Abstract. Dai-Liang Peng 1, Jing-Feng Huang 1, Cheng-Xia Cai 2, Rui Deng 1 and Jun-Feng Xu 1. Journal of Integrative Plant Biology 2008

Transcription

1 Journal of Integrative Plant Biology 2008 Assessing the Response of Seasonal Variation of Net Primary Productivity to Climate Using Remote Sensing Data and Geographic Information System Techniques in Xinjiang Dai-Liang Peng 1, Jing-Feng Huang 1, Cheng-Xia Cai 2, Rui Deng 1 and Jun-Feng Xu 1 ( 1 Institute of Agriculture Remote Sensing and Information System Application, Zhejiang University, Hangzhou , China; 2 Xinjiang Institute of Meteorology, Urumchi , China) Abstract Net primary productivity (NPP) is a key component of energy and matter transformation in the terrestrial ecosystem, and the responses of NPP to global change locally and regionally have been one of the most important aspects in climatevegetation relationship studies. In order to isolate causal climatic factors, it is very important to assess the response of seasonal variation of NPP to climate. In this paper, NPP in Xinjiang was estimated by NOAA/AVHRR Normalized Difference Vegetation Index (NDVI) data and geographic information system (GIS) techniques. The impact of climatic factors (air temperature, precipitation and sunshine percentage) on seasonal variations of NPP was studied by time lag and serial correlation ageing analysis. The results showed that the NPP for different land cover types have a similar correlation with any one of the three climatic factors, and precipitation is the major climatic factor influencing the seasonal variation of NPP in Xinjiang. It was found that the positive correlation at 0 lag appeared between NPP and precipitation and the serial correlation ageing was 0 d in most areas of Xinjiang, which indicated that the response of NPP to precipitation was immediate. However, NPP of different land cover types showed significant positive correlation at 2 month lag with air temperature, and the impact of which could persist 1 month as a whole. No correlation was found between NPP and sunshine percentage. Key words: climate; geographic information system techniques; net primary productivity; remote sensing; seasonal variation; serial correlation ageing; time lag. Peng DL, Huang JF, Cai CX, Deng R, Xu JF (2008). Assessing the response of seasonal variation of net primary productivity to climate using remote sensing data and geographic information system techniques in Xinjiang. J. Integr. Plant Biol. doi: /j x Available online at Terrestrial net primary production (NPP), the difference between gross primary production (GPP) and autotrophic respiration, is vital to human society because it not only provides essential materials, such as food, fiber and wood, but also Received 11 Jun Accepted 7 Feb Supported by the Hi-Tech Research and Development (863) Program of China (2006AA ); the China Meteorological Administration (CCSF ) and Xinjiang Meteorological Bureau (QSR ). Author for correspondence. Tel: ; Fax: ; <xjf11@zju.edu.cn>. C 2008 Institute of Botany, the Chinese Academy of Sciences doi: /j x creates environments suitable for human inhabitation (Peng and Apps 1999; Zhao et al. 2005). In recent years, NPP has received more attention as an important component in the ecosystem process which removes carbon dioxide (CO 2 )from the atmosphere and stores it in short-living (foliage and fine roots) and long-living (wood) tissues (Canadell et al. 2000; Ahl et al. 2004). With the advancement of studies of global change and terrestrial ecosystems, the estimation of NPP of natural vegetation with modeling, especially with remote sensing data, and the responses of NPP to global change locally and regionally have been one of the most important aspects in climate-vegetation relationship studies (Zhang et al. 2002). NPP is sensitive to many controls, including climate, soils, plant characteristics, disturbance regime, and a number of other natural and anthropogenic factors. Climate is a major driver of variation in NPP (Christopher et al. 1995) but a clear

2 2 Journal of Integrative Plant Biology 2008 understanding of the impact of climate change on NPP is lacking (Canadell et al. 2000; Ahl et al. 2004). The responses of NPP to climatic factors are not only related to different climatic factors and different land cover types, but to different temporal dimensions as well (Braswell et al. 1997). NPP interannual fluctuations of different magnitudes do exist (Maisongrande et al. 1995; Kindermann et al. 1996; Malmstrom et al. 1997). Mohamed et al. (2004) explored the patterns of interannual variability in NPP of different land cover types in relation to climatic factors. It is important to study the relationship between seasonal variations of NPP and climatic variability in order to isolate causal climatic factors. On short time scales, air temperature, precipitation and other climatic factors