Modeling the effects of recent land use change On the carbon cycle in the zhu jiang delta region Of southern china

An important feature the LUE model is that the variables S, FPAR, and f can be estimated using existing satellite remote sensing techniques (Goward and Dye, 1996). We validated the model results by comparing them with NPP figures reported in the research literature for comparable ecosystems and environmental conditions.

2.3 Data Sources

2.3.1 Net Primary Production.
We used monthly average normalized difference vegetation index (NDVI) data to estimate FPAR, based on the empirical relationship employed by Ruimy et al. (1994). The NDVI data were from the NOAA/NASA Pathfinder AVHRR Land (PAL) data set (Agbu and James, 1994). Monthly PAR data were from the satellite-derived data set described by Dye and Shibasaki (1995). The spatial resolution of the NDVI and PAR data sets were 8 km and 100 km, respectively. We set the e value for NAT to 1.39 g MJ, based on field data from Peng and Zhang (1995). We set e for AGR to 1.74 g MJ-1, which is value given by Ruimy et al. (1994) for tropical cultivation. Based on available climate data, we assumed that environmental constraints in our study area were negligible and set the value of f in Equation (1) to 1. We referred to the 30 m land use maps from Seto et al. (2000) to select sample locations in the course resolution FPAR images to represent the NAT and AGR land use classes. From the sample data we produced time series of monthly FPAR and PAR values for input to the NPP model.

2.3.2 Carbon Storage.
Our estimates of the carbon content of soils in the ZJDR are based on data reported by Cai (1996). We estimated carbon storage in phytomass in NAT by referring to data given by Chen et al. (1993) and other published sources. For AGR, we referred to data reported by Atjay et al. (1977). The carbon storage data were specified on a per-unit-area basis (density). In the current analysis we assume the carbon densities associated with the land use classes remain fixed at their specified levels during the study period.

2.4 Estimating Regional Effects
To estimate the regional effects of land use change we assumed that ecosystem NPP and carbon storage were spatially homogeneous within the land use classes. We calculated regional changes by multiplying the area-based NPP and carbon storage values by the total land area lost or gained for each land use class.

3. Results

3.1 Validation of NPP Model Results
Based on the observed range of values from field studies reported in the literature, we specified high, medium, and low NPP estimates for NAT and AGR. For NAT, our model estimate of NPP differs from the medium field estimate by +13% (20.5 and 18.2 t DM ha-1 yr-1, respectively). The high quality of the field data used in the NAT comparison strengthens our confidence that our model estimate of NPP is reasonably accurate for NAT. The results suggest that the model was moderately successful in estimating NPP for AGR. Because of shortcomings in the reported data for AGR, caution is necessary when interpreting the AGR comparison results.

3.2 Effects of Land Use Change on NPP
The NPP analysis indicates that land use change in the ZJDR between 1988 and 1996 reduced the annual amount of atmospheric C assimilated into phytomass by approximately 1.55 Mt (-7.5%) (Table 1). More than half (55%) of this reduction is attributable to the loss of agricultural land.

Table 1. LUE model results showing NPP (rate and regional total) for 1988 and 1996, and the net change, DNPPtot. Conversion from dry matter (DM) to carbon assumes .45 g C per g DM.

Table 3. Net change in regional total phytomass (988-1996) based on estimates from reported field data and land use change (Fig. 1).

 

Land Use Class	Net change in Phytomass(104 t C)
	high	med	low
NAT	-1347.3	-464.5	-269.1
AGR	-203.4	-101.9	-0.3
Total	-1550.7	-566.4	-269.4

3.3 Effects of Land use Change on Carbon Storage
Our results indicate that conversion of natural and agricultural land to urban uses in the ZJDR from 1988 to 1996 is the source of a release of between 140×104 and 310×104 tons of soil organic C, with a medium estimate of 225×104 tons (Table 2). These figures account for the expected cumulative release during the 20 years following conversion. Patterns of carbon loss observed in studies of cultivated soils (Schlesinger et al., 2000) suggest that the majority of the carbon would be released in the several years immediately following conversion.