% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@INPROCEEDINGS{Chen:910819,
author = {Chen, Shuying and Poll, Stefan and Goergen, Klaus and
Heinrichs, Heidi and Hendricks-Franssen, Harrie-Jan},
title = {{R}enewable {E}nergy {P}otential {A}nalysis {B}ased on
{H}igh-{R}esolution {R}egional {A}tmospheric {M}odeling over
{S}outhern {A}frica},
reportid = {FZJ-2022-04169},
year = {2022},
abstract = {Renewable Energy Potential Analysis Based on
High-Resolution Regional Atmospheric Modeling over Southern
AfricaS. Chen, S. Poll, K. Goergen, H. Heinrichs, H.-J.
Hendricks-Franssen A large part of the global population
without reliable access to electricity lives in Africa.
Here, renewable energy could be a sustainable, cost
efficient, and climate-friendly solution, especially given
the large unexplored wind and solar energy potentials across
the African continent. Reliable and highly resolved
information is needed to assess renewable energy sources
adequately. Most often weather data like MERRA2 or ERA5 are
used for the assessment of renewable energy sources,
sometimes combined with a simple spatial downscaling based
on the Global Wind or Solar Atlas neglecting surface and
vertical atmospheric physical laws. However, those
meteorological input datasets typically have a relatively
coarse spatial resolution (e.g., ERA5 reanalysis at about
30km). With the aim to provide more robust data at high
spatial resolution, we produce a prototypical
high-resolution dataset over southern Africa from dedicated
atmospheric simulations. Such results can serve in future
research studies to estimate renewable energy potentials
with a higher spatial precision compared to previous
studies. As a basis for our study, we use the ICOsahedral
Nonhydrostatic (ICON) Numerical Weather Prediction
(ICON-NWP) model in its Limited Area Mode (ICON-LAM), based
on a configuration used also by the German Weather Service
(DWD) for operational weather forecasting. The study domain
over southern Africa is chosen due to its known favorable
meteorological conditions for solar and wind energy.
ICON-LAM dynamically downscales the global deterministic
ICON-NWP forecasts dataset from a grid spacing of 13km to a
convection-permitting resolution of 3.3km, where deep
convection parameterization is switched off. The
high-resolution triangulated grid cells of the 3.3km domain
are exactly inscribed in the 13km global grid cells. This
southern Africa ICON-LAM implementation is novel and has not
been run before. Simulations cover the time span from 2017
to 2019 with contrasting meteorological conditions. To keep
the ICON-LAM close to the observed atmospheric state, which
is assimilated into the driving global ICON-NWP runs, the
model atmosphere is reinitialized every 5 days, with a
preceding spin-up of one day. The land surface and
subsurface are run transient. The simulated 10m wind speed,
surface solar irradiance, 2m air temperature, and
precipitation are validated by using satellite data,
composite products, and in situ observations from three
networks (SASSCAL, TAHMO, and NCEI). This is done both for
the coarser driving model, the ERA5 reanalysis as well as
our ICON-LAM setup. Here we show initial results pointing to
reliable ICON-LAM simulations. Spatio-temporal Mean Bais
(MB) of 10m wind speed is 1.24 m s-1 and $84\%$ of simulated
frequency distributions overlap more than $60\%$ area with
that of observations. Correlation coefficients (R) of
surface solar irradiance have been well captured with an
average value over 0.9, and the spatial mean MB is 22.84 W
m-1. Low bias of 2m air temperature exists with a spatial
mean MB of 0.28 ֯C. The precipitation bias increases from
West to East, which may relate to the prevailing
precipitation regimes. All the simulations were run on the
cluster partition of supercomputer JUWELS at the Jülich
supercomputer center. The whole simulation costs 219
re-initialization cycles, for each cycle, 15 nodes and three
wall clock hours have been allocated.},
month = {Sep},
date = {2022-09-29},
organization = {NIC Symposium, Jülich (Germany), 29
Sep 2022 - 30 Sep 2022},
cin = {IEK-3 / IBG-3},
cid = {I:(DE-Juel1)IEK-3-20101013 / I:(DE-Juel1)IBG-3-20101118},
pnm = {1111 - Effective System Transformation Pathways (POF4-111)
/ 1112 - Societally Feasible Transformation Pathways
(POF4-111)},
pid = {G:(DE-HGF)POF4-1111 / G:(DE-HGF)POF4-1112},
typ = {PUB:(DE-HGF)1},
url = {https://juser.fz-juelich.de/record/910819},
}