PARTITIONING GPP NT RESULTS DESCRIPTION

This file includes weekly Gross Primary Production (GPP) estimates and related uncertainty. GPP is calculated as GPP=RECO-NEE where RECO is estimated as function of temperature using the Reichstein et al. 2005 partitioning method and the model parameterized using nighttime measured NEE data. The parameterization of the model is done for each calendar year separately. The meteorological data used to apply the model have been gapfilled using MDS method and the ERA downscaled (consolidated version, see the meteo info file for details)
GPP is estimated starting from the 40 different NEE estimates (based on different ustar thresholds, see NEE info document) and for the two different ustar threshold extraction (ustar threshold varying among years and constant ustar in case at least three years of data are present).
To correctly understand the meaning of the uncertainty it is suggested to read also the NEE documentation. The quality of the GPP is linked to the quality of the NEE data and for this reason the NEE quality flags (included in the NEE products) should be considered. These 40 GPP versions have been first aggregated to weekly time resolution (from daily) and then used as basis for all the derived variables provided.


VARIABLES DEFINITION:
LEGEND: HH (half-hourly or hourly), DD (daily), WW (weekly), MM (monthly), YY (yearly)

TIMESTAMP_START (YYYYMMDD): ISO timestamp - short format, first day of the aggregated WEEK

TIMESTAMP_END (YYYYMMDD): ISO timestamp - short format, last day of the aggregated WEEK

WEEK (WW): Week of the year, from 1 to 52, last week including all days needed to complete the year

GPP_NT_VUT_REF (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, reference version selected from GPP versions using a model efficiency approach. Based on corresponding NEE_VUT_XX version 
    HH: (umolCO2 m-2 s-1)
    DD: calculated from half-hourly data (gC m-2 d-1)
    WW-MM: average from daily data (gC m-2 d-1)
    YY: sum from daily data (gC m-2 y-1)

GPP_NT_VUT_USTAR50 (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, based on NEE_VUT_USTAR50 
    HH: (umolCO2 m-2 s-1)
    DD: calculated from half-hourly data (gC m-2 d-1)
    WW-MM: average from daily data (gC m-2 d-1)
    YY: sum from daily data (gC m-2 y-1)

GPP_NT_VUT_MEAN (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, average from GPP versions, each from corresponding NEE_VUT_XX version 
    HH: average from 40 half-hourly GPP_NT_VUT_XX (umolCO2 m-2 s-1)
    DD: average from 40 daily GPP_NT_VUT_XX (gC m-2 d-1)
    WW: average from 40 weekly GPP_NT_VUT_XX (gC m-2 d-1)
    MM: average from 40 monthly GPP_NT_VUT_XX (gC m-2 d-1) 
    YY: average from 40 yearly GPP_NT_VUT_XX (gC m-2 y-1)

GPP_NT_VUT_SE (see temporal resolution for units) Standard Error for Gross Primary Production, calculated as (SD(GPP_NT_VUT_XX) / SQRT(40)) 
    HH: SE from 40 half-hourly GPP_NT_VUT_XX (umolCO2 m-2 s-1)
    DD: SE from 40 daily GPP_NT_VUT_XX (gC m-2 d-1)
    WW: SE from 40 weekly GPP_NT_VUT_XX (gC m-2 d-1)
    MM: SE from 40 monthly GPP_NT_VUT_XX (gC m-2 d-1) 
    YY: SE from 40 yearly GPP_NT_VUT_XX (gC m-2 y-1)

GPP_NT_VUT_XX (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, based on corresponding NEE_VUT_XX (with XX = 05, 16, 25, 50, 75, 84, 95) 
    HH: (umolCO2 m-2 s-1)
    DD: calculated from half-hourly data (gC m-2 d-1)
    WW-MM: average from daily data (gC m-2 d-1)
    YY: sum from daily data (gC m-2 y-1)

GPP_NT_CUT_REF (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, reference selected from GPP versions using a model efficiency approach. Based on corresponding NEE_CUT_XX version 
    HH: (umolCO2 m-2 s-1)
    DD: calculated from half-hourly data (gC m-2 d-1)
    WW-MM: average from daily data (gC m-2 d-1)
    YY: sum from daily data (gC m-2 y-1)

