Correction of a 1 km daily rainfall dataset for modelling forest ecosystem processes in Italy. (17th March 2016)

• Record Type: Journal Article
• Authors: Fibbi, Luca; Chiesi, Marta; Moriondo, Marco; Bindi, Marco; Chirici, Gherardo; Papale, Dario; Gozzini, Bernardo; Maselli, Fabio
• Is Part Of: Meteorological applications. Volume 23:Number 2(2016)
• Page Start: 294
• Page End: 303
• Publication Date: 2016-03-17
• Subjects: meteorological dataset -- E‐OBS -- BIOME‐BGC -- GPP modelling; Meteorology -- Periodicals; Meteorological services -- Periodicals; 551.5
• DOI: 10.1002/met.1554
• Languages: English
• ISSNs: 1350-4827

Abstract

Abstract: The availability of daily meteorological data over extended wide areas is a common requirement to model ecosystem processes on a regional scale. The meteorological dataset E‐OBS (European Observations) has been downscaled in 2012 to a 1 km spatial resolution in Italy. The assessment of this dataset against ground measurements revealed a marked under‐estimation of rainfall, which is a major driving factor for vegetation growth in water‐limited Mediterranean environments. The objective of the present study is to correct the downscaled rainfall dataset through the use of additional ground meteorological data collected over a 30 year period (1981–2010). The corrected dataset was evaluated using the ecosystem process model BIOME‐BGC (BioGeochemical Cycles). After identifying the most under‐estimated areas, seven sites that were representative of different forest ecosystems were selected; for each of these sites, gross primary production (GPP) was simulated using the E‐OBS original and corrected rainfall datasets. The test was completed for one site in central Italy (IT‐Ro2), assessing the outputs of the BIOME‐BGC data versus the daily GPP data obtained by the eddy correlation technique. In all cases, the use of the corrected dataset produced more realistic GPP simulations; for IT‐Ro2, the mean bias error was 0.18 instead of 2.33 g C m −2 day −1 . The beneficial effect of the correction depended on the aridity of the examined site, that is, it was the maximum in the driest Mediterranean ecosystems and almost null in the most humid environments. It can therefore be concluded that the corrected dataset is suitable to drive the long‐term simulation of forest ecosystem processes in all Italian eco‐climatic conditions.