affect the physiological processes that control plant photosynthesis and growth (Christopher et al. 1995). Usually, the effects of climatic factors on NPP are not instantaneous, and sometimes exert delayed effects and serial correlation ageing (Steele et al. 2005; Peng et al. 2007). However, the study of time lag and serial correlation ageing of the impact of climatic factors on seasonal change for NPP has been neglected. In this paper, Xinjiang was selected as the study area. NPP in this region was estimated by NOAA/AVHRR Normalized Difference Vegetation Index (NDVI) data and geographic information system (GIS) techniques. Linear regressions of means and coefficients of variation of NPP and those of air temperature, precipitation and sunshine percentage were performed, and the time lag and serial correlation ageing method was adopted, and all of these analyses were used to assess the responses of seasonal variation of NPP for different land cover types to climate. Results NPP validation and its seasonal variation Based on remote sensing data and spatial interpolated meteorological data, the NPP of eight land cover types at a temporal resolution of 1 month were estimated using the lightuse efficiency and process models. In order to validate the accuracy of NPP estimation, the reference NPP values and actually observed NPP values were used in this study (Table 1). As shown in Table 1, except shrub land and desert steppe, the differences between NPP values of this study and those of the reference, actually observed NPP values were small, which indicated that the results of NPP estimation in this study were credible. It was found that the seasonal variation profile of NPP for all land cover types was single-peaked (Figure 1), the minimal value was even 0 in winter, and the maximum NPP appeared in summer (July). Different land cover types have different values of NPP. Within the forest land cover types, evergreen trees had higher NPP values than those of deciduous trees. For croplands, because the farming system in most areas of Xinjiang is one crop/year with the growing season mainly April October, the NPP was maximum in July. The NPP value of grassland was greater than shrub land and desert steppe, which was 57 g C m 2 month 1 in summer. The NPP value of desert steppe was the lowest compared with those of other land cover types, which was almost 0 in winter and still less than 39 g C m 2 month 1 in summer. According to the coefficients of variation (CVs) of NPP for eight land cover types (Figure 2), the CVs of deciduous broadleaf trees, deciduous needleleaf trees and crops were 1.16, 1.13 and 1.07, respectively, and which were rather larger than other vegetation types. The immediate cause was the short growing season, with onset of greenness in spring, optimal growth in summer, defoliation in autumn and minimal NPP in winter. The difference of CVs for the other five land cover types were small. For desert steppe, shrub land and grassland, due to the stunted and sparse distribution with low coverage, and their seasonal variations could not be identified by the sensors on the satellite. For evergreen trees, however, the CVs were small because of their physiological characteristics. Correlation between seasonal variation of climate and NPP First of all, the correlation coefficients (R) between three climatic factors (air temperature, precipitation and sunshine percentage) and NPP of eight land cover types at different time lags Table 1. Comparison of annual NPP of different land cover types from different studies (g C m 2 year 1 ) Land cover type Land cover Reference value Observed (Hu et al. 1990; This study type code (Sun and Zhu 2000) Liu et al. 1993; Wang 1993) Evergreen broadleaf trees EBT Deciduous broadleaf trees DBT Evergreen needleleaf trees ENT Deciduous needleleaf trees DNT Cropland CRO Grassland GRA Shrub land SHR Desert steppe DES

3 Seasonal Variation of NPP and Climate 3 NPP (g C m 2 month 1 ) EBT END DBT CRO DNT GRA SHR DES The results of correlation between the CVs of seasonal NPP for eight land cover types and those of three climatic factors are shown in Figure 3. The CVs of NPP for four land cover types (evergreen needleleaf trees, cropland, deciduous broadleaf trees and deciduous needleleaf trees) exhibited positive significant correlation with the CV of precipitation, while no positive/negative significant correlations were found with CVs of air temperature and sunshine percentage Month Figure 1. Seasonal variations of net primary productivity (NPP) for eight land cover types in Xinjiang. CRO, cropland; DBT, deciduous broadleaf trees; DES, desert steppe; DNT, deciduous needleleaf trees; EBT, evergreen broadleaf trees; ENT, evergreen needleleaf trees; GRA, grassland; SHR, shrub land. in every county were calculated. Subsequently, the maximal absolute values (R) were obtained from different time