6533b856fe1ef96bd12b2cce

RESEARCH PRODUCT

Capability of a high resolution hydroclimatic modelling chain to simulate soil water deficit indices for Douglas-fir and common Beeches over Burgundy.

subject

description

During the 2003 drought and heat wave event, douglas-fir and common beech stands in Burgundy have been heavily affected, and presented symptoms of dieback and mortality. This event seems to be the first occurrence of expected climatic changes in the near future and questions their sustainability in Burgundy since their climate vulnerability is mainly due to the amplitude and accumulated water constraints exercised during their growing cycle. In the context of climate change and in order to provide information to forest managers who partly rely on a mapping of the climatic constraints until the end of this century, this work explores the ability of a high resolution hydroclimatic modelling chain, coupling the regional climate model WRF to the daily lumped water balance model Biljou© in order to simulate soil water deficit indices for these two species.The first part of this paper analyzes the capacity of WRF model to simulate each surface atmospheric variable used as input for the water balance computation. The analysis of model's ability to simulate these variables is based on (i) a direct and comparative approach between WRF simulated data and observations recorded by the Météo-France stations network and SAFRAN reanalyses across the whole region, over stations and forest stands, (ii) on an indirect approach using the potential evapotranspiration and soil water deficit index calculated by Biljou©. Results show a significant improvement upon the ERA-Interim data for each variable and a strong ability to produce reliable data at high resolution. However, the WRF capability to estimate a realist potential evapotranspiration, combined to the the low correlation between the average annual soil water deficit and radial growth indexes, show that the WRF deficiencies in simulating water deficit are mainly attributable to its precipitation biases.The second part proposes to apply a statistical post-correction to the WRF precipitation data. Although this method significantly improves the spatial distribution of precipitation, their seasonal and interannual variability and precipitation amounts, post-corrected data do not produce a water deficit index sufficiently close to those ones estimated from observations or SAFRAN reanalysis. Two new simulations explicitly solving convective processes and using a spectral nudging have shown that this deficiency is mainly attributable to the inability of the correction method to solve timing differences of the transient climate variability simulated by WRF.This work showed that two types of climate modeling errors occurring independently, are major issues for impact studies: (i) the timing of precipitations events ; (ii) the statistical distribution of daily precipitation. Combined together, they control the number of days crossing the 40% threshold of relative extractable water and indirectly the soil water deficit index intensity.