Assessing the impact of urbanization on net primary productivity using multi-scale remote sensing data: a case study of Xuzhou, China

Abstract

An improved Carnegie Ames Stanford Approach (CASA) model based on two kinds of remote sensing (RS) data, Landsat Enhanced Thematic Mapper Plus (ETM+) and Moderate Resolution Imaging Spectroradiometer (MODIS), and climate variables were applied to estimate the Net Primary Productivity (NPP) of Xuzhou in June of each year from 2001 to 2010. The NPP of the study area decreased as the spatial scale increased. The average NPP of terrestrial vegetation in Xuzhou showed a decreasing trend in recent years, likely due to changes in climate and environment. The study area was divided into four sub-regions, designated as highest, moderately high, moderately low, and lowest in NPP. The area designated as the lowest sub-region in NPP increased with expanding scale, indicating that the NPP distribution varied with different spatial scales. The NPP of different vegetation types was also significantly influenced by scale. In particular, the NPP of urban woodland produced lower estimates because of mixed pixels. Similar trends in NPP were observed with different RS data. In addition, expansion of residential areas and reduction of vegetated areas were the major reasons for NPP change. Land cover changes in urban areas reduced NPP, which could chiefly be attributed to human-induced disturbance.

Keywords

multi-scale remote sensing / net primary productivity / improved Carnegie Ames Stanford approach model / urbanization

Introduction

As the most fundamental biological component of the Earth’s surface, vegetation plays an important role in the regional and global environment, as well as in the survival and growth of other biological communities. A better understanding of the climatic, environmental, and ecological changes in the Earth system, including those relating to Earth’s atmospheric aerosol transport, radiation budget, biology and climate interactions, and carbon cycle, are needed to better understand how climate change may lead to environmental degradation ( Prieto-Blanco et al., 2009a). Net Primary Productivity (NPP) is regarded as an ecological index of environmental quality and state of health to evaluate the sustainability of an ecosystem ( Fang, 2000). Among the pools and fluxes that make up the terrestrial carbon cycle, NPP is regarded as the annual net carbon flux from the atmosphere to the biosphere ( Hazarika et al., 2005). NPP is defined as the amount of organic matter produced by green plants per unit of time and space ( Field et al., 1995; Liu et al., 1999). In general, NPP, the amount of total organic carbon fixed by photosynthesis after accounting for respiration, should effectively reflect the viability of plant communities under natural conditions and changes in climate that may influence growth of vegetation ( Sun and Zhu, 1999). As the study of NPP has progressed, NPP estimates are applied not only to carbon cycling studies, but also to government environmental decision making and regional management ( Huang et al., 2011). With the rapid development of urbanization in China, urban ecological and environmental problems have become increasingly serious, thus the study of NPP has become increasingly valuable. Urbanization is one of the most important aspects of global change. Land cover changes directly reduce regional NPP. Thus, the increasing intensity of human-induced disturbance could be considered an important factor in the decrease in forest NPP in urbanized areas ( Imhoff et al., 2004; Xu et al., 2007). Indications have shown that settlement, population, and Gross Domestic Product (GDP) have had a strongly negative impact on NPP in Southeastern China ( Lu et al., 2010). The loss of NPP due to urban expansion could be estimated by integrating remote sensing with GIS techniques ( Tian and Qiao, 2014).

