Loading…
Bias correction of surface downwelling longwave and shortwave radiation for the EWEMBI dataset
Many meteorological forcing datasets include bias-corrected surface downwelling longwave and shortwave radiation (rlds and rsds). Methods used for such bias corrections range from multi-year monthly mean value scaling to quantile mapping at the daily timescale. An additional downscaling is necessary...
Saved in:
Published in: | Earth system dynamics 2018-05, Vol.9 (2), p.627-645 |
---|---|
Main Author: | |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Many meteorological forcing datasets include bias-corrected surface
downwelling longwave and shortwave radiation (rlds and rsds). Methods used
for such bias corrections range from multi-year monthly mean value scaling to
quantile mapping at the daily timescale. An additional downscaling is
necessary if the data to be corrected have a higher spatial resolution than
the observational data used to determine the biases. This was the case when
EartH2Observe (Calton et al., 2016) rlds and rsds were bias-corrected
using more coarsely resolved Surface Radiation Budget
(Stackhouse Jr. et al., 2011) data for the production of the meteorological
forcing dataset EWEMBI (Lange, 2016). This article systematically
compares various parametric quantile mapping methods designed specifically
for this purpose, including those used for the production of EWEMBI rlds and
rsds. The methods vary in the timescale at which they operate, in their way
of accounting for physical upper radiation limits, and in their approach to
bridging the spatial resolution gap between E2OBS and SRB. It is shown how
temporal and spatial variability deflation related to bilinear interpolation
and other deterministic downscaling approaches can be overcome by downscaling
the target statistics of quantile mapping from the SRB to the E2OBS grid such
that the sub-SRB-grid-scale spatial variability present in the original E2OBS
data is retained. Cross validations at the
daily and monthly timescales reveal that it is worthwhile to take empirical
estimates of physical upper limits into account when adjusting either
radiation component and that, overall, bias correction at the daily timescale
is more effective than bias correction at the monthly timescale if sampling
errors are taken into account. |
---|---|
ISSN: | 2190-4987 2190-4979 2190-4987 |
DOI: | 10.5194/esd-9-627-2018 |