Transcript S2-Colombo
Estimation of the Net Ecosystem Exchange in Agro-Forestry Ecosystems by Assimilation of Satellite-Derived Information in a Process Model R. Colombo(1), M. Meroni(1) , M. Migliavacca(1), L. Busetto(1-2), C. Panigada(1), G. Seufert(3) (1) Lab. Telerilevamento Dinamiche Ambientali DISAT-UNIMIB, Milano, Italy [email protected] www.disat.unimib.it/telerilevamento (2) CNR (3)Joint - IIA, Roma, Italia Research Centre, IES - Ispra (VA), Italia Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Main objectives Mapping carbon exchange in fast growing forests (poplar plantation) • ground estimation of RS parameters related to carbon; • prediction of canopy parameters at different RS scales; • recalibration strategy for modelling carbon exchange at regional level. Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Index of presentation 1. Experimental site overview; 2. NEE vs biophysical parameters; 3. Ground measurements techniques; 4. Canopy characteristics from remote sensing data; 5. Carbon mapping by coupling RT and BIOME-BGC models Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Experimental site • Type: I-214 clone; • Age: 14 years; • Tree density: 278 trees/ha; • Height: 27 m; • Area: 10.5 m2 ha-1 M ila n o Natural forest r ive or cin Ti After logging regenerating forest S tu d y S ite T ic in o R iv e r P a rk Poplar plantation Flux tower Rice-paddies • Exchange rates of sensible heat, water vapour and carbon dioxide are measured by the eddy covariance technique, since march 2002. • The site belongs to the CARBOEUROPE-IP network (managed by JRC-IES) Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy LTDA Canopy characteristics and NEE at the experimental site 1/2 Temporal trend of Fipar and NEE Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Canopy characteristics and NEE at the experimental site 2/2 2,50 1,50 2 2 LAI (m /m ) 2,00 Temporal trend of 1,00 LAI 0,50 0,00 30 60 90 120 150 180 Day of Year Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy and NEE Data collection Field data Remote Sensing Lab. data Pigment, water, dry matter content and N (PROSPECT) Hyperspectral airborne data LAI, fIPAR, SPAD, Fc, fluorescence Field radiometric measurements 0.6 0.5 R 0.4 0.3 0.2 0.1 25 0 0 24 00 23 00 2 20 0 2 10 0 2 00 0 1 90 0 18 0 0 17 00 1 6 00 1 5 00 1 40 0 1 30 0 1 20 0 90 0 11 0 0 800 10 00 700 6 00 5 00 40 0 0.0 w l (nm ) Spectroscopy and Leaf samples chemical extraction MODIS data Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy LAI measurements Field: canopy structural characterisation, LAI, fIPAR LAI - 2000 eSAIo ePAIo/u+o Digital Camera eLAIg Belowu+o - Belowo LAIu A:Belowo with B:Belowo+u Colombo et al., 2003 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy eLAIo fIPAR measurements Field: canopy structural characterisation, LAI, fIPAR, fAPAR vs fIPAR IA [W/m2] IA RA IB RB (IA - RA) - (IB - RB) / IA = Fapar IA- IB / IA = Fipar Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Relating canopy properties to RS observations Physically based models Reflected (6-12% of PAR) Field data VS hyperspectral data direct use 0 .9 0 .8 0 .7 0 .6 NDVI Canopy level (e.g. SAILH) PAR incoming 0 .5 - LAI, Fapar, Fc, rcl, tcl - Transmitted (0-40% of PAR) inversion Semi-empirical models Absorbed (48-94%) 0 .4 0 .3 2 r = 0 ,8 3 0 .2 0 .1 Heat (75-97%) 0 0 .0 0 0 .5 0 1 .0 0 1 .5 0 2 .0 0 2 .5 0 Fluorescence (3-5%) Photochemical (0-20%) 3 .0 0 eLAIo Leaf level (e.g. PROSPECT) Regression model between NDVI and LAI Colombo et al., 2004 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy - Cab, Cw, Cm, N; rfl, tfl - CR Model: direct use SAILH R canopy Canopy model PROSPECT Leaf R and T Leaf model Canopy structure Acquisition geometry Soil reflectance Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Chlorophyll Water Dry matter N 35 g cm-2 0.02 cm 0.01 g cm-2 1.3 Laef biochemistry CR Model: direct use SAILH R canopy Canopy model PROSPECT Leaf R and T Leaf model Canopy structure Acquisition geometry Soil reflectance Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Chlorophyll Water Dry matter N 35 g cm-2 0.02 cm 0.01 g cm-2 1.3 Laef biochemistry CR Model: direct use SAILH R canopy Canopy model PROSPECT Leaf R and T Leaf model Canopy structure Acquisition geometry Soil reflectance LAI MTA hot spot …. 