Traditionally, NPP has been estimated using climate-productivity relationship models ( Del Grosso et al., 2008; Pranuthi et al., 2012) and eco-physiological processing models ( Li and Ji, 2001; Sasai et al., 2011). However, NPP estimation based on Remote Sensing (RS) and Geographic Information System (GIS) technology has been shown to have vast advantages over previous approaches, especially for the study of regional and global scales ( Zhao et al., 2005; Prieto-Blanco et al., 2009b; Yuan et al., 2014). Thus, RS and GIS have been widely applied in recent years ( Feng et al., 2006), making RS data one of the main focus areas in the development of new NPP estimation models ( Zhao et al., 2007). At home and abroad, the majority of NPP estimations has been based on RS data with low spatial resolution, such as the National Oceanic and Atmospheric Administration (NOAA)/ Advanced Very High Resolution Radiometer (AVHRR) and MODIS ( Huang et al., 2011; Singh et al., 2011 ; Liu et al., 2013). Based on the MODIS data, the process-based Boreal Ecosystem Productivity Simulator (BEPS) model was utilized to study the changes of NPP in China from 2000 to 2010 ( Liu et al., 2013). However, there is a disadvantage to using low spatial resolution for studies of urban areas due to the smaller area size and more sporadic distribution of woodlands at these locations. Notably, the study suggested that MODIS NPP data exhibit general correlations with local climate and land use, but tend to be overestimated in low-productivity sites and underestimated in high-productivity sites ( Turner et al., 2006). Meteorological data, fractional photosynthetically active radiation (FPAR) data, and the parameterization of light use efficiency in a simple light use efficiency model (MOD17) have provided insights into the site-specific causes for the differences between MODIS and BigFoot products ( Turner et al., 2006).

The Carnegie Ames Stanford Approach (CASA) model, based on resource balance perspective, was established to estimate regional or global NPP ( Potter et al., 1993). RS data, including climate, vegetation, and soil data, are the fundamental data CASA utilizes to estimate NPP ( Dong and Ni, 2011). In this paper, we apply an improved CASA model using two kinds of RS data (Landsat ETM+ and MODIS/Normalized Difference Vegetation Index (NDVI)) at two scales along with corresponding meteorological data to evaluate the NPP of the city of Xuzhou, China from June of 2001 through June of 2010. We also analyze the temporal and spatial variations in NPP at different spatial scales to help determine the robustness of this model. Lastly, it was considered that the urbanization was the main factor in NPP change.

Study area

Xuzhou is a city in the northwest of Jiangsu Province and is adjacent to Shandong, Henan and Anhui Provinces. It is located in 116°22′‒118°40′E, 33°43′‒34°58′N, and covers an area of approximately 11,258 km². The urban area is about 963 km². The location of Xuzhou urban area is shown in Fig.1. The eastern part of the region has a higher elevation than the west, and the altitude of Xuzhou ranges over 19–45 m above sea level. The middle and northeast parts of the region contain many hills and low mountains, while the northwest is an alluvial plain in which Feng and Pei counties are located. Old Yellow River and the Beijing-Hangzhou Grand Canal flow across the whole territory. Xuzhou located in a mountainous area and has a warm temperate monsoon climate. The main vegetation type is warm, temperate, deciduous broadleaf forest.

Data and method

Data and processing

In this study, Landsat ETM+ images and 16-day NDVI products (250 m spatial resolution) derived from MODIS data (June of 2001‒2010) were used to estimate NPP. The ETM+ images were classified into six classes by support vector machines ( Melgani and Bruzzone, 2004; Tan and Du, 2008): woodland, lawn, farmland, urban, water area, and barren. The small-scale NDVI was calculated by ETM+ band 3 and band 4. The monthly mean temperature, precipitation, and hours of sunshine for the study area were gathered from the Chinese Meteorological Data Share Service System (http://cdc.cma.gov.cn/home.do) and spatially interpolated using the Kriging interpolation method to obtain two scales of spatial resolution, one for each of the two kinds of RS data. Finally we obtained two NPP production estimates at two different spatial resolutions: 30 m for ETM+ data and 250 m for MODIS data.

Improved CASA model

The CASA model, based on a resource balance point of view, was established to estimate regional or global NPP ( Potter et al., 1993). CASA utilizes spatial data of climate, vegetation, and soil to estimate NPP ( Dong and Ni, 2011). Based on the CASA model, vegetation types are considered when determining the maximum and minimum of NDVI and the maximum of the lightest utilization rate. Three kinds of meteorological data (temperature, precipitation, solar radiation) were used to estimate the water stress factor in the CASA model ( Franklin et al., 1997). The accuracy of the improved CASA model characterized by simple operation has been verified (Zhu et al., 2005).The framework of CASA model is shown in Fig.2.

