These directories contain the input and output data related to the
two-sided Partial Radiative Perturbation (PRP) decomposition in the
time domain described in Clark et al. (2021).

Results of the forward PRP calculations are contained in

      /RRTM_OUTPUT/PRP_FORWARD/~variable~/~variable~.YYYY.nc

Results of the backward PRP calculations are contained in

      /RRTM_OUTPUT/PRP_BACKWARD/~variable~/~variable~.YYYY.nc

where ~variable~ refers to either

"TOTAL", "TEMP", "H2O", "CO2", "CH4", "N2O", "CFC11", or "CFC12".
(Note that the "TOTAL" results are contained only in the PRP_FORWARD
directory.) Both backward and forward files are stored for the years
from 1984 through 2017, with each year in a different file indicated 
by the four-digit number YYYY. Each file contains surface downwelling
longwave irradiance and top-of-atmosphere (TOA) upwelling longwave
irradiance computations. We provide the TOA upwelling longwave irradiance
computations for completeness, and they were not examined in Clark et
al. (2021). In Clark et al. (2021), the PRP calculations were for the
surface, and there are no TOA PRP calculations that would examine the
impact of skin temperatures on TOA Irradiance. The TEMP PRP calculations
of Clark et al. (2021) examined air temperatures only.

The directory

PRP_FORWARD/TOTAL

contains results of the full surface downwelling longwave irradiance (SDLI):

      RRTMG(T,H2O,CO2,CH4,N2O,CFC-11,CFC-12),

where each RRTMG input is what was observed and contains all of the variability
therein.

The directory

PRP_FORWARD/TEMP

contains results of the forward PRP calculations for temperature:

      RRTMG(T_bar,H2O,CO2,CH4,N2O,CFC-11,CFC-12),

where each RRTMG input is what was observed, EXCEPT for temperature which is
fixed to its seasonal cycle.

The directory

PRP_BACKWARD/TEMP

contains results of the backward PRP calculations for temperature:

      RRTMG(T,H2O_bar,CO2_bar,CH4_bar,N2O_bar,CFC-11_bar,CFC-12_bar),

for which ONLY temperature is what is observed and everything else is fixed
to a seasonal cycle.

The other files in the PRP_FORWARD and PRP_BACKWARD directories contain analogous
results to TEMP but for the other variables. 

The directory RRTM_INPUT contains the files input into the RRTM calculations:

      seasonal_cycles.nc - The seasonal cycles that we used for each input.

NOAA gases can be located following the links included in the Acknowledgements
section of Clark et al. (2021). ERA-Interim provides the remaining inputs required 
to run RRTMG globally as was done in Clark et al. (2021). If your aim is to run 
RRTMG with ERA-Interim data, you will need to calculate the pressures following 
Berrisford et al. (2011) on both full and half-levels. Using the pressures on both
full and half-levels, you will also need to interpolate the ERA-I data to 
half-levels, as RRTMG expects both half and full level inputs for several variables.

TIPS FOR EXAMINING THE DATA:

To get the contributions to the SDLI anomalies by temperature, for example, using
the forward PRP as in Clark et al. (2021), compute the difference

      PRP_FORWARD/TEMP/TEMP.YYYY.nc - CLIMATOLOGY

where CLIMATOLOGY is defined as the seasonal cycle. A script for doing so would look
something like this:

      (preallocate TEMP_PRP_FORWARD as [NYEARS,NHOURS,NLAT,NLON])

      do yr = 1984,2017
            TEMP_PRP_FORWARD(yr - 1984,:,:,:) = (retrieve data from "/PRP_FORWARD/TEMP." + yr + ".nc")
      end do

      CLIMATOLOGY = (average year dimension of TEMP_PRP_FORWARD;
                     smoothing the NHOUR dimension is optional)

      do yr = 1984,2017
            # compute PRP
            TEMP_PRP_FORWARD(yr,:,:,:) = TEMP_PRP_FORWARD(yr - 1984,:,:,:) - CLIMATOLOGY
      end do 	

Some studies define the climatology as radiative transfer outputs computed with
climatological inputs. We used the climatology from the corresponding PRP calculations.

To get the contribution to SDLI anomalies by temperature, for example, using the
backward PRP as in Clark et al. (2021), compute the difference

      TOTAL/TOTAL.YYYY.nc - PRP_BACKWARD/TEMP/TEMP.YYYY.nc - CLIMATOLOGY

A script for doing so would be something like this:

      (preallocate TEMP_PRP_BACKWARD as [NYEARS,NHOURS,NLAT,NLON])

      do yr = 1984,2017
            TEMP = (retrieve data from "/PRP_BACKWARD/TEMP." + yr + ".nc")
            TOTAL = (retrieve data from "/PRP_FORWARD/TOTAL." + yr + ".nc")
            TEMP_PRP_BACKWARD(yr - 1984,:,:,:) = TOTAL - TEMP
      end do

      CLIMATOLOGY = (average year dimension of TEMP_PRP_BACKWARD;
                     smoothing the NHOUR dimension is optional)

      do yr = 1984,2017
            # compute PRP
            TEMP_PRP_BACKWARD(yr,:,:,:) = TEMP_PRP_BACKWARD(yr - 1984,:,:,:) - CLIMATOLOGY
      end do 	

To get the contribution to SDLI anomalies by temperature, for example, using the
two-sided PRP as in Clark et al. (2021), compute

      TEMP_PRP_TWOSIDED = (TEMP_PRP_FORWARD + TEMP_PRP_BACKWARD) / 2.0

Clark et al. (2021) averaged these results at a monthly timescale, but maintaining
6-hourly variability is possible.

For inquiries, please contact

      Joseph P. Clark
      jc7377@princeton.edu
      Joseph.P.Clark@noaa.gov

References:

      Berrisford, P., Dee, D., Fielding, K., Fuentes, M., Kallberg, P., Kobayashi, S.,
         & Uppala, S. (2009). The ERA Interim archive: Version 1.0. ERA Report Series, 1, 1-16.

      Clark, J.P., E.E. Clothiaux, S.B. Feldstein, and S. Lee 2021: Drivers of Clear
         Sky Global Surface Downwelling Longwave Irradiance Trends from 1984 through 2017.
         Geophys. Res. Lett.