3 m2m-2 50 deg 0.05 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Chlorophyll Water Dry matter N 35 g cm-2 0.02 cm 0.01 g cm-2 1.3 Laef biochemistry CR Model: direct use SAILH R canopy Canopy model PROSPECT Leaf R and T Leaf model Canopy structure Acquisition geometry Soil reflectance LAI MTA hot spot …. 3 m2m-2 50 deg 0.05 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Chlorophyll Water Dry matter N 35 g cm-2 0.02 cm 0.01 g cm-2 1.3 Laef biochemistry CR Model: inverse mode SAILH R canopy Canopy model PROSPECT Leaf R and T Leaf model Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Laef biochemistry CR Model: inverse mode SAILH R canopy Canopy model PROSPECT Leaf R and T Leaf model Canopy structure LAI MTA hot spot …. 3 m2m-2 50 deg 0.05 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Chlorophyll Water Dry matter N 35 g cm-2 0.02 cm 0.01 g cm-2 1.3 Laef biochemistry CR Model inversion: method The models parameters are iteratively “adjusted” until the modeled reflectance closely “resembles” the measured one modeled observed A sum of squares cost function may be the measure of this “resemblance” to be minimized: n dir C obs n spec meas R k 1 k , i , obs R k , i,mod 2 i 1 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Model output: biophysical maps Multitemporal LAI maps from MODIS reflectance LAI map from hyperspectral MIVIS data Meroni et al., 2004 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Carbon modeling: BIOME-BGC overview BIOME-BGC estimates fluxes and storage of energy, water, carbon, and nitrogen for vegetation and soil on a daily basis Major Features: • Daily time step (day/night partitioning based on daily information) • Single, uniform soil layer hydrology (bucket model) •1 canopy layer (sunlit/shaded leaf partitioning) • Dynamic phenology and C/N allocation (e.g. LAI, biomass, soil and litter) • Variable litter and soil C decomposition rates (3 litter and 4 soil C pools) Heinsch et al., 2001 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Biome-BGC: modified version COUPLING Biome-BGC WITH THE PROSAIL RADIATIVE TRANSFER MODEL NDVI • At daily time-step LAI values simulated by the BIOME-BGC model are used as input for the PROSAIL model • PROSAIL calculate daily shortwave, PAR transmitted and absorbed by the canopy and site shortwave albedo • APAR and shortwave absorbed are forced into Biome-BGC (radtrans routine) and then are divide into sunlit and shaded canopy portion. LAI Site characteristics and Inizialization data SW and PAR Albedo, Abs, Trans. BIOME - BGC Meteorological Data MODELED GPP input output/input output model Migliavacca et al., 2004 R a d ia t iv e T r a n f e r D e s c r ip t io n B io m e - B G C v 4 . 1 1 B io m e - B G C c o u p le d w it h P R O S A I L L a m b e r t B e e r d e s c r ip t io n A s s u m p t io n o f c o n s t a n t s it e a lb e d o P la t e m o d e l P R O S P E C T f o r t h e d e s c r ip t io n o f le a f o p t ic a l p r o p e r t ie s 1 D t u r b id m e d iu m S A I L 4 f lu x e s K - M th e o ry fo r N D V I S a m e m o d e l, 2 f lu x e s a p p r o x im a t io n f o r a lb e d o , t r a n s m is s io n a n d a b s o r p t io n Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy r NIR r R NDVI MODIS PROSAIL Biome Characteristics •Ecophysiological parameters •Phenology (onday, offday) r NIR r R Recalibration strategy RECALIBRATION OF THE MODIFIED BIOME-BGC MODEL An inverse modeling approach is used to recalibrate the input parameters of the PROSAIL-BGC model. The optimization technique is based on a Quasi-Newtonian algorithm that minimizes the sum of