Improving terrestrial CO<sub>2</sub> flux diagnosis using spatial structure in land surface model residuals

Abstract

<jats:p>Abstract. We evaluate spatial structure in North American CO2 flux observations using a simple diagnostic land surface model. The Vegetation Photosynthesis Respiration Model (VPRM) calculates net ecosystem exchange (NEE) using locally observed temperature and photosynthetically active radiation (PAR) along with satellite-derived phenology and moisture. We use observed NEE from a group of 65 North American eddy covariance tower sites spanning North America to estimate VPRM parameters for these sites. We investigate spatial coherence in regional CO2 fluxes at several different time scales by using geostatistical methods to examine the spatial structure of model data-model residuals. We find that persistent spatial structure does exist in the data-model residuals at a length scale of approximately 1000 km. This spatial structure defines a flux-tower-based VPRM residual covariance matrix. The residual covariance matrix is useful in constructing prior fluxes for atmospheric CO2 concentration inversion calculations, as well as for constructing a VPRM North American CO2 flux map optimized to eddy covariance observations. Finally, the estimated VPRM parameter values do not separate clearly by plant functional type (PFT). This calls into question whether PFTs partition ecosystems by carbon cycle participation when the viewing lens is a simple model.</jats:p>

Terrestrial CO2 Fluxes, Concentrations, Sources and Budget in Northeast China: Observational and Modeling Studies

“…In addition, land use types (referred to as plant functional types in Hilton et al, 2013) used in WRF‐VPRM to calculate CO 2 fluxes, currently classified into seven categories, namely, evergreen forest, deciduous forest, mixed forest, shrubland, savannah, cropland, and grassland, are also derived from MODIS land cover product (MOD12Q1) (Zhang et al, 2017b).…”

“…PAR is calculated using the shortwave (SW) downward radiation as SW/0.505 (Mahadevan et al, 2008). The values of four empirical parameters ( α , β , λ , and PAR 0 ) for seven land use categories used in this study (summarized in Table 1) were calibrated previously by Hilton et al (2013) using data from 65 EC towers over North America. Note that the calibrated parameters varied between different towers among the same land use category over the North America, and we used the median values from different towers in each category for all the seven categories.…”

“…The commonly used LUE models include the Carnegie‐Ames‐Stanford Approach (Field et al, 1995), Carbon Fix (Veroustraete et al, 2002), Carbon Flux (Turner et al, 2006), the Eddy‐Covariance (EC)‐LUE model (Yuan et al, 2007), the Vegetation Photosynthesis Model (Xiao et al, 2004), and the Vegetation Photosynthesis and Respiration Model (VPRM) (Mahadevan et al, 2008). GEE and NEE estimated using LUE models still have large uncertainties and discrepancies, mainly due to model structure differences (Yuan et al, 2014) and model parameter choices (Hilton et al, 2013). In general, top‐down and bottom‐up estimates differ widely across all spatiotemporal scales.…”

“…In order to reconcile estimates of NEE using top‐down and bottom‐up methods, researchers have been incorporating the in situ and satellite observations into modeling studies. The high‐temporal‐resolution EC measurements of CO 2 fluxes and concentrations over diverse biomes crossing wide latitudes have been widely applied to optimize model parameters and validate modeling results (Dayalu et al, 2018; Hilton et al, 2013; Jamroensan, 2013; Mahadevan et al, 2008; Yuan et al, 2014). In addition, column‐averaged CO 2 concentrations (XCO 2 ) retrieved from satellites are effective to constrain CO 2 fluxes and can be used to evaluate simulation results on wide spatial scales, especially in the regions with limited in situ measurements (Basu et al, 2013; Crowell et al, 2019; Hu et al, 2019b, 2020; Rayner et al, 2014; Rayner & O'Brien, 2001).…”

“…In the past decade, the application and evaluation of WRF‐VPRM mainly focused on North America (Feng et al, 2016; Hu et al, 2019b, 2020; Park et al, 2018; Ye et al, 2017) and Europe (Pillai et al, 2011). The model performances remain unknown in Asia including China due to the lack of observations, as well as model uncertainties due to uncertainties of VPRM parameters in the region (Dayalu et al, 2018; Hilton et al, 2013; Liu et al, 2015; Zhang et al, 2017a). Diao et al (2015) used WRF‐VPRM to investigate the spatiotemporal characteristics of NEE and surface CO 2 concentrations over the Yangtze River Delta region in China for 5 days, from 28 July to 2 August 2010.…”