lags and considered as the correlation coefficient between one climatic factor and NPP for each land cover type. It was found that the mean of NPP showed significant positive correlation with monthly mean air temperature and monthly total precipitation in almost every county at P = 0.01; but there was no significant correlation between the mean values of NPP and those of sunshine percentage as a whole. The results indicated that the NPP for different land cover types had a similar correlation with any one of the three climatic factors in Xinjiang, and different climatic factors showed different correlations with NPP. Time lags and serial correlation ageing Because monthly mean sunshine percentage showed no significant correlation with seasonal variation of NPP, it was decided not to analyze the time lags and serial correlation ageing for this variable. Air temperature anomaly was used to examine the role of one of the major climatic factors that govern the activity of land cover, and NPP for different land cover types showed significant positive correlation at 2 month lag with air temperature in Xinjiang. In addition, the impact of air temperature could persist 1 month in most areas in Xinjiang. However, significant positive at 0 lag and no serial correlation ageing were found between NPP for different land cover types and seasonal variation precipitation in Xinjiang (Figure 4). Discussion The uncertainty of impact of climate on NPP not only includes different climatic factors and different land cover types, but also includes different temporal dimensions. To improve our understanding and ability to assess the impact of climate on seasonal variation of NPP accurately, the linear regressions of DNT DBT CRO SHR DES ENT EBT GRS Figure 2. Coefficient of variation of seasonal variation of net primary productivity for eight land cover types in Xinjiang. CRO, cropland; DBT, deciduous broadleaf trees; DES, desert steppe; DNT, deciduous needleleaf trees; EBT, evergreen broadleaf trees; ENT, evergreen needleleaf trees; GRA, grassland; SHR, shrub land. Figure 3. Correlation between the coefficient of variation (CV) of seasonal net primary productivity for eight land cover types and CV of air temperature (white bars), precipitation (black bars) and sunshine percentage (dashed bars). CRO, cropland; DBT, deciduous broadleaf trees; DES, desert steppe; DNT, deciduous needleleaf trees; EBT, evergreen broadleaf trees; ENT, evergreen needleleaf trees; GRA, grassland; SHR, shrub land.

4 4 Journal of Integrative Plant Biology 2008 Figure 4. Lagged correlations (r value) of seasonal anomalies of net primary productivity for eight land cover types versus lagged anomalies of air temperature and precipitation. Zero lag (white bars), 1 month lag (black bars) and 2 month lag (dashed bars). CRO, cropland; DBT, deciduous broadleaf trees; DES, desert steppe; DNT, deciduous needleleaf trees; EBT, evergreen broadleaf trees; ENT, evergreen needleleaf trees; GRA, grassland; SHR, shrub land. mean and CVs of NPP for different land cover types and those of three climatic factors were performed, and the time lag and serial correlation ageing analysis methods were adopted in this paper. The correlations of mean and CVs of NPP for different land cover types and those of three climatic factors (air temperature, precipitation and sunshine percentage) indicated that NPP of different land cover types showed significant correlation with the different climatic factors in Xinjiang at P = 0.01, and precipitation was the major climatic factor which impacted the seasonal variation of NPP. Not only the correlation between mean values of NPP and total precipitation were significant in seasonal variation, but also the correlation between the CVs of NPP for four land cover types and precipitation was as well. This indicated that the seasonal variation of NPP mainly contributed to precipitation in Xinjiang as a whole, and the relatively small variation in precipitation considerably determined the seasonal variation of NPP. The CVs of the grassland, shrub land and desert steppe showed no significant correlation with the CVs of precipitation at P = 0.01; the main reason may have been because the stunted and sparse distribution with the low coverage, and their seasonal variation were not able to be identified by the sensors on the satellite, and therefore the sensitivity of this vegetation to precipitation by remote sensing sensors was not obvious. The correlation between mean values of NPP and those of air temperature was statistically significant at P = 0.01, but the CVs of seasonal variation NPP showed no significant correlation with air temperature. This indicated that, although air temperature did not seem to contribute to the seasonal variation of NPP, it remained an essential predictor of the mean NPP value. However, there was no significant correlation of sunshine