squared differences between observed and modeled data. • The recalibration strategy was based on a two steps approach: Inversion Step Scale Objective Observed Data Target Variables 1 Local Scale To determine the uncertain of some ecophysiological parameters Eddy Covariance Measurement s (GPP 2002) 2 Regional Scale To retrieve phenological parameters and site characteristics application at regional scale – Model NDVI MODIS 250 m Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy C:NLeaf Fraction of leaf N in RUBISCO (PLNR) Fine root to leaf carbon allocation ratio (FR:LC) Maximum stomatal conductance (gsmax) yearday to start new growth (onday) yearday to end litterfall (offday ) first -year maximum leaf carbon (LC ) Recalibration steps First – Step Recalibration OPTIMIZATION ALGORITHM - (Leaf C:N; PLNR, gsmax, FR:LC) An inversion against GPP measurements (year 2002) was applied in order to obtain a better ecophysiological parameterization PROSAIL Biome Characteristics •Ecophysiological parameters •Phenology (onday, offday) LAI Online framework for gap-filling and fluxpartitioning of eddy covariance flux data (Reichstein 2003) SW and PAR Albedo, Abs, Trans. Site characteristics and Inizialization data BIOME - BGC Meteorological Data Eddy Covariance Partitioning n obs f ( GPP i MOD GPP i OBS ) 2 i0 The new ecophysiological dataset will be use to estimate carbon budget of poplar plantations at regional scale COST FUNCTION EVALUATION MODELED GPP Second – Step Recalibration OPRIMIZATION ALGORITHM (On date ; Off date) PROSAIL Biome Characteristics •Ecophysiological parameters •Phenology (onday, offday) LAI SW and PAR Albedo, Abs, Trans. n obs f ( NDVI i0 MODIS NDVI PROSAIL ) 2 MODIS FILTERED NDVI TIME SERIES COST FUNCTION EVALUATION Method to estimate phenological parameters by comparing the NDVI values retrieved from MODIS data and the NDVI values estimated with the PROSAIL-BGC model Site characteristics and Inizialization data BIOME - BGC MODIS - TERRA Meteorological Data MODELED GPP Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy NDVIPROSAIL-BGC estimated for the same day and for the same sun-surface-geometry of MODIS observations. Assimilation of RS data Regional scale: deconvolution and filtering MODIS 250m 16-days composite NDVI data (Product MOD13Q1) Time period 2000-2003 Original vs Filtered MODIS NDVI 10000 9000 8000 NDVI (x10000) 7000 6000 5000 4000 3000 2000 Original MODIS NDVI 1000 FILTERED MODIS NDVI 0 1/ 1/ 00 1/ 7 /0 0 31 / 12 /0 1/ 7 0 /0 1 31 / 12 /0 1 1/ 7/ 02 31 / 12 /0 2 1/ 7 /0 3 31 / 12 /0 3 DOY Recursive application of a Savitzky-Golay filter MODIS 2003 NDVI [250x250m; 2 bands; 0.4-0.9m; daily; regional application] Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Preliminary results: 1st step local scale The new set of the eco-physiological parameters was then used to simulate annual GPP for the 2003 growing season, achieving a good improvement in GPP estimation. • Year Year GPP measured GPP measured -2year -1 gC m gC m year 2003 200 3 1473 1473 GPP before rec. GPP before _rec _ GPP after_rec _rec. GPPafter -2 year -1 gC gCmm year -2 year -1 gC gCmm year 1304 126 5 RMSE: EF: 1569 1569 Loague et al., 1991 Before rec.: 1.81 gC m-2 day-1 Before rec.: 0.77 After rec.: 1.41 gC m-2 day-1 After rec.: 0.86 Cumulative GPP 2002/2003 (error %): Prosail - Biome-BGC (before rec.) 12% Prosail - Biome-BGC (after rec.) 2% Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy 0.76 0.0006 Preliminary results: 2nd step for regional scale The recalibration against MODIS data was used in order to obtain an estimate of the start and the end of the growing season. • Simulation: Year EPC DATA: Recalibrated ecophysiological parameters START - END FIRST GUESS: Random Is the optimisation technique able to determine the corrects start and end of growing season? 2002 2003 * Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy S ta rt End O b s .* 91 267 E s t. 88 279 O b s .