Annual Gross Primary Production from Vegetation Indices: A Theoretically Sound Approach

Abstract

Abstract:A linear relationship between the annual gross primary production (GPP) and a PAR-weighted vegetation index is theoretically derived from the Monteith equation.A semi-empirical model is then proposed to estimate the annual GPP from commonly available vegetation indices images and a representative PAR, which does not require actual meteorological data. A cross validation procedure is used to calibrate and validate the model predictions against reference data. As the calibration/validation process depends on the reference GPP product, the higher the quality of the reference GPP, the better the performance of the semi-empirical model. The annual GPP has been estimated at 1-km scale from MODIS NDVI and EVI images for eight years. Two reference data sets have been used: an optimized GPP product for the study area previously obtained and the MOD17A3 product. Different statistics show a good agreement between the estimates and the reference GPP data, with correlation coefficient around 0.9 and relative RMSE around 20%. The annual GPP is overestimated in semiarid areas and slightly underestimated in dense forest areas. With the above limitations, the model provides an excellent compromise between simplicity and accuracy for the calculation of long time series of annual GPP.

Empirical vs. light-use efficiency modelling for estimating carbon fluxes in a mid-succession ecosystem developed on abandoned karst grassland

“…The heterogeneous nature of the ecosystem in this study, made of a mix of tree patches and grassland, partly explains why EVI performed better than NDVI. In fact, EVI proved to be a better proxy for GPP in forest ecosystems where NDVI can get saturated [ 42 , 44 ]. Although r 2 and rmse suggested EVI to be the best proxy for GPP, the low ΔAICc between the EVI and SAVI-based models led to conclude that there was no substantial difference of performance between the two models [ 41 ].…”

A carbon sink-driven approach to estimate gross primary production from microwave satellite observations

“…Gross Primary Production (GPP), is a key ecosystem process. Estimation of GPP from satellite observations commonly uses optical data together with empirical or semi-empirical models (Gilabert et al, 2017;Running et al, 2004) or machine learning approaches (Beer et al, 2010;Jung et al, 2011;Tramontana et al, 2016;Yang et al, 2007). Biophysical properties obtained from optical remote sensing that are often used to estimate GPP include the fraction of Absorbed Photosynthetically Active Radiation (fAPAR), Normalized Difference Vegetation Index (NDVI), or Leaf Area Index (LAI).…”

“…These approaches rely on the light-use efficiency theory (Monteith, 1972) whereby GPP depends on the incoming Photosynthetically Active Radiation (PAR), the fraction of PAR that is absorbed, i.e. fAPAR, and the efficiency of converting light to assimilated carbon (Beer et al, 2010;Gilabert et al, 2017;Jung et al, 2011;Running et al, 2004;Tramontana et al, 2016;Yang et al, 2007). Another variable retrieved from optical data is Solar-Induced chlorophyll Fluorescence (SIF), which is a measure for photosynthetic activity (Frankenberg et al, 2011;Guan et al, 2016).…”

Quantifying the Impacts of Land-Use and Climate on Carbon Fluxes Using Satellite Data across Texas, U.S.

“…An understanding of the relationship between regional carbon fluxes in the forms of gross primary productivity (GPP) and net ecosystem exchange (NEE) with environmental changes is necessary to understand the response of ecosystems to global climate changes [35][36][37][38]. GPP, a critical parameter for carbon cycle and climate change, is the total amount of carbon fixed by plants through photosynthesis in an ecosystem [39][40][41][42]. Accurate measurement or estimate of GPP is essential for the carbon cycle and climate change studies [43,44].…”