GPP_NT_CUT_USTAR50 (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, based on NEE_CUT_USTAR50 
    HH: (umolCO2 m-2 s-1)
    DD: calculated from half-hourly data (gC m-2 d-1)
    WW-MM: average from daily data (gC m-2 d-1)
    YY: sum from daily data (gC m-2 y-1)

GPP_NT_CUT_MEAN (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, average from GPP versions, each from corresponding NEE_CUT_XX version 
    HH: average from 40 half-hourly GPP_NT_CUT_XX (umolCO2 m-2 s-1)
    DD: average from 40 daily GPP_NT_CUT_XX (gC m-2 d-1)
    WW: average from 40 weekly GPP_NT_CUT_XX (gC m-2 d-1)
    MM: average from 40 monthly GPP_NT_CUT_XX (gC m-2 d-1) 
    YY: average from 40 yearly GPP_NT_CUT_XX (gC m-2 y-1)

GPP_NT_CUT_SE (see temporal resolution for units) Standard Error for Gross Primary Production, calculated as (SD(GPP_NT_CUT_XX) / SQRT(40)) 
    HH: SE from 40 half-hourly GPP_NT_CUT_XX (umolCO2 m-2 s-1)
    DD: SE from 40 daily GPP_NT_CUT_XX (gC m-2 d-1)
    WW: SE from 40 weekly GPP_NT_CUT_XX (gC m-2 d-1)
    MM: SE from 40 monthly GPP_NT_CUT_XX (gC m-2 d-1) 
    YY: SE from 40 yearly GPP_NT_CUT_XX (gC m-2 y-1)

GPP_NT_CUT_XX (see temporal resolution for units) Gross Primary Production, from Nighttime partitioning method, based on corresponding NEE_CUT_XX (with XX = 05, 16, 25, 50, 75, 84, 95) 
    HH: (umolCO2 m-2 s-1)
    DD: calculated from half-hourly data (gC m-2 d-1)
    WW-MM: average from daily data (gC m-2 d-1)
    YY: sum from daily data (gC m-2 y-1)


MODEL EFFICIENCY SELECTION FOR GPP_NT_CUT_REF and GPP_NT_VUT_REF:
FAILURE OF THE MODEL PARAMETRIZATION:

When not enough data are available to parameterize the model the GPP can not be calculated. This can happen for entire years or for NEE versions filtered with high USTAR thresholds (that remove more data).
In this site the partitioning model was not applied in the years:
2007
2008
2009


Model Efficiency selection:

The reference GPP has been selected on the basis of the Model Efficiency. Starting from the 40 different GPP estimations (obtained parameterizing the model using 40 NEE versions filtered with different USTAR thresholds) it has been calculated the Model Efficiency between each version and the others 39. The reference GPP has been selected as the one with higher Model Efficiency sum (so the most similar to the others 39).
In this dataset have been selected as reference:

GPP_NT_CUT_REF obtained parameterizing the model with NEE filtered using the USTAR percentile 43.75 (USTAR value: 0.3778)
GPP_NT_VUT_REF obtained for YEAR 1999 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.439553)
GPP_NT_VUT_REF obtained for YEAR 2000 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.449424)
GPP_NT_VUT_REF obtained for YEAR 2001 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.419243)
GPP_NT_VUT_REF obtained for YEAR 2002 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.389571)
GPP_NT_VUT_REF obtained for YEAR 2003 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.382209)
GPP_NT_VUT_REF obtained for YEAR 2004 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.371542)
GPP_NT_VUT_REF obtained for YEAR 2005 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.3075)
GPP_NT_VUT_REF obtained for YEAR 2006 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.292571)
GPP_NT_VUT_REF obtained for YEAR 2007 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.262578)
GPP_NT_VUT_REF obtained for YEAR 2008 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0)
GPP_NT_VUT_REF obtained for YEAR 2009 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0)
GPP_NT_VUT_REF obtained for YEAR 2010 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.249091)
GPP_NT_VUT_REF obtained for YEAR 2011 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.2794)
GPP_NT_VUT_REF obtained for YEAR 2012 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.297)
GPP_NT_VUT_REF obtained for YEAR 2013 parameterizing the model with NEE filtered using the USTAR percentile 31.25 (USTAR value: 0.320975)