* 78 315 E s t. 70 310 Start and End of growing season were determined as the dates of NEE inversion. Ongoing: spectral measurements at the experimental site Long term continuous spectral measurements (2 spectra h-1): passive fluorescence with FLDP, r400-1100, VIs (e.g. CO2H2O,T sonic anemometer & IR gas analyzer inlet PRI, REP) at the eddy covariance site; Spectral system optics micromet sensors Discontinuous field campaign for spectral Eddy and Spectral system housing heterogeneity sampling. Gitelson et al., 2004 soil met sensors Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Conclusions A) NEE and Canopy characteristics • Temporal and spatial behavior of canopy characteristics (e.g. LAI and fIPAR) exerts a control in annual NEE; efforts in their measurements; B) RS and biophysical retrieval • Inversion of leaf & canopy reflectance models provide accurate estimation of LAI (MVA, LAI RMSE=.40 m2m-2); C) Carbon modeling • The assimilation methods can provide important informations for the application of the BIOME-BGC process models at regional scale; • Inverse modelling approach against eddy covariance measurements allow to determine the ecophysiological dataset that improve GPP estimation and to determine start and end of growing season Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy thank you Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Biome-BGC Modification – 2/4 Main features of PROSAIL Model (Jacquemoud) SAILH R canopy Canopy PROSPECT Leaf R and T Leaf Biochemical Leaf Model model Canopy Structure Sun-Surface-Sensor geometry Soil Reflectance LAI MTA …. ... 50 deg • Schematic description of fluxes within the canopy • Two diffuse fluxes (E+ upward – E- downard ) • E0 - Direct solar flux • Es - Flux in the direction of observation Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Clorophyll Water Thick. SLA N Biome-BGC modification COUPLING Biome-BGC WITH THE PROSAIL RADIATIVE TRANSFER MODEL Goals of modification • • To Improve Biome-BGC simulations at experimental site using a different description of canopy radiation regime • Radiative transfer regime describe using K-M approssimation • Shortwave and PAR albedo linked to phenological cycle To simulate NDVI MODIS for a given sun-surface-sensor geometry Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Biome-BGC Modification – 4/4 Comparison between Biome-BGC and PROSAIL-BGC Biome -BGC PROSAIL-BiomeBGC Observed GPP Year G P P m e a su re d gC m -2 year -1 G P P B io m e - B G C gC m -2 year -1 G P P P R O S A IL-B G C gC m -2 year 2002 1567 1330 1362 2003 1473 1265 1304 -1 Slightly improves the accuracy of GPP estimations (for both year). Cumulative GPP 2002/2003 (error %): Biome-BGC 15% Prosail - Biome-BGC 12% Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy G PP ( 0 .0 0 6 0 .0 0 4 Preliminary Results – 1/4 0 .0 0 2 First-Step Recalibration Results (against GPP) 0 .0 0 0 50 100 150 200 250 300 350 DOY 0 .0 1 6 0 .0 1 6 G PP 2 0 0 2 - O b serv ed GPP A fte r R e c = 1546 gCm -2 -2 year year -1 -1 G P P (K g C m 0 .0 0 8 0 .0 0 6 0 .0 0 4 0 .0 0 2 0 .0 0 0 50 100 150 200 250 A fte r re ca lib ra tion -1 year = 1327 gCm day = 1578 gCm B e fo r e R e c -2 0 .0 1 0 obs 0 .0 1 4 300 350 0 .0 1 2 -2 ) -1 GPP -1 M o d e le d G P P ( K g C m GPP -2 ) G P P 2 0 0 2 - M o d - B e fo r e R e c . 0 .0 1 2 day B e fore re ca lib ra tion G P P 2 0 0 2 - M o d - A fte r R e c . 0 .0 1 4 0 .0 1 0 0 .0 0 8 0 .0 0 6 0 .0 0 4 B efore-R ec. slope: 0.68 Intercept: 0.001 r ²: 0.88 0 .0 0 2 0 .0 0 0 0 .0 0 0 0 .0 0 2 0 .0 0 4 DOY 0 .0 0 6 0 .0 0 8 0 .0 1 0 O b se rv e d G P P (K g C m -2 A fter-R ec. slope: 0.83 Intercept: 0.0009 r ²: 0.93 0 .0 1 2 0 .0 1 4 -1 day ) 0 .0 1 6 B e fo re A fte r R e ca lib ra tio n R e ca lib ra tio n 0 .0 1 4 A ft e r re c a lib ra t io n Recalibrated ecophysiological parameters 0 .0 1 2 M o d e le d G P P (K g C m -2 day -1 ) B e fo re re c a lib ra t io n 0 .0 1 0 F R C :L C - 0 .3 3 3 Le a f C :N (K g C /K g N ) 1 5 .5 9 2 1 .6 6 PLN R g sM A X - 0 .0 8 8 0 .0 0 6 0 .1 0 5 4 0 .0 0 4 1 0 .0 0 8 0 .0 0 6 0 .0 0 4 0 .0 0 2 Workshop on Climatic Analysis and Mapping for Agriculture A fte r-R e c . 