Industrial development and human activities have greatly altered land cover over the past several decades.Besides

the increased cutting of forests and burning of fossil fuels have raised carbon dioxide concentrations in the atmosphere and has led to global temperature increases.Photosynthesis by vegetation removes carbon dioxide from the atmosphere and so plays an important role in the carbon cycle.To measure net primary production（NPP） is a way to understand the photosynthesis capabilities of the vegetation. NPP has been assessed using satellite data by several methods

which includes the light-use efficiency（LUE） model and normalized difference vegetation index（NDVI）both of which are commonly employed using the red and near-infrared channels.In order to study zonal net primary production（NPP） effectively using multi-spectral satellite data

a new vegetation index based on pattern decomposition method and a NPP estimation model taking into account photosynthetic saturation are developed by field experiments.In this paper

we focus on estimating NPP for a temperate forest zone（Kii Peninsula

Japan

mainly covered by temperate forest） using MODIS data of 2001 with a spatial resolution of 500m.To understand the photosynthetic capability of different vegetation types

we calculated NPP values for each land cover type using both the proposed method and a LUE model.Based on the land cover classification

global solar exposure

air temperature

and monthly average effective day length for vegetation photosynthesis

with the proposed method we estimated the following annual NPP values（in units of kg CO2/m2/a）: evergreen

2.04;deciduous

2.23;farm

1.74;paddy

1.42;and urban area

1.06.In comparison

the LUE model estimated the following values: evergreen

1.99;deciduous

2.09;farm

1.76;paddy

1.53;and urban area

1.23.An IPCC report has listed NPP estimates for temperate forests as 2.29 and 2.86 kg CO2/m2/a.The annual values of zonal NPP for the evergreen category calculated using the proposed method agree with those listed by the IPCC report within the algorithm error of 26%. To validate the proposed method

results were compared NPP based on land surveys of temperate forest with paddy areas.The forest survey took place at an 80×80 m plot on Yoshino Mountain in Nara Prefecture.The forest results were 1.52±0.36 kg CO2/m2/a for the proposed method

1.15 kg CO2/m2/a for the LUE model

and 1.50±0.75 kg CO2/m2/afor the survey data.The NPP estimations by the survey and proposed method were more agreed within the permissible error.However

the paddy NPP estimated using satellite data（1.42 kg CO2/m2/a） was nearly 60% that of the field survey（2.48 kg CO2/m2/a）.Because that paddies receive various nutrient and water supplements

unlike a natural forest

and this may have affected the parameter calculation.Additional field surveys of paddy areas are planned in the Kii Peninsula to develop more precise paddy parameters. Although the NPP estimate for paddy was only 60% of the survey NPP

paddy only affected 3% of the Kii Peninsula zonal NPP and thus could be ignored here.Accounting for the 26% estimated error of the algorithm

for the whole region from 32°30′ to 36°24′ N

134°30′ to 137°06′ E（area=3.94×104 km2）

the annual zonal NPP was calculated as 6.11±1.62 kg CO2/a. This study shows that the proposed NPP estimation method can be applied to temperate forest regions

such as the Kii Peninsula.Verifying the method in other vegetation areas will lead to greater precision and allow for NPP estimation on a global scale.