percentage with NPP even though there was abundant sunshine in Xinjiang, especially in summer, the season of maximum plant growth, which could be attributed to the greater evaporation in this season with increasing water loss and decreasing productivity. The increase in plant growth under warm temperatures results in an increase of biomass input to the soil pool, while the associated precipitation increases soil moisture and facilitates the transformation of organic matter into readily-available inorganic nutrients of mainly phosphorous and nitrogen resulting in delayed vegetation growth (Neill et al. 1995; Tian et al. 1998; Mohamed et al. 2004). This is the main reason for a significant positive 2 month lagged and 1 month serial correlation between NPP for different land cover types and air temperature on seasonal variation in most areas of Xinjiang. Generally, the response of NPP to precipitation was not instantaneous. Precipitation infiltrates the soil, is absorbed by root system, converted into photosynthetic material which then absorbs and reflects sunlight, which is acquired by the satellite sensor to be used as a measure of terrestrial reflectance and vegetation response. This process could account for the time lags for response of NPP to precipitation. However, in Xinjiang, there was significant positive correlation at 0 lag and a lack of serial correlation ageing between NPP for different land cover types and precipitation on seasonal variation, which indicated a directly immediate enhancement of vegetation production following a positive precipitation anomaly. The main reason for this was that the climate in Xinjiang features a temperate zone continental climate with annual precipitation of 145 mm, which was 23% of the national average annual precipitation (630 mm) and most of the plants were lacking water, so the response of vegetation to precipitation was instantaneous and could not persist for a long time. Unfortunately, the temporal resolution of the NPP was 1 month, and the phenology of the growing season and the effect of climate on NPP could not be accurately captured, which would have resulted in some errors for assessing the impact of climate on NPP. Besides climatic factors, there are many other factors which disturb NPP, especially anthropogenic factors, such as land use change (deforestation, fire and urban construction), irrigation and multiple cropping indices, which suggests that there are still many questions to be answered about the impacts of climate and land use on NPP.

5 Seasonal Variation of NPP and Climate 5 Materials and Methods Study area Xinjiang is approximately located between N and E. The elevation ranges m a.s.l. The land cover statistics revealed that desert steppe (68.84%) and grassland (18.86%) were the dominant land cover types. Other land cover types included evergreen needleleaf trees (0.53%), evergreen broadleaf trees (0.04%), deciduous needleleaf trees (0.01%), deciduous broadleaf trees (0.07%), cropland (2.92%) and shrub land (8.90%). The climate of Xinjiang features a temperate continental arid/semiarid climate with an annual precipitation of 145 mm, which is 23% of the national average (630 mm), and an annual temperature range of C. Xinjiang is the largest area in China, with the lowest rates of productivity, and it is therefore important to estimate NPP and assess the impacts of climatic variability on NPP in this large semiarid area. Data and pre-processing Remote sensing and terrain data In this study, the following remote sensing data were used: land cover (spatial resolution, 1 km) supplied by the National Aeronautics and Space Administration (NASA); NOAA/AVHRR NDVI from (spatial resolution, 8 km) obtained from the National Satellite Meteorological Center; digital elevation model (spatial resolution 1 km); and an administrative map of Xinjiang (1: ) from the Xinjiang Institute of Meteorology. Using GIS techniques, ERDAS Imagine and ArcMap software, these data were resampled to the 8 km AVHRR data and projected to Geographic (Lat/Lon) (WGS84 datum). These remote sensing images were calibrated, and areas of interest (AOI) were created for eight land cover types. Finally, the 10-year mean values of NOAA/AVHRR NDVI were calculated at a temporal resolution of 1 month. Meteorological data The accuracy of the input data to the models is critical to the accurate estimation of NPP. However, it is very difficult to obtain high-quality weather data at large scales and the input data has to rely on the limited number of ground meteorological stations. For each climatic variable, the weather station data have to be interpolated/extrapolated to the whole study area by using spatial interpolation algorithms. The quality of the interpolated data depends on the number and distribution of weather stations available in the study region (Bunkei et al. 2004). In this study, the