Dear Tan Phan, actually, there are just a view topics you have to consider: opposite to weather forecasts there are important long term processes in climate, e.g. soil-moisture, snowcover effects, and so on. So, you have to choose a proper land surface model and initialize it with most realistic data. If necessary, interpolate between the soil layers of your driving data to get the soil parameters at the layers you have in your land surface model. Schedule at least one extra year in your long term simulations to reduce spin up effects within in the soil. Another important point is the sea surface temperature. If you have real SST data, use it. But be careful with INTERPF. INTERPF should calculate daily averages of SST when it finds skin temperature (SKINTEMP), but instead it averages the complete simulation period(!) and you will get a constant sea surface temperature. Additionally, think about the goemetric dimension. Avoid cutting through high mountains. If you have a large domain with many grid points MM5 creates its own weather patterns, but is that what you want? (This leads to inconsistant lateral boundary conditions.) If you want to keep MM5 closely connected to your driving data (e.g. flat terrain), reduce domain sizes and use 3d grid nudging when you have enough disk space. All of this really depends on the characteristic atmospheric processes which you have in your geographic region and which MM5 should be able to generate. Within our project "Research for Climate Protection - Model Run Evaluation (reclip:more)" (http://systemsresearch.arcs.ac.at/projects/climate/), my colleagues and I had to perform several 1-year simulations with different domain settings and paramterizations to find out the most proper configuration. I strongly recommend such sensitivity studies bevor you start with a 10-year run. I hope that helps a little. Greetings, Heimo -- --------------------------------------------------------- Heimo Truhetz Wegener Center for Climate and Global Change (WegCenter) University of Graz, Austria phone: +43-316-380-8442 fax: +43-316-380-9830 eMail: heimo.truhetz@uni-graz.at web: http://www.wegcenter.at --------------------------------------------------------- Dear Tan, > Following your advices, I am trying to focus on LSM and SST input for > MM5. Firstly, I found that I can use OI-SST and also ERA40, for > example, from ICTP as boundary conditions. But for LSM, I don't know how > to process. As far as I can say, there is no difference in the way of ingestion of SST between "with" and "without" LSM. When you use REGRID and ERA40, you have to keep in mind that pregrid only handles regular lat-lon grids. Concerning OI-SST I have no experiences. From literature available via the MM5-homepage it is shown, that SST influences the simulations seriously. So, try to get most realistic SSTs - highly resolved in time as well. > For weather forecast using MM5, I use AVN data. So it is easy for me to > setup the model by changing script *.sh file during preprocessing (like, > setup terrain.deck, pregridder.sh, and so on). AVN data includes > atmospheric, snow cover, and soil moisture fields. In the case of using > ERA40 and OISST data for RegCM, the BATS-LSM moduls can processes > itself. I don't know if Nosh-LSM in MM5 can do it in the same way? Yes, NOAH-LSM simulates the soil parameters. But you need a good initialisation. In MM5 you can freely choose the depth of the soil layers. If you have different driving data, it does make sense to interpolate between these layers and keep MM5 always running with the same layer-depths. When you use REGRID and ERA40, you have to be careful with the lowest soil layer: the "level 2"-level in GRIB-files is represented as a 1 byte value and has therefore a maximum of 255, even if the layer is much deeper (e.g. 289 cm in the case of ERA40). So, you have to use "255" in your Vtable, otherwise pregid can't find this level. I hope, I was able to give some answers to your questions. Kindest, Heimo > Could > you please show me more detail about it. > I am looking forward to hearing from you. > With my best wishes. > > Tan > > > */Heimo Truhetz /* wrote: > > Dear Tan, > > yes of course, please ask me if you want. > > So, I am interested in your work as well. > Could you please tell me more about it? > > > Best wishes, > Heimo