Atmospheric CO 2 modeling at the regional scale

Atmospheric CO 2 modeling at the regional scale: A bottom – up approach applied to the Carbo. Europe Regional Experiment campaign (CERES) Claire Sarrat, Joël Noilhan, Pierre Lacarrère, Sylvie Donier et al.

OUTLINE I II Objectives of CERES and meso scale modeling Atmospheric CO 2 modeling at the regional scale with Meso-NH Ø A ‘golden day’ case study: may-27 Ø A ‘lagrangian experiment’ case study : june-06 III Intercomparisons of atmospheric meso-scale models

Objectives of the CERES campaign (Dolman et al. , BAMS, 2006) I Objectives II Atmospheric CO 2 modeling III Intercomparisons of models Ø Objectives : to establish a regional budget of CO 2 : • 10 surface flux sites (energy, water and CO 2) on different types of land cover (forest, vineyards, maize, wheat, rapeseed, beans, grassland, bare soil) • Atmospheric Boundary Layer (ABL) data: RS, aircrafts, radar UHF… • CO 2 concentrations observations in and above the ABL: Biscarosse, La Cape Sud, Marmande + aircrafts: Piper Aztec, Dimona, Sky Arrow • ‘Flux divergence’ flights • LAI monitoring • Surface and soil properties (Ts, soil water content…) Ø Experiment in Les Landes, S-W of France: - from may-16 to june-25 2005 - 21 IOP days

Objectives of the modeling activity I Objectives II Atmospheric CO 2 modeling III Intercomparisons of models ØCO 2 regional budget using a meteorological meso-scale model Meso-NH and the CERES data: - Test the model ability to simulate the strong surface heterogeneities - Simulate the CO 2 transfers at the boundaries: surface – ABL and entrainment at the ABL top -Simulate the complex interactions of CO 2, heat and water surface fluxes within a regional model - Simulate correctly the concentrations in the PBL as a necessary condition to retrieve the surface fluxes by inverse modeling (see T. Lauvaux presentation) Meteorological Model Meso-NH Atmospheric [CO 2] Lafore et al. , 98 Surface ISBA-A-gs Noilhan et al. 89 Calvet et al. , 98 LE, H, Rn, W, Ts… CO 2 Fluxes Anthropogenic Sea

Atmospheric CO 2 modeling Meso-NH configuration I Objectives II Atmospheric CO 2 modeling III Intercomparisons of models Nesting 2 ways Land use: Ecoclimap (Masson et al. , 2003) Initialization and lateral boundaries forcing: ECMWF model Anthropogenic CO 2 emissions from Stuttgart Univ. at 10 km resolution Large domain : France Horizontal resolution: 10 km Altitude (m) 900 km Small domain: CERES domain Horizontal resolution: 2 km 320 km

Atmospheric CO 2 modeling may– 27 Sea breeze effects (Sarrat et al. , JGR, 2006) CO 2 concentrations (ppm) may-27 9 HUTC I Objectives II Atmospheric CO 2 modeling III Intercomparisons of models CO 2 concentrations (ppm) may-27 14 HUTC S-W Wind direction FOREST AREA AGRICUL. AREA S-W S-E CO 2 concentrations S-E AGRICUL. AREA FOREST AREA

Atmospheric CO 2 modelling may– 27 Boundary layer heterogeneity Simulated vertical cross section of the mixing ratio at 14 UTC obs model Zi = 1600 m Forest obs model OCEAN FOREST AREA AGRICUL. AREA Zi = 900 m Crops

Atmospheric CO 2 modelling : may– 27 A scheme of main processes

Atmospheric CO 2 modelling : june-06 Lagrangian experiment N-W I Objectives II Atmospheric CO 2 modeling III Intercomparisons of models

Atmospheric CO 2 modelling : june-06 Lagrangian experiment : Budget calculation N-W 15 UTC 6 UTC CO 2 variation CO 2 turb. flux CO 2 advection