following meteorological data were used: monthly mean air temperature; monthly total precipitation; monthly mean vapor pressure; monthly sunshine percentage; and monthly mean wind speed. Ten year mean values ( ) for every climatic factor from 100 weather stations were calculated at a temporal resolution of 1 month. The mathematical models were developed by stepwise regression to study the relationship between every climatic factor and variable (longitude, latitude and elevation), and then grid maps of the climatic factors were made through trend surface simulation and residual interpolation using GIS techniques. In the model-building process, the data from 80 weather stations were used for modeling, and other weather stations data were used to validate the precision of the models (Figure 5). The results of this validation showed that the relative errors of precipitation were high, most of which were greater than 10%. For other climatic factors, the relative errors were less than 5% in more than 80% areas of Xinjiang; in the other 20% areas, the relative errors were less than 10% (Table 2). The monthly average absolute errors between simulated and measured values for every climatic factor showed that the monthly average absolute errors of precipitation were less than 1 mm except in winter, and the absolute errors of air temperature, sunshine percentage, vapor pressure and wind speed were less than 0.7 C, 0.01%, 0.3 hpa and m/s, respectively. NPP estimates Accurate estimate of NPP is critical to understanding the carbon dynamics within the atmosphere-vegetation-soil continuum and the response of the terrestrial ecosystem to climate (Bunkei et al. 2004). There are numerous models to estimate global and regional NPP. Ruimy et al. (1999) suggested three types of models which were generally used to estimate terrestrial NPP: (i) statistical models (Lieth 1975); (ii) parametric models (Potter et al. 1993; Prince et al. 1995; Ruimy et al. 1999); and (iii) ecological process models (Running and Coughlan 1988; Foley 1994). Each group of models has strengths and limitations. Statistical models are well known for their simplicity but limited generality (Lieth 1975). Parametric models have the advantage of remotely sensing data, especially at large scales, but lose the link to some critical ecological processes by using empirical relationships/constants. Process models are based on current knowledge of major ecological/biophysical processes, but suffer from a high level of complexity, large computational demand and the difficulty to calibrate. The integration of light-use efficiency and process model algorithms is a potentially effective approach to estimate global and regional NPP using remote sensing data and ecological/biophysical processes (Bartelink et al. 1997; Franklin et al. 1997; Goetz and Prince 1998). In this paper, NPP was defined as: NPP = GPP (R mo + R g ) (1) Where GPP is gross primary production, R mo is maintenance respiration by all other living parts except leaves and fine roots, and R g is growth respiration; GPP = εg FPAR PAR f 1(T) f 2(β) (2)

6 6 Journal of Integrative Plant Biology 2008 Figure 5. Weather stations over Xinjiang. Where εg is light-use efficiency, FPAR is the fraction of photosynthetically active radiation absorbed by green vegetation and PAR is the photosynthetically active radiation absorbed by green vegetation, and f1 (T) and f2 (β) are the effects of air temperature and soil moisture on photosynthesis, respectively (Christopher et al. 1995; Sun 1996; Sun and Zhu 2001; David et al. 2002; Zhao et al. 2005). Statistical analysis Based on the grid maps of monthly mean values of NPP for different land cover types, air temperature, sunshine percentage and monthly total precipitation, the mean values of these variables were extracted at the county level using the AOI for different land cover types using GIS techniques. The impact of climate on NPP based on regression analysis, the time lag and serial correlation ageing analysis were assessed. Regression analysis The monthly mean values of NPP, air temperature and sunshine percentage, and monthly total precipitation for eight land cover types were calculated at the county level using GIS techniques. The correlation between the monthly mean values of NPP and those of air temperature, sunshine percentage and monthly total precipitation for eight land cover types were performed and correlation coefficients were obtained. We have adopted the assumption of Fang et al. (2001) that if the correlation between the CVs of NPP and CVs of air temperature, precipitation and sunshine percentage were found statistically significant, the seasonal variation in NPP is attributed to the temporal variability in these climatic factors. Mohamed et al. (2004) used the same assumption to prove that, in the mid-northern latitudes, the relatively small variation in