Downscaling of ECMWF seasonal forecast in the tropical region Central Sulawesi, Indonesia, using the climate limited area model, CLM of the German Weather Service

Principal Investigator

Prof. Dr. Gode Gravenhorst
Institute of Bioclimatology, University of Göttingen, Germany
Büsgenweg 2
D-37077 Göttingen, Germany

ggraven@gwdg.de

Project description

Introduction

Implementation of CLM model as proposed in Special Project for 2005 and the following computer runs will deliver the model climatology as well as the hourly data which could be used for another study proposed in this Special Project for the period of 2006 to 2008. Running of 6 months EPS seasonal predictions will continue in year 2006 to get the updated predictions. Validation of seasonal predictions will be extended by comparing the model precipitation data with the satellite based precipitation estimate as well as with the recent data from ground based meteorological network. The DEKLIM and IMPENSO project will end in August 2006, but the modelling efforts will continue. Supplying the atmospheric parameters for hydrological models also continuous to the years 2006 - 2008. In years 2007 – 2008 the parameterization for main physical processes of CLM will be tested in tropical areas. In the domains, where the CLM originally is developed, mainly in mid latitudes, studies to compare fields and model results have been of more important for the LM Member States. Improved parameterization of physical processes occurring in tropical areas will enhance the model’s capability to simulate atmospheric parameters in broadly different conditions. A Soil-Vegetation-Atmosphere-Transfer model for a plot size scale (SVAT-Regio) developed by Institute of Bioclimatology, University of Göttingen (Oltchev, et.al 2003) simulates plant growth, CO2 fluxes and the water budget on an hourly time step. SVAT-Regio needs atmospheric data to drive the model. On off-line way, the CLM output of hourly simulated atmospheric parameters will be supplied for input of the SVAT-Regio model. Having some experiences using the output of a CLM model, a coupling study of CLM and SVAT-Regio model for land surfaces will be performed. In the year 2008 comparison studies with mesoscale atmospheric MM5 model of the University Corporation for Atmospheric Research and Penn State University (UCAR/PSU), will be carried out. This leads to a Multi-Model study for seasonal predictions of ENSO effects in Central Sulawesi and other Indonesia region. To study model uncertainties because of uncertain initial conditions, the ensemble model of CLM will be used.

Physical parameterization studies

Cumulus parametritzation . For study of cumulus parameterization the experiment that has been implemented by Daniela and Sokol (2002) will be adopted. Enhancing the resolution of NWP model we have to consider the applicability of existing physical parameterizations especially for the cumulus parameterization. The LM-DWD and its climate model version CLM use the cumulus parameterization by Tiedtke (Doms and Schaettler, 1999; Tiedtke, 1989). It is possible to switch the parameterization off and to run the CLM only with explicit expressions of cloud and rain processes. Therefore, the experiment will compare the precipitation simulation produced at switching the cumulus parameterization scheme ON/OFF. Other cumulus parameterization schemes will tested.

Regional SVAT model

Model description and validation. The basic concept used in the SVAT-Regio model is a complex description of energy, Н2О-and CO2-exchange between a non-uniform land surface and the atmosphere in local and regional scales (Oltchev et al. 1996). Multi-layer representation of plant canopy and soil allows to dynamically couple the penetration of solar radiation within a plant canopy, the turbulent transfer of sensible heat, Н2О and CO2 between soil, canopy and the atmosphere, the canopy microclimate, and the plant and the unsaturated soil water balance with meteorological conditions of the atmospheric boundary layer, canopy architecture, soil morphology, and biological properties of over storey and under storey vegetation. These parameterizations enable us to predict and quantify rather realistically the response of energy, water and carbon budgets of a plant canopy on changed environmental conditions. Values of atmospheric parameters for input of the SVAT-Regio model are necessary for the mean daily or hourly data for air temperature, relative humidity, wind speed, precipitation amount, and global solar radiation (cloud amount).

Spatial resolution of SVAT-Regio model depends on size of the entire modelled area and on spatial heterogeneity of relief, land-use, vegetation and soil. The present model version uses regular grids with the cell number not exceeding 500 × 500 cells. Depending on study objectives and available computer recourses these initial dimensions can be either decreased or increased. Grid cell resolution can vary from 100 m × 100 m to 10 km × 10 km or more. The minimal size of selected grid cells, however, should be not smaller than 100 m × 100 m. The main reason of such restriction is the quality requirement for application of one-dimensional SVAT models. The horizontal size of each grid cell has to be at least several times larger than the maximal height of the grid vegetation. This restriction allows neglecting the boundary effects arising in the surface sub-layer and inside of the plant canopy at the borders between different land-use and vegetation types (e.g. edge of a forest, lake shore, etc.).

Planning

Application of Special Projects in the year 2006 to 2008 will be scheduled according to our work plan as follows:

Year 2006

• Continuing the runs for 6 months EPS seasonal prediction updated up to April 2006 for stakeholder workshop’s material.
• Comparison of seasonal prediction with model climatology and observations.
• Off-line feeding of hydrological model WASMOD-ETH in IMPENSO research group from CLM model output.

Year 2007,

• Physical parameterization studies including comparison of several schemes of convective, land surface and radiation parameterization.
• Comparison of CLM model output with MM5 and other model output.
• Multi-model studies for seasonal prediction using Bayesian Model Averaging.

Year 2008

• Ensemble CLM model studies.
• Off-line feeding of SVAT-Regio model developed in Institute of Bioclimatology, Göttingen, for vegetation-atmosphere water and CO2 flux studies.
• Coupling the SVAT-Regio model into the CLM model as one of the Land Surface Model (LSM) options.

The estimated annual use of computer resources of the High Performance Computing Facility for the period of 2006 to 2008; will be 100,000 units and data storage will be 200 Gbytes.

References

Daniela,R. and Z. Sokol. 2002 Results of numerical experiments with LM DWD - first attempts to verify precipitation forecast by radar-derived rainfall fields (1998 flood event) Working document of the WG2 COST717 , WG2 meeting Oberpfaffenhofen March 8-9, 2002 . Available at http://www.smhi.se/cost717/doc/WDF_02_200204_2.pdf

Doms, G. and U. Schaettler, 1999: The Nonhydrostatic. Limited-Area Model LM (Lokal-Modell) of DWD, Part 1: Scientific Documentation, (available from http://www.cosmo-model.org).

Oltchev, A. ; Ibrom, A. ; Kreilein, H. ; Falk, U. ; Grelle, T. ; Gravenhorst, G. 2003. Application of a regional SVAT model for simulation of energy and water fluxes in mountainous area of Central Sulawesi ( Indonesia). EGS - AGU - EUG Joint Assembly, Abstracts #1080.

Oltchev A., Constantin J., Gravenhorst G., Ibrom A., Heimann J., Schmidt J., Falk M, Morgenstern K., Richter I., Vygodskaya N. (1996) Application of a Six-Layer SVAT model for simulation of evapotranspiration and water uptake in a spruce forest. J. Phys. Chem. Earth, Vol. 21 (3): 195-199.

Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models, Mon. Wea. Rev., 117: 1779-1799

For more details, please refer to the latest progress report.

Additional information

Project started in 2005.