Conclusion (1) : Atmospheric CO 2 modeling with Meso-NH ØThe CERES database is well adapted to study the CO 2 and water budget at the regional scale Ø The meso-scale dynamical processes such as sea and vegetation breezes have a strong impact on the spatial and temporal variability of CO 2 concentrations in the ABL ØThe atmospheric CO 2 budgeting using meso-scale modelling allows to estimate the contribution of advection and turbulent transport processes on the spatio-temporal variation of the regional CO 2 concentration

Intercomparison of 5 meteorological models I Objectives II Atmospheric CO 2 modeling III Intercomparisons of models Participation of 5 models: RAMS from Amsterdam Vrije Univ. , RAMS from Alterra, RAMS from CEAM, WRF from MPI, Meso-NH from CNRM Experimental Protocol agreed on: Ø Domain of simulation at 2 km resolution Ø Initialization and lateral boundaries forcing for meteorological and surface variables with ECMWF model Ø Land cover by the Ecoclimap database including 61 surface classes, summer crops/winter crops ØCO 2 anthropogenic emissions at 10 km resolution from Stuttgart Univ. Ø 2 golden days of the CERES campaign: may-27 and june-06 2005

Intercomparison of 5 meteorological models: Surface fluxes Auradé winter crop Ø Auradé winter crop site is well simulated by all the models may-27 Le Bray forest RN H Ø Simulations for Le Bray forest site more difficult for all models RN H LE LE Ø Bsimu [. 5, 2] Ø CO 2 flux overestimated due to too high respiration? SFCO 2

Intercomparison of 5 meteorological models: Atmospheric Boundary Layer Ø Most of the models simulate the nocturnal stable ABL and humidity accumulation at low level Ø At 14 H large variation for ABL development : -> 800 m RAMS-ALTE ->1500 m WRF-MPI day Z (m) night obs Potential temp RS june-06 05 H FOREST Potential temp RS june-06 14 H FOREST

Intercomparison of 5 meteorological models Vertical profiles of CO 2 concentrations (may-27) zi Forest zi Crops afternoon vertical profile morning vertical profile CO 2 concentrations ABL height vs CO 2 concentrations: Ø the CO 2 concentrations decrease when the ABL is developing due to vertical mixing and assimilation Ø the CO 2 depletion is higher over the crops area whereas the vertical mixing in lower than over the forest Ø Generally, the models reproduce the observed trend.

Conclusion (2) : Intercomparison of 5 regional meteorological models Ø 5 models have simulated two contrasted days of CERES according a similar model configuration ØThe surface fluxes are easier to simulate over fully developed crops than over the pine forest. The windy june-06 case is better simulated. Ø The surface CO 2 fluxes on the warm may-27 are poorly simulated by most models. Ø Large discrepancies are observed in the simulation of the ABL development and potential temperature ØThe CO 2 concentrations simulated in the ABL present a correct evolution between the morning and the afternoon profiles.

Atmospheric CO 2 modelling Conditions of simulation : Initialisation of CO 2 the day before the simulated day at 18 HUTC with a homogeneous vertical profile over the domain of simulations Meteorological and surface moisture initialisation, lateral boundaries forcing : ECMWF analyses CO 2 anthropogenic emissions from Stuttgart Univ. at 10 km Land use : Ecoclimap (Masson, 2003, Champeaux et al. , 2005) 62 classes of vegetation: Ecoclimap processed from CORINE 2000 and Vegetation NDVI. Anthropogenic emissions interpolated at 2 km

Intercomparison of 5 meteorological models Vertical profiles of CO 2 concentrations june-06 Sensitivity to initial conditions

Intercomparison of 5 meteorological models Aircraft fluxes Ø The observed aircraft fluxes over forest and crops present large horizontal variations BOWEN RATIO FOR CROP 1 . 7 MNH-CNRM 1. 5 . 7 RAMS-AMVU . 8 . 7 RAMS-ALTE 1. 7 1. 3 RAMS-CEAM 2. 3 June-06, 9 -11 UTC 1 OBS ØFor MNH-CNRM and RAMS-ALTE CO 2 fluxes look consistents Ø For MNH-CNRM the LE fluxes are overestimated over crops because of an overestimation of the LAI Forest Crops