cloud cover considerably determines the interannual variation of NPP while temperature variability is inversely correlated to the variability in NPP. The CVs of NPP and those of air temperature, precipitation, and sunshine percentage for eight land cover types were calculated at the county level. The correlation between CVs of NPP and those of three climatic factors for eight land cover types were performed and correlation coefficients were obtained. Time lag and serial correlation ageing analysis Usually, the response of NPP to climate is not instantaneous, and sometimes exerts delayed effects. This delayed NPP response was thought to incorporate both immediate physiological alterations and delayed biogeochemical adjustments of global

7 Seasonal Variation of NPP and Climate 7 Table 2. The monthly average relative errors and absolute errors between simulated and measured values for every climatic factor Air temperature Precipitation Sunshine percentage Vapor pressure Wind speed Month Relative Absolute Relative Absolute Relative Absolute Relative Absolute Relative Absolute error (%) error ( C) error (%) error (mm) error (%) error error (%) error (hpa) error (%) error (m/s) vegetation due to variable climate (Mohamed et al. 2004; Steele et al. 2005; Peng et al. 2007). The method of time lag analysis can be defined as (Fik and Mulligan 1998; Chen and Liu 2002; Zhang et al. 2005): r k (x, y) = S k(x, y) (3) S x S y+k Where S k (x,y) is the sample covariance; S x S y+k are standard deviations, and they can be calculated by the following formulae: S k (x, y) = 1 (x i x i )(y i+k y i+k ) S x = [ S y+k = [ The means are defined as: 1 1 (x i x i ) 2 ] 1/2 x i = 1 y i = 1 (y i+k y i+k ) 2 ] 1/2 x i (4) (5) y i+k Where n is the sample number of x i and y i, k is the number of time lag, and k n/4 or k = n 10, in this study, n is 12, and k = 0, 1, 2. When k is 0, it means there is no time lag, which reflects immediate NPP response to climate variation. The marginal values of significant correlation at P = 0.01 are different with different time lags, which are 0.661, and when k is 0, 1 and 2, respectively. When the correlation coefficient (R k ) is greater than the marginal value and is maximal in all correlation coefficients at different time lags, the time lags equal to the values of k multiply 1 month. The impact of climate on seasonal variation NPP may be not instantaneous, and sometimes this impact is long-term. In this paper, the serial correlation ageing meant the time of the impact of climatic factors on NPP could persist, and it was calculated by accumulating values of the time of serial significant correlation at P = 0.01 between NPP and three climatic factors at all time lags. For example, it endowed the correlation coefficient with R 0, R 1 and R 2 when k was 0, 1 and 2, respectively. If R 0 and R 1 were both greater than the marginal values of significant correlation at P = 0.01, the serial correlation ageing would be 1 month. If R 0, R 1 and R 2 were all greater than the marginal values of significant correlation at P = 0.01, the serial correlation ageing would be 2 months. Acknowledgements The authors acknowledge the data support of the National Aeronautics and Space Administration (NASA), the National

8 8 Journal of Integrative Plant Biology 2008 Satellite Meteorological Center and the Xinjiang Institute of Meteorology. References Ahl DE, Gower ST, Mackay DS, Burrows SN, Norman JM, Diak GR (2004). Heterogeneity of light use efficiency in a northern Wisconsin forest: implications for modeling net primary production with remote sensing. Remote Sens. Environ. 93, Bartelink HH, Kramer K, Mohren GMJ (1997). Applicability of the radiation-use efficiency concept for simulating growth of forest stands. Agr. For. Meteorol. 88, Braswell BH, Schimel DS, Linder E, Moore B III (1997). The response of global terrestrial ecosystems to interannual temperature variability. Science 278, Bunkei M, Ming X, Jin C, Satoshi K, Masayuki T (2004). Estimation of regional net primary productivity (NPP) using a process-based ecosystem model: how important is the accuracy of climate data? Ecol. Model. 178, Canadell JG, Mooney HA, Baldochi DD, Berry JA, Ehleringer JR, Field CB (2000). Carbon metabolism of the terrestrial biosphere: a multi-technique approach for improved understanding. Ecosystems 3, Chen YG, Liu JS (2002). Derivation and generalization of the urban gravitational model using fractal idea with an application to the spatial cross correlation between Beijing and Tianjin. Geogr. Res. 21, (in Chinese with an English abstract). Christopher BF, James TR, Carolyn MM (1995). Global net primary production: combining ecology and remote sensing. Remote Sens. Environ. 51, David PT, Stith TG, Warren BC, Matthew G, Tom KM (2002). Effects of spatial variability in light use efficiency on satellite-based NPP monitoring. Remote Sens. Environ. 