NPP is calculated as a function of absorbed photosynthetically active radiation (APAR) and real light utilization rate (ϵ) (Zhu et al., 2005):

(1) NPP\left(x,t\right)=APAR\left(x,t\right)\times \mathrm{ϵ}\left(x,t\right),

where x is a given pixel and t represents the month.

APAR calculation

(2) APAR\left(x,t\right)=SOL\left(x,t\right)\times FPAR\left(x,t\right)\times \mathrm{0.5},

where SOL(x, t) is the total solar radiation at pixel x at month t, measured in MJ·m^‒2·month^‒1 and FPAR(x, t) is a fraction of photosynthetically active radiation absorbed by canopy within the same pixel over the same time frame.

Few meteorological stations are available to collect the solar radiation in the study area, so a model between sunshine hours and solar radiation has been established to calculate daily solar radiation ( Schuol et al., 2007).

(3) H_{\mathrm{0}}={\displaystyle \frac{\mathrm{24}}{\mathrm{\pi }}}{I}_{SC}{E}_{\mathrm{0}}\left[\mathit{\omega }{T}_{SR}\sin \mathit{\delta }\sin \mathit{\psi }+\cos \mathit{\delta }\cos \mathit{\psi }\sin \left(\mathit{\omega }{\mathit{T}}_{SR}\right)\right],

where E₀ is the Earth orbital eccentricity correction factor, I_SC is the solar constant (4.921 MJ·m^‒2·h^‒1), ω is the rotational angular velocity of the earth (0.2618 rad·h^‒1), T_SR is sunshine duration, δ is solar declination, and ψ is geographic latitude.

E₀ can be calculated using the model established by Duffie ( Duffie and Beckman, 2006).

(4) E_{\mathrm{0}}={\left(r_{\mathrm{0}}/r\right)}^{\mathrm{2}}=\mathrm{1}+\mathrm{0.033}\cos \left(\mathrm{2}\mathrm{\pi }{d}_n/\mathrm{365}\right),

where r₀ is the average Earth-Sun distance, r is the Earth-Sun distance of a day, and d_n is the number of days in a year.

Next, the formula to calculate δ as follows ( Hejun, 2006),

(5) \mathit{\delta }=\mathrm{0.006918}-\mathrm{0.399912}\cos \theta +\mathrm{0.010257}\sin \theta -\mathrm{0.006758}\cos \mathrm{2}\theta +\mathrm{0.000907}\sin \mathrm{2}\theta \theta =\mathrm{2}\mathrm{\pi }{\delta }_{\nu }/\mathrm{365},

where {\delta }_{\nu } is the number of the day in the year starting from the first day of January.

T_SR can be calculated through the following formula ( Neitsch et al., 2002),

(6) T_{SR}={\cos }^{-\mathrm{1}}\left(\left|-\tan \mathit{\delta }\tan \mathit{\psi }\right|\right)/\mathit{\omega }.

where δ, ψ, and ω have the same values as for Eq. (3).

In order to establish values of SOL(x, t) using the calculated value of H₀, we finally used the following empirical formulas to calculate real daily solar radiation (H):

(7) H_L=\mathrm{0.8}{H}_{\mathrm{0}}.

(8) H=H_L\times \left(a+b\times S/S_L\right),

where a, b are constants simulated according to the measured value of the solar radiation and S and S_L are sunshine duration and day length respectively. Previously established results for H_L and H₀ , a and b are 0.248 and 0.752, respectively (Zuo et al., 1963).

FPAR could be received by the linear relationship between NDVI and FPAR ( Ruimy et al., 1994):

(9) FPAR\left(x,t\right)={\displaystyle \frac{\left(NDVