February 12, 2018 at 8:43 am #3512
We would like to downscale Temperature and Precipitation from a GCM.
Which are the most correlated weather patterns that we can use?
We were thinking about NDVI, soil moisture and land surface temperature.
Any suggestion is more than welcome!
PauFebruary 13, 2018 at 9:08 am #3557
Could you please specify your question so that you answer the items below?
What is the target you want to downscale to? E.g. point source station data, or to a finer grid.
It seems you want to do a downscaling that is tied to weather patterns. This is a topic that is only explored for certain regions and might not be globally applicable. For which region and for which period of the year are you targeting?
I am not sure what you intend by including NDVI, soil moisture and land surface temperature. What do you want to do with these variables?
Please clarify these questions and we’ll do our best to help you.
PeterFebruary 13, 2018 at 12:05 pm #3558
Thanks for your answers and your interest
We would like to downscale to finer grids as well as for in-situ stations. We understand that the only option for the stations case would be to downscale only in the specific positions where the stations are located. Therefore we might interpolate the stations to grids so that we can also apply a ‘downscaling to finer grids’ procedure.
Our aim is to introduce alternative historical high resolution products that might have a correlation with the parameters in the grids that we wish to downscale. We have put an eye into the ‘MACA’ technique (Multivariate Adaptive Constructed Analogs) where they suggest a multi-linear fitting between high resolution grids and the coarse climate models (the ones that we wish to downscale).
Our first thought was to integrate many kinds of observed variables and let the model automate the importance of the diverse inputs in the correlation model. However it will be wiser to carefully choose the most suitable variables that are indeed related with our target grids to be downscale.
We are wondering which variables are most appropriate to guide the downscaling for Precipitation and Temperature.
PauFebruary 14, 2018 at 10:03 am #3561
Thank you for the specification. I would suggest to stick with precipitation and temperature as these are the most studied and understood variables from the climate models. Soil moisture is mostly not so well modelled due to much simplified surface hydrology and water fluxes. Also the surface (interpreted as skin) temperature is sensitive to cloud and radiation in the model. The 2-metre temperature is better calibrated to observations.
You must be logged in to reply to this topic.