|
CERES/ARM Validation Experiment (CAVE)
|
Here, one can plot CERES/SARB validation results.
Radiation transfer model runs at CERES footprint resolution (archived in the CERES CRS data product CRS - defined )
are compared to observed SW and LW fluxes at the surface and TOA. Footprints fall within 15km and 7 minutes
of surface observations. Surface SW & LW results are ploted along with tables of statistics as shown in the example below.
Here's how to decode the tables.
Some of the validation plots differ between Aqua and Terra satellite results for various "grouping" of the CAVE sites. Why is this the case?
Click here for an explanation.
LW Dn Sfc = Broadband longwave (thermal IR) downward at surface
LW Up Sfc = Broadband longwave (thermal IR) upward at surface
Observations are from pyrgeometers
SW Dn Sfc = Broadband shortwave (solar) downward at surface
Observations from pyrheliometer plus shaded pyranometer
SW Up Sfc = Broadband shortwave (solar) upward at surface
Observations from pyranometer
LW Up TOA = Outgoing Longwave Radiation (OLR)
SW Up TOA = Reflected Shortwave Flux
Observations from CERES broadband instruments processed
by the advanced SSF method (i.e., better than ES8 ERBE-like)
Obs Mean = Observed flux in Wm-2
Surface observations are 15 minute means. CERES observations
and SARB/CRS products are instantaneous. SW observations at
the surface have been adjusted to correspond to the solar
zenith angle (SZA) of the satellite observation.
a = cos(SZAsatellite)
b = cos(SZAeachminute)
c = 15 minute mean of b
adjusted sfc obs flux = (sfc obs flux)*a/c
LW was computed and observed for day and night footprints.
SW was computed and observed for daytime footprints only.
To estimate a 24 hour mean SW flux or bias from the tables,
divide each SW flux or bias by 2.
N = number of samples
Bias = SARB/CRS product minus Observations
Std Dev = standard deviation
RMS = root mean square difference of SARB/CRS product and observations
Mod Frc All-Clr = Model (SARB/CRS product) cloud forcing as difference of
computed all sky flux (available for all footprints)
and computed clear sky flux (available for all footprints)
Forcing All-CNA= Aerosol forcing as the difference of
computed all sky flux (including clouds & aerosols)
and computed cloudy sky flux (Includes clouds but NO aerosols)
Dif Bias Obs-CRS = Bias (Observations - SARB/CRS product)
of the diffuse downwelling SW (Tuned results only)
AOT Frc Clr-Prs= Aerosol forcing as the difference of
computed clear sky flux (including aerosols)
and computed pristine flux (clear sky with NO aerosols)
All Sky = all available CERES fooprints (commonly 15-40 km)
Clear Sky VIRS = CERES footprints identified as clear using
high resolution VIRS imager pixels (~1 km)
Clear Sky VIRS & L/AA = CERES footprints identified as clear using both
high resolution VIRS imager pixels (~1 km)
AND a minute-by-minute time series analysis
of surface pyranometer data as per the
Long and Ackerman algorithm
Overcast VIRS = CERES footprints identified as overcast using
high resolution VIRS imager pixels (~1 km)
Overcast VIRS & L/AA = CERES footprints identified as overcast using both
high resolution VIRS imager pixels (~1 km)
AND a minute-by-minute time series analysis
of surface pyranometer data as per the
Long and Ackerman algorithm
Accuracy
The following table shows desired surface flux accuracies from BSRN:
From Ohmura et al., 1998
| Quantity at Surface |
Capability |
| Direct solar irradiance |
2 Wm-2 |
| Diffuse solar radiation |
5 Wm-2 |
| Global (SW) radiation |
5 Wm-2 |
| Downwelling longwave radiation |
10 Wm-2 |
| Ohmura, A., E. Dutton, B. Forgan, C. Frohlich,
H. Gilgen, H. Hegne, A. Heimo, G. Konig-Langlo, B. McArthur,
G. Muller, R. Philipona, C. Whitlock, K. Dehne, and M. Wild,
1998: Baseline Surface Radiation Network (BSRN/WCRP): New precision
radiometery for climate change research. Bull. Amer. Meteor. Soc.
79 No. 10, 2115-2136. |
Accuracies specified by BSRN in the Longwave region are seen as overly
pessimistic as pointed out in
Charlock et al. 2001. This is also found in the statistics tables
above in model to observation comparisons.
|