Net Primary Production Studies Over Large Scales

1
Net Primary Production Studies Over Large
Scales Daolan Zheng Dept. of Earth Ecological,
and Environmental Sciences, Univ. of Toledo,
Toledo, OH 43606 Contributors T. Hame, K.
Hibbard, R. Olson, W. Parton, S. Prince, and the
Participants of the Global Primary Production
Data Initiatives and the Ecosystem Model/Data
Inter-comparisons
2
1) Global Primary Production Data Initiatives
(GPPDI) dataset (0.5 x 0.5 degree NPP) 2)
Ecosystem Model/Data Inter-comparison (EMDI)
3
GPPDI
ABSTRACT Net Primary Production (NPP) is an
important component of the carbon cycle but
direct field measurement of NPP is tedious and
not practical for large areas and so models are
generally used to study the carbon cycle at a
global scale. Most NPP data are for relatively
small field plots that cannot represent the 0.50
x 0.50 grid cells that are commonly used in
global NPP models. We summarize and present a
series of methods (4) that were used by original
authors OR us to prepare a consistent data set of
NPP for 0.50 grid cells for a range of biomes.
The grid cells are grouped to the biome level and
are compared with existing compilations of field
NPP and the results of the Miami potential NPP
model. The full dataset currently contains 3654
cells. An edited subset consists of 2335 cells
in which outliers were removed and all replicate
measurements were averaged for each unique
geographic location. (http//daacl.esd.ornl.gov/np
q/GPPDI/Combined_GPPDI_des.html).
4
Major methods Aggregation of fine-scale (plot
or stand-level) vegetation inventory Direct
correlation of extensive data sets of ground
measurements with remotely sensed spectral
vegetation indices Local modeling of NPP
using key independent variables Regression
analysis to link productivity with controlling
environmental variables.
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
EMDI
Abstract. Validation of global net primary
production (NPP) models on spatial and temporal
dimensions is much needed. This study compared
the predicted NPP from 9 global models with the
observed NPP estimates developed in the United
States and Australia at 0.50 x 0.50 spatial
resolution. Differences among NPP estimates
varied significantly over space and time.
Overall, mean model results overestimated by
about 65-83 and 148-198 gC/m2/yr for ANPP (0-600)
and TNPP (0-1000), respectively, compared to the
observed data. Such biases increased slightly for
higher NPP values. Trend of inter-annual NPP
variation derived from model mean for Queensland
(1960-95), Australia did not significantly differ
(p0.1). While in Iowa (1982-95), USA between
model/data mean differed (flood, 1993)..
11
Objectives
a) Generate continuous NPP maps based on field
inventory b) Compare global model
predictions with field-based NPP c) Examine
spatial patterns of modeled NPP d) Compare
model outputs with key environmental variables.
12
Sources
Iowa
Queensland
Source Barrett
Source Parton/Sala/Tieszen/Brown
13
Models
 
 
 
14
TNPP residual (Model-Data)
USA
Australia
15
U.S.A.
(Xi-Xmin) / (Xmax Xmin)
Data
Model mean
R2 0.84
16
Australia
Data
Model mean
(Xi-Xmin) / (Xmax Xmin)
R2 0.66
17
Mo-Tnpp
Total Net Primary Production (TNPP)
18
Mo_Anpp
Aboveground Net Primary Production (ANPP)
19
Mod-AET
Comparison of model TNPP means with Actual
Evapotranspiration (AET)
20
Mod-NDVI
Comparison of model TNPP means with Normalized
Difference Vegetation Index (NDVI)
21
Mod-PPT
Comparison of model TNPP means with Annual total
precipitation
22
Mod-TEMP
Comparison of model TNPP means with Annual mean
temperature
23
Aust.-Indiv. (-21.25S/143.75E)
24
AUS-mean (Annual)
Australia field data model mean(-21.25S/143.7
5E)36 year sequence
25
Iowa individual models14 year sequence
26
IOWA-Mean (Annual)
Iowa field data and model mean14 year sequence
27
Conclusions
First model/data comparison at 0.5 grain
with continental extent. Models differ
substantially. Models overestimate NPP (or
data underestimated) Uncertainty from this
study (different methods in estimating litterfall
in U.S., lack of data in Aus., grazing effects,
BNPP allocation). Reasonable agreement
between NPP data and model outputs (in both
spatial patterns and temporal variations).
ANPP comparisons may reduce uncertainty.
28
Acknowledgments

• Many individuals and modeling groups who
  contributed data and model runs to GPPDI and
  EMDI.
• International Geosphere-Biosphere Programme
  (IGBP), Data and Information System (DIS).
• National Center for Ecological Analysis and
  Synthesis (NCEAS) for EMDI I, II, and III
  workshops, data organization and distribution.