In this study, BIOME-BGC was applied in a pixel mode of 250 Â 250 m, the
same resolution as MODIS data. Initial modeling inputs, daily meteorological data,
and parameters other than foliar C:N ratio for the study area are described in
He (2008).
8.3.2 Mapping C:N Ratio Using Remote Sensing Data
To derive a C:N ratio map from remote sensing, experimental data and methods
from recent literature were adopted. Specifically, the foliar C:N values were
calculated by using a constant C CNT value (i.e., mean C value measured at the
27 grass sampling plots measured by Psomas et al. (2008), C CNT ¼ 44.05) over the
N predictions from a MODIS NDVI exponential model (y ¼ 9.98
(À3.49x) ; Hansen
and Schjoerring 2003). In this study, MODIS Band 3 (459–479 nm) was used to
replace r 440 in the equation, and MODIS Band 4 (545–565 nm) was used to replace
r 573 (Hansen and Schjoerring 2003). Two MODIS images acquired on July 11 of
2005 (the same period that field data were collected) were obtained from the
Canadian Centre of Remote Sensing and used to calculate a vegetation index and
C:N ratio map.
8.3.3 Sensitivity Analysis of the Model to C:N Ratio
Sensitivity analysis (SA) is a commonly-used method to quantify the variation of
the model outputs to variation in model parameters (Saltelli et al. 2000). SA of
model parameters is carried out by changing them and observing the corresponding
response in the output variables. When local SA techniques are applied, parameter
values are changed one at a time, while fixing all other parameter values (Bar
Massada and Carmel 2008). Global SA alters a subset or all the parameters
simultaneously in a given model simulation (Helton et al. 2006). Two foliar C:N
ratio scenarios were applied in this study to examine the sensitivity of the BIOMEBGC model, and to estimate the advantage of spatially distributed foliar C:N ratio
over the constant values. First, the “global C:N” scenario was applied using C:N
values ranging from 5 to 45 to drive the model. This range of foliar C:N ratio values
was determined by a previous study in which grassland foliar C:N ratio was found
to vary from 5.83 to 44.98 (Psomas et al. 2008). At the same time, special attention
was paid to the accuracy of simulated NPP while using the default C:N ratio defined
by the BIOME-BGC model (i.e., C:N ¼ 24.0) which is most frequently used in
ecosystem process modeling studies. Second, the “Remote Sensing C:N” scenario
was applied using the foliar C:N map derived from MODIS data to drive the model.
160
Y. He et al.
same resolution as MODIS data. Initial modeling inputs, daily meteorological data,
and parameters other than foliar C:N ratio for the study area are described in
He (2008).
8.3.2 Mapping C:N Ratio Using Remote Sensing Data
To derive a C:N ratio map from remote sensing, experimental data and methods
from recent literature were adopted. Specifically, the foliar C:N values were
calculated by using a constant C CNT value (i.e., mean C value measured at the
27 grass sampling plots measured by Psomas et al. (2008), C CNT ¼ 44.05) over the
N predictions from a MODIS NDVI exponential model (y ¼ 9.98
(À3.49x) ; Hansen
and Schjoerring 2003). In this study, MODIS Band 3 (459–479 nm) was used to
replace r 440 in the equation, and MODIS Band 4 (545–565 nm) was used to replace
r 573 (Hansen and Schjoerring 2003). Two MODIS images acquired on July 11 of
2005 (the same period that field data were collected) were obtained from the
Canadian Centre of Remote Sensing and used to calculate a vegetation index and
C:N ratio map.
8.3.3 Sensitivity Analysis of the Model to C:N Ratio
Sensitivity analysis (SA) is a commonly-used method to quantify the variation of
the model outputs to variation in model parameters (Saltelli et al. 2000). SA of
model parameters is carried out by changing them and observing the corresponding
response in the output variables. When local SA techniques are applied, parameter
values are changed one at a time, while fixing all other parameter values (Bar
Massada and Carmel 2008). Global SA alters a subset or all the parameters
simultaneously in a given model simulation (Helton et al. 2006). Two foliar C:N
ratio scenarios were applied in this study to examine the sensitivity of the BIOMEBGC model, and to estimate the advantage of spatially distributed foliar C:N ratio
over the constant values. First, the “global C:N” scenario was applied using C:N
values ranging from 5 to 45 to drive the model. This range of foliar C:N ratio values
was determined by a previous study in which grassland foliar C:N ratio was found
to vary from 5.83 to 44.98 (Psomas et al. 2008). At the same time, special attention
was paid to the accuracy of simulated NPP while using the default C:N ratio defined
by the BIOME-BGC model (i.e., C:N ¼ 24.0) which is most frequently used in
ecosystem process modeling studies. Second, the “Remote Sensing C:N” scenario
was applied using the foliar C:N map derived from MODIS data to drive the model.
160
Y. He et al.