Developing a continental-scale measure of gross primary production by combining MODIS and AmeriFlux data through Support Vector Machine approach

Yang, Fan; Ichii, Kazuhito; White, Michael A.; Hashimoto, Hirofumi; Michaelis, Andrew; Votava, P.; Zhu, A‐Xing; Huete, Alfredo; Running, Steven W.; Nemani, Ramakrishna R.

doi:10.1016/j.rse.2007.02.016

Cited by 182 publications

(170 citation statements)

References 42 publications

Supporting

Mentioning

159

Contrasting

Unclassified

Order By: Relevance

“…Both TEC GPP and MOD17A2 GPP estimates were more accurate for forest ecosystems than that for non-forest ecosystems. A similar result was reported by Yang et al (2007) and Heinsch et al (2006). Yearly estimated MOD17A2 GPP (Fig.…”

Section: [ ( F I G _ 3 ) T D $ F I G ]supporting

confidence: 89%

“…The GPP algorithm does not include the effect of soil water stress, and sensitivity of GPP to D is increased in the MOD17 GPP model to partially account for the problem (Heinsch et al, 2006). Although MOD17A2 GPP is the only global-scale operational GPP product in high temporal and spatial resolutions, recent validation studies (Turner et al, 2003;Rahman et al, 2005;Yang et al, 2007) show that it has considerable errors due to problems associated with: inputs of climate data (Turner et al, 2006;Zhao et al, 2005); use of biome properties look-up table (Heinsch et al, 2006;Turner et al, 2006); the MOD17 algorithm itself (Heinsch et al, 2006).…”

Section: Modis Gpp Product (Mod17a2)mentioning

confidence: 99%

“…Several statistical models such as the simple temperature and greenness model (TG model;Sims et al, 2008), the regression tree approach (Xiao et al, 2010), the support vector machine model (SVM; Yang et al, 2007), model tree ensembles (MTE; Jung et al, 2011), remote sensing based greenness and radiation model (GR; Wu et al, 2011), the total canopy chlorophyll content and potential incident photosynthetically active radiation model (Gitelson et al, 2012), the temperature and greenness rectangle model (TGR; Yang et al, 2013), and the photosynthetic capacity model (PCM; Gao et al, 2014) have been developed to estimate GPP. Calibrations are required to build statistical GPP models.…”

Section: Introductionmentioning

confidence: 99%

“…Since satellites can supply large-scale observation of terrestrial vegetation, a diverse set of satellite-based process models have developed quickly during recent years. These have the potential to accurately predict GPP and NPP from regional to continental scales (Potter et al, 1993;Ruimy et al, 1994;Field et al, 1995;Running et al, 2000;Xiao et al, 2005a;Yuan et al, 2007;Yang et al, 2007). Remote sensing-based process models are principally based on the light-use-efficiency theory -photosynthesis production correlates with the absorbed photosynthetically active radiation (APAR) (Monteith, 1972;Asrar et al, 1984) and FPAR is derived from remote sensing data, GPP ¼ e Â APAR ¼ e Ã ÂS stress Â FPAR Â PAR (1) where GPP is the gross primary production (gC m À2 month À1 ), e is the actual LUE (gC MJ À1 ) including environmental stresses and is often defined as e* Â S stress , e* is the maximum LUE and S stress refers to environmental stresses, FPAR is the fraction of PAR absorbed by the canopy, and PAR is the incident photosynthetically active radiation (MJ m À2 month À1 ).…”

Section: Introductionmentioning

confidence: 99%

“…Further analysis of flux data illustrates environmental controls over carbon dioxide and water vapor exchange of terrestrial vegetation . Thus, more flux data measured at different ecosystem types have been used for calibration and evaluation of global LUE GPP models (Wang et al, 2004;Xiao et al, 2005a;Heinsch et al, 2006;Yuan et al, 2007;Yang et al, 2007;Sjöström et al, 2013).…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Improved global simulations of gross primary product based on a new definition of water stress factor and a separate treatment of C3 and C4 plants

Yan

Wang

Billesbach

et al. 2015

Ecological Modelling

View full text Add to dashboard Cite

A B S T R A C TAccurate simulation of terrestrial gross primary production (GPP), the largest global carbon flux, benefits our understanding of carbon cycle and its source of variation. This paper presents a novel light use efficiency-based GPP model called the terrestrial ecosystem carbon flux model (TEC) driven by MODIS FPAR and climate data coupled with a precipitation-driven evapotranspiration (E) model (Yan et al., 2012). TEC incorporated a new water stress factor, defined as the ratio of actual E to Priestley and Taylor (1972) potential evaporation (E PT ). A maximum light use efficiency (e*) of 1.8 gC MJ À1 and 2.76 gC MJ À1 was applied to C3 and C4 ecosystems, respectively. An evaluation at 18 eddy covariance flux towers representing various ecosystem types under various climates indicates that the TEC model predicted monthly average GPP for all sites with overall statistics of r = 0.85, RMSE = 2.20 gC m À2 day À1 , and bias = À0.05 gC m À2 day À1 . For comparison the MODIS GPP products (MOD17A2) had overall statistics of r = 0.73, RMSE = 2.82 gC m À2 day À1 , and bias = À0.31 gC m À2 day À1 for this same set of data. In this case, the TEC model performed better than MOD17A2 products, especially for C4 plants. We obtained an estimate of global mean annual GPP flux at 128.2 AE 1.5 Pg C yr À1 from monthly MODIS FPAR and European Centre for Medium-Range Weather Forecasts (ECMWF) ERA reanalysis data at a 1.0 spatial resolution over 11 year period from 2000 to 2010. This falls in the range of published land GPP estimates that consider the effect of C4 and C3 species. The TEC model with its new definition of water stress factor and its parameterization of C4 and C3 plants should help better understand the coupled climate-carbon cycle processes.

