Climatic data is an important component of ecological models. In B4EST we want to quantify the potential impacts of climate change on forests in Europe by understanding how different tree phenotypes (trees of different characteristics) are likely to perform under different projected future climates. To do this we will be reanalysing existing European growth trials of tree provenance (geographic seed origin) and progeny (genetic/family origin) using data on historic and projected future climate.
As part of our work we will need to deal with biases in climate projections as both global and regional climate models include some interpolation errors which need be known and understood. In B4EST, we will be quantifying these errors, or discrepancies, between different data sources so we can better understand the uncertainties in our ecological models.
A particular challenge to our task is that whilst a wide range of climatic data is now available from web portals, time series data (data indexed in time order) are rarely available at high resolution (5 km or less).
Frequently used time-series databases include:
- the Climate Research Unit (CRS) data TS.4.02 from the University of East Anglia, UK,
- the Copernicus climate data portal (EU), and
- the UKCP18 data from the Centre for Environmental Data Analysis (CEDA).
The data provided at these portals are too coarse for site sensitive scientific studies, such as those used in B4EST. For this reason, many groups have developed different downscaling methods, which increase the resolution of the data enabling it to be used at smaller spatial scales.
Climate data software ClimateWNA and the ClimateEU (https://sites.ualberta.ca/~ahamann/data.html) provide an interface in the form of a standalone MS Windows application, enabling users to extract and downscale PRISM 1961-1990 monthly normal climate data (1.25 x 1.25 arcmin, 4 km approximately) to scale-free point data based on latitude, longitude and elevation.
The climatic time series data availability for ClimateEU is unfortunately not suitable for the requirements of B4EST. In order to overcome this time limitation, as well as additionally replicating and controlling the downscaling process, B4EST (Work Package One) has followed the main steps developed by the ClimateEU team, to migrate the ClimateEU method into R programming language (for more detail, see the explanation below).
The testing procedure is currently under development. Once the statistical process has been defined, CRU time series, UKCP18 surfaces and other climatic surfaces with historical data and future projections will be made freely available to enable scientists to develop a range of climate data for ecological and genetic studies in a user-friendly R-shiny web app (https://shiny.rstudio.com/). In this way scientists working in B4EST will be able to access reliable climatic time series data to characterize the historic and future climate of their provenance trials and experiments.
Migrating the ClimateEU method into R programming language – the process
Using Worldclim ver 1.4 as a test climatic surface (representing the 1961-1990 climatic normal period) with 30 arc-second of spatial resolution and other data obtained from meteorological stations across Europe, a scale-free bi-linear interpolation process with a dynamic local lapse rate adjustment has been completed. Other spatial interpolation techniques such as inverse distance weighting have been tested to increase the range of interpolation methods with which to compare results.