80, Fang J, Piao S, Tang Z, Peng C, Ji W (2001). Interannual variability in net primary production and precipitation. Science 293, 1723a. Fik TJ, Mulligan GF (1998). Functional Form and spatial interaction models. Environ. Plan. A 30, Franklin SE, Lavigne MB, Deuling MJ, Wulder MA, Hunt J (1997). Landsat TM derived forest cover types for modeling net primary production. Can. J. Remote Sens. 23, Foley JA (1994). Net primary productivity in the terrestrial biosphere: the application of a global model. J. Geophys. Res. 99, Goetz SJ, Prince SD (1998). Variability in carbon exchange and light utilization among boreal forest stands: implications for remote sensing of net primary production. Can. J. For. Res. 28, Hu ZZ, Sun JX, Zhang YS (1990). Preliminary studies on calorific value and nutrient composition in tianzhu alpine polygonum viviparum meadow. Acta Phytoecol. et Geobot. Sin. 14, (in Chinese with an English abstract). Kindermann J, Wurth G, Kohlmaier GH, Badeck FW (1996). Interannual variation of carbon exchange fluxes in terrestrial ecosystems. Global Biogeochem. Cy. 10, Lieth H (1975). Modeling the primary productivity of the world. In: Lieth H, Whittaker RH, eds. Primary Productivity of the Biosphere. Vol. 14. Springer-Verlag, New York. pp Liu SR, Xu DY, Wang B (1993). Impacts of climate change on productivity of forests in China 1: geographic distribution of actual productivity of forest in China. For. Res. 6, (in Chinese with an English abstract). Malmstrom CM, Thompson MV, Juday GP, Los SO, Randerson JT, Field CB (1997). Interannual variation in global-scale net primary production: testing model estimates. Global Biogeochem. Cy. 11, Maisongrande P, Ruimy A, Dedieu G, Saugier B (1995). Monitoring seasonal and inter-annual variations of gross primary productivity using a diagnostic model and remote-sensed data. Tellus B 47, Mohamed MAA, Babiker IS, Chen ZM, Ikeda K, Ohta K, Kato K (2004). The role of climate variability in the inter-annual variation of terrestrial net primary production (NPP). Sci. Total Environ. 332, Neill C, Piccolo MC, Steudler PA, Melillo JM, Feigl BJ, Cerri CC (1995). Nitrogen dynamics in soils of forests and active pastures in the Western Brazilian Amazon Basin. Soil Biol. Biochem. 27, Peng CH, Apps MJ (1999). Modelling the response of net primary productivity (NPP) of boreal forest ecosystems to changes in climate and fire disturbance regimes. Ecol. Model. 122, Peng DL, Huang JF, Wang XZ (2007). Correlative analysis between regional vegetation seasonal fluctuation and climate factors based on MODIS-EVI. Chin. J. Appl. Ecol. 18, (in Chinese with an English abstract). Potter CS, Randerson JT, Field CB, Matson PA, Vitousek PM, Mooney HA et al. (1993). Terrestrial ecosystem production: a process model based on global satellite and surface data. Global Biogeochem. Cy. 7, Prince SD, Goets SJ, Goward SN (1995). Monitoring primary production from earth observation satellites. Water Air Soil Poll. 82, Running SW, Coughlan JC (1988). A general model of forest ecosystem processes for regional applications. I, Hydrological balance, canopy gas exchange and primary production processes. Ecol. Model. 42, Ruimy A, Kergoat L, Bondeau A (1999). Comparing global models of terrestrial net primary productivity (NPP): analysis of differences in light absorption and light-use efficiency. Global Change Biol. 5, Steele BM, Reddy SK, Nemani RR (2005). A regression strategy for analyzing environmental data generated by spatio-temporal processes. Ecol. Model. 181, Sun R (1996). Research of the Terrestrial Vegetation Net Primary Production (NPP) in China Base on AVHRR-NDVI (D). Beijing Normal University, Beijing (in Chinese).

9 Seasonal Variation of NPP and Climate 9 Sun R, Zhu QJ (2001). Effect of climate change of terrestrial net primary productivity in China. J. Remote Sens. 5, (in Chinese with an English abstract). Sun R, Zhu QJ (2000). Distribution and seasonal change of net productivity in China from April, 1992 to March, Acta Geogr. Sin. 55, (in Chinese with an English abstract). Tian H, Melillo JM, Kicklighter DW, McGuire AD, Helfrich JVK III, Moore B III et al. (1998). Effect of interannual climate variability on carbon storage in Amazonian ecosystems. Nature 396, Wang YF (1993). The biomass and productivity of typical steppe in innermongotial. Plants 4, Zhao MS, Heinsch FA, Nemani RR, Running SW (2005). Improvements of the MODIS terrestrial gross and net primary production global data set. Remote Sens. Environ. 95, Zhang H, Gao SY, Zheng QH (2002). Responses of NPP of salinized meadows to global change in hyperarid regions. J. Arid Environ. 50, Zhang XX, Ge QS, Zheng JY (2005). Impacts and lags of global warming on vegetation in Beijing for last 50 years based on remotely sensed data and phenological information. Chin.J.Ecol.24, (in Chinese with an English abstract). (Handling editor: Jiquan Chen)