show abstract

Section: [ ( F I G _ 3 ) T D $ F I G ]supporting

confidence: 89%

Section: Modis Gpp Product (Mod17a2)mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Improved global simulations of gross primary product based on a new definition of water stress factor and a separate treatment of C3 and C4 plants

Yan

Wang

Billesbach

et al. 2015

Ecological Modelling

View full text Add to dashboard Cite

show abstract

Large‐scale estimation and uncertainty analysis of gross primary production in Tibetan alpine grasslands

Liu

Xiao

et al. 2014

JGR Biogeosciences

View full text Add to dashboard Cite

Gross primary production (GPP) is an important parameter for carbon cycle and climate change research. Previous estimations of GPP on the Tibetan Plateau were usually reported without quantitative uncertainty analyses. This study sought to quantify the uncertainty and its partitioning in GPP estimation across Tibetan alpine grasslands during 2003-2008 with the modified Vegetation Photosynthesis Model (VPM). Monte Carlo analysis was used to provide a quantitative assessment of the uncertainty in model simulations, and Sobol' variance decomposition method was applied to determine the relative contribution of each source of uncertainty to the total uncertainty. The results showed that the modified VPM successfully reproduced the seasonal dynamics and magnitude of GPP of 10 flux tower sites on the plateau (R 2 = 0.77 À 0.95, p < 0.001). The 6 year mean GPP in Tibetan alpine grasslands was estimated at 223.3 Tg C yr À1 (312.3 g C m À2 yr À1 ). The mean annual GPP increased from western to eastern plateau, with the increase of annual temperature and precipitation and the decrease of elevation, while the decrease of GPP from southern to northern plateau was primarily driven by air temperature. Furthermore, the mean relative uncertainty of the annual GPP was 18.30%, with larger uncertainty occurring in regions with lower GPP. Photosynthetic active radiation, enhanced vegetation index, and the maximum light use efficiency (LUE) are the primary sources of uncertainty in GPP estimation, contributing 36.84%, 26.86%, and 21.99%, respectively. This emphasizes the importance of uncertainty in driving variables as well as that of maximum LUE in LUE model simulation.

show abstract

Change in surface energy balance in Alaska due to fire and spring warming, based on upscaling eddy covariance measurements

et al. 2014

Self Cite

View full text Add to dashboard Cite

Warming in northern high latitudes has changed the energy balance between terrestrial ecosystems and the atmosphere. This study evaluated changes in regional surface energy exchange in Alaska from 2000 to 2011 when substantial declines in spring snow cover due to spring warming and large-scale fire events were observed. Energy fluxes from a network of 20 eddy covariance sites were upscaled using a support vector regression (SVR) model, by combining satellite remote sensing data and global climate data. Based on site-scale analysis, SVR reproduced observed net radiation, sensible heat flux, latent heat flux, and ground heat flux; 8 day root-mean-square errors for these variables were 15, 10, 9, and 3 W m À2 , respectively. Based on upscaled fluxes, decreases in spring snow cover induced an increase in surface net radiation, a net heating effect, of 0.56 W m À2 decade À1. This heating effect was comparable to the net cooling effect due to increased fire extent during the study period (up to 0.59 W m À2 decade À1). These land cover effects were larger than the change in the energy forcing associated with CO 2 balance for the Alaska region. Spring warming and postfire land cover change increased the regional latent heat flux. The regional sensible heat flux decreased with the postfire land cover change. Our results highlight the importance of positive spring snow albedo feedback to climate and a postfire negative feedback under the expected warming climate in the Arctic.

show abstract

Developing a continental-scale measure of gross primary production by combining MODIS and AmeriFlux data through Support Vector Machine approach

Cited by 182 publications

References 42 publications

Improved global simulations of gross primary product based on a new definition of water stress factor and a separate treatment of C3 and C4 plants

Improved global simulations of gross primary product based on a new definition of water stress factor and a separate treatment of C3 and C4 plants

Large‐scale estimation and uncertainty analysis of gross primary production in Tibetan alpine grasslands

Change in surface energy balance in Alaska due to fire and spring warming, based on upscaling eddy covariance measurements

Contact Info

Product

Resources

About