Data gathering is a compulsory process, which allowed us to know from where we were starting from and which additional data we needed. This step correspond with the A1 action of the FORECCAsT project.
One of the significant challenges of the project was to find a way to characterize as precisely as possible the tree dieback of the largest possible number of tree species using the lowest possible number of climate variables. To do so, we relied on previous studies carried on several oak species and linking the climatic water balance (P-ETP), the temperature during summer and tree dieback. Other studies on different tree species and in different locations have also shown the impact of these same variables on the description of climatic niches with the BioClimSol model. In other words, these variables allow to evaluate if the species are in or out of its climatic niche and if their risk of tree dieback increase.These variables have been detailed into six climate variables:
These variables are the six ones used in the BioClimSol method, the method we chose to use in our application ( see step 1). We also asked a group of students from AgroParisTech Nancy to evaluate the relevance of these 6 climate variables and to test some new ones. They approved these six variables and suggested others such as the precipitation seasonality or the temperature seasonality. Further investigation is now required to validate their suggestions.
The climate database we used to obtain the six previously cited variables in order to describe the past climate evolution is the AURITALIS model, which is developed by Jean Lemaire. AURITALIS is a hybrid model created by combining data from the AURELHY model (Météo France) and from the DIGITALIS model (AgroParisTech-Lerfob). The aim of this combination is to benefit simultaneously from the precision of the temperature and precipitation data from Météo France and from the precision of the solar radiation of DIGITALIS.
From thirty-year average (1981 to 2010) taken from these databases, we created one raster per climate variable at the scale of the French territory. These six rasters are the one we used for our model creations ( see step 5).
Example of raster for the TMAN (average annual temperature) variable. The warmer the color, the warmer the temperature.
In order to know which model to use for the prediction of future values for these climate variables, we tested and compare several models from Météo France produced through the DRIAS project. Over the 13 different models available on the DRIAS website, we chose for now to use the CNRM because it is the only one to provide values of solar radiation which are necessary to calculate the climatic water balance.
The BioClimSol method we use in our mobile application will include a reforestation module ( see step 1). To calculate the climatic thresholds of the ecological niches of the tree species included in this module, we cross-referenced their natural distribution area and a world climate database.
We used three different data sources to determine the natural distribution areas. The most important one was the European forest genetic resources programme ( EUFORGEN), which we completed from the work of Caudullo et al. (2017) and Little (1971).
After a comparative analysis of four databases, we chose WorldClim2 as our world climate database.
To use the "FORECCAsT by BioClimSol" mobile application, the user must be able to indicate very precisely its localisation on a digital map. After a careful comparison of three background maps, we decided to use OpenStreetMap, which is both precise enough and easy to include in our project.
To extract the topographic data needed by our models, we had to chose a DEM that was both precise enough and adapted to our project. After comparing four DEMs with a reference one, we decided to use the SRTM DEM, developped by the NASA at a 30 meters step.
In order to have a better understanding of Haut-Languedoc forest soils and to use this information in our mobile application, we set an in situ sampling campaign of 100 soil pits scattered in the Haut-Languedoc territory (see step 3). To ensure that these soil pits are representative of the soil diversity in our study area, we combined geological, soil, climate and topographic data to divide the territory in three pedo-climatic regions:
Division of the PNRHL territory in 3 pedo-climatic regions. Orange: western region. Blue: central region. Purple: eastern region. Green: distribution of the douglas fir (reference species for this work) in the area.
For each region, we scattered our soil pits so that they took place in topographic conditions as varied as possible, notably concerning slope and exposition.
We collected the data available concerning Haut-Languedoc Regional Nature Park Natura 2000 areas and the existing Habitats of Community Interest (HCI) cartography layers.
Example of the natural habitat cartography for a Natura 2000 site ("Le Caroux et l'Espinouse")
From this data, we determined that the most frequent HCI in our study area was the "Atlantic acidophilous beech forests with Ilex and sometimes also Taxus in the shrublayer" habitat. The occurence of this habitat istrongly depends on climate and it is linked to wood production stakes. For these reasons, we decided to focus on this HCI for out study of natural forest habitats.
Subscribe to our international newsletter about forest and climate change! The goal of this newsletter is to enable a better diffusion of all information concerning these topics, and to help networking activities between projects. News from field projects, trending publications, events to come... it covers all these subjects and is published every three months.