Artificial amplification of warming trends across the mountains of the western United States. Geophys. Res. Lett., 42, 153–161, https://doi.org/10.1002/2014GL062803.
Pascolini-Campbell, M., J. T. Reager, H. A. Chandanpurkar, and M. Rodell, 2021: Retracted article: A 10 percent increase in global land evapotranspiration from 2003 to 2019. Nature, 593, 543–547, https://doi.org/10.1038/s41586-021-03503-5.
Prein, A. F., and Coauthors, 2015: A review on regional convection‐permitting climate modeling: Demonstrations, prospects, and challenges. Rev. Geophys., 53, 323–361, https://doi.org/10.1002/2014RG000475.
Prein, A. F., R. M. Rasmussen, K. Ikeda, C. Liu, M. P. Clark, and G. J. Holland, 2017: The future intensification of hourly precipitation extremes. Nat. Climate Change, 7, 48–52, https://doi.org/10.1038/nclimate3168.
Prein, A. F., C. Liu, K. Ikeda, R. Bullock, R. M. Rasmussen, G. J. Holland, and M. Clark, 2020: Simulating North American mesoscale convective systems with a convection-permitting climate model. Climate Dyn., 55, 95–110, https://doi.org/10.1007/s00382-017-3993-2.
Prein, A. F., R. M. Rasmussen, D. Wang, and S. E. Giangrande, 2021: Sensitivity of organized convective storms to model grid spacing in current and future climates. Philos. Trans. Roy. Soc., A379, 20190546, https://doi.org/10.1098/rsta.2019.0546.
PRISM Climate Group, 2014: PRISM climate data. Oregon State University, accessed 16 December 2020, https://prism.oregonstate.edu.
Rasmussen, R. M., and Coauthors, 2011: High-resolution coupled climate runoff simulations of seasonal snowfall over Colorado: A process study of current and warmer climate. J. Climate, 24, 3015–3048, https://doi.org/10.1175/2010JCLI3985.1.
Rasmussen, R. M., and Coauthors, 2012: How well are we measuring snow? The NOAA/FAA/NCAR winter precipitation test bed. Bull. Amer. Meteor. Soc., 93, 811–829, https://doi.org/10.1175/BAMS-D-11-00052.1.
Rasmussen, R. M., and Coauthors, 2014: Climate change impacts on the water balance of the Colorado headwaters: High-resolution regional climate model simulations. J. Hydrometeor., 15, 1091–1116, https://doi.org/10.1175/JHM-D-13-0118.1.
Rasmussen, R. M., and Coauthors, 2018: Evaluation of the Wyoming Weather Modification Pilot Project (WWMPP) using two approaches: Traditional statistics and ensemble modeling. J. Appl. Meteor. Climatol., 57, 2639–2660, https://doi.org/10.1175/JAMC-D-17-0335.1.
Rasmussen, R. M., and Coauthors, 2023: Four-kilometer long-term regional hydroclimate reanalysis over the conterminous United States (CONUS), 1979–2020. U.S. Geological Survey, accessed 21 February 2023, https://doi.org/10.5066/P9PHPK4F.
Rupp, D. E., C. Daly, M. K. Doggett, J. I. Smith, and B. Steinberg, 2022: Mapping an observation-based global solar irradiance climatology across the conterminous United States. J. Appl. Meteor. Climatol., 61, 857–876, https://doi.org/10.1175/JAMC-D-21-0236.1.
Scaff, L., A. F. Prein, Y. Li, C. Liu, R. Rasmussen, and K. Ikeda, 2020: Simulating the convective precipitation diurnal cycle in North America’s current and future climate. Climate Dyn., 55, 369–382, https://doi.org/10.1007/s00382-019-04754-9.
Sengupta, M., Y. Xie, A. Lopez, A. Habte, G. Maclaurin, and J. Shelby, 2018: The National Solar Radiation Data Base (NSRDB). Renewable Sustainable Energy Rev., 89, 51–60, https://doi.org/10.1016/j.rser.2018.03.003.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.
Stephan, C. C., C. Strube, D. Klocke, M. Ern, L. Hoffmann, P. Preusse, and H. Schmidt, 2019: Intercomparison of gravity waves in global convection-permitting models. J. Atmos. Sci., 76, 2739–2759, https://doi.org/10.1175/JAS-D-19-0040.1.
Strachan, S., and C. Daly, 2017: Testing the daily PRISM air temperature model on semiarid mountain slopes. J. Geophys. Res. Atmos., 122, 5697–5715, https://doi.org/10.1002/2016JD025920.
Tegen, I., P. Hollrig, M. Chin, I. Fung, D. Jacob, and J. Penner, 1997: Contribution of different aerosol species to the global aerosol extinction optical thickness: Estimates from model results. J. Geophys. Res., 102, 23 895–23 915, https://doi.org/10.1029/97JD01864.
Thompson, G., and T. Eidhammer, 2014: A study of aerosol impacts on clouds and precipitation development in a large winter cyclone. J. Atmos. Sci., 71, 3636–3658, https://doi.org/10.1175/JAS-D-13-0305.1.
Thornton, P. E., M. M. Thornton, B. W. Mayer, Y. Wei, R. Devarakonda, R. S. Vose, and R. B. Cook, 2016: Daymet: Daily surface weather data on a 1-km grid for North America, version 3. Oak Ridge National Laboratory Distributed Active Archive Center, accessed 15 July 2016, https://doi.org/10.3334/ORNLDAAC/1328.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800, https://doi.org/10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
Trenberth, K. E., A. Dai, R. M. Rasmussen, and D. B. Parsons, 2003: The changing character of precipitation. Bull. Amer. Meteor. Soc., 84, 1205–1217, https://doi.org/10.1175/BAMS-84-9-1205.
Uppala, S. M., and Coauthors, 2005: The ERA‐40 re‐analysis. Quart. J. Roy. Meteor. Soc., 131, 2961–3012, https://doi.org/10.1256/qj.04.176.
Wu, X., and J. Yan, 2011: Estimating the outgoing longwave radiation from the FY-3B satellite visible infrared radiometer channel 5 radiance observations. Chin. Sci. Bull., 56, 3480–3485, https://doi.org/10.1007/s11434-011-4686-6.
Xu, Z.-F., and Z.-L. Yang, 2015: A new dynamical downscaling approach with GCM bias corrections and spectral nudging. J. Geophys. Res. Atmos., 120, 3063–3084, https://doi.org/10.1002/2014JD022958.