14-17 June 2005, Bologna, Italy B e fo re -R e c . slo p e : 0 .6 8 In te rc e p t: 0 .0 0 1 slo p e : 0 .8 3 In te rc e p t: 0 .0 0 0 9 m /s 2 .0 0 1 0 .0 1 6 Assimilation of RS data • MODIS 250m 16-days composite NDVI data (Product MOD13Q1) • Time period 2000-2003 • RETRIEVE PROFILES HIGH QUALITY MODIS NDVI • Selection high quality observation (according to MODIS Quality Assurance flags) • Recursive application of a Savitzky-Golay filter Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Assimilation of RS data Regional scale: deconvolution Low Res./High Freq PIXEL 250 m rice (fc = 70 %) MODIS 2003: unmixed NDVI deciduous (fc = 15%) poplar (fc = 15%) [250x250m; 2 bands; 0.4-0.9m; daily; regional application] Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Final comparison Comparison between measured and modeled annual GPP (2003) after recalibrations 0.01 6 0 .01 8 B e fo re R e c a lib ra t io n GPP 2 00 3 - Ob s GPP 2 00 3 - A fte r Re c. (2 Step ) 0 .01 6 0.01 4 A ft er R e c a lib ra t io n GPP 2 00 3 - B efo re R ec. ) -1 -1 0 .01 2 -2 y da day -2 0 .01 0 0.01 2 ) 0 .01 4 0.01 0 0.00 8 0 .00 8 0.00 6 0 .00 6 0.00 4 0 .00 2 Recalibrated ecophysiological parameters 0 .00 0 0 50 10 0 1 50 2 00 DOY 250 30 0 35 0 Mo de le d G PP (Kg C m GPP (Kg C m 0 .00 4 Bef ore-R e c. slope : 0. 76 Inte rce pt: 0.00 06 r ²: 0. 78 0.00 2 0.00 0 0 .000 0 .002 0 .00 4 0.00 6 0.00 8 0.0 10 O b se rve d G PP (Kg C m -2 0.0 12 0.0 14 0. 016 -1 da y ) RMSE: 1.59 gC m-2 day-1 EF: 0.81 Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Af ter-R ec . slope : 0. 77 Inte rcep t: 0.00 21 r ²: 0. 78 METHODS – Application of Biome-BGC at Regional Scale Ecophysiological Parameters (Clone I-214) Solar Radiation, PAR Temperature, Precipitation, VPD Standing Biomass (Leaf Carbon etc..) Site Characteristics (Texture, Soil Depth, etc.) Credits: Bigfoot Project Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Data Assimilation Approaches • Determination of Model Inizialization Parameters • Forcing • Estimation And Optimization Through Model Inversion Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy NEE characterisation 6 15 4 10 2002 5 2 2003 NE E (µ m o lCO 2 m -2 -1 NEE (gC m-2 day-1) s ) 0 -5 -10 -15 -20 -25 0 -2 -4 -6 -8 -30 29 Giug no 2 0 02 2 9 G iug no 20 0 3 -10 -35 0. 00 2.00 4.00 6.00 8.00 10.00 12. 00 14.00 H o ur o f Day 16. 00 18.00 20. 00 March - April 22.00 August - September -12 0 50 100 150 200 250 Day of Year Daily variability Inter-annual variability Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy 300 350 400 Canopy characteristics and NEE at the experimental site 3/3 Footprint Poplar Spatial trend of LAI 4 2 N E E (µ m o lC O 2 m -2 -1 s ) 0 -2 -4 -6 -8 -10 NEE = 0 + 1LAI + 2ln(PAR) -12 -14 -16 0 200 400 600 800 -2 1000 1200 1400 -1 P A R (µ m ol m s ) Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy Conclusions • The proposed methodology can provide important informations for the application of process models at regional scale • Modified version of Biome-BGC sligthly improve the accuracy of GPP estimation and permit to simulate NDVI MODIS. • Inverse Modelling approach against Eddy Covariance Measurements permit to determine an ecophysiological dataset that improve GPP estimation • Inverse Modelling approach against MODIS NDVI permit to determine start ad end of growing season •To improve this methodology in order to recalibrate other phenological parameters and site characteristics (e.g. transfert growth and litterfall period; maximum leaf carbon, soil effective depth). •Optimization algorithm (for second step rec.) is not still stable. Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy LTDA Thank you Workshop on Climatic Analysis and Mapping for Agriculture 14-17 June 2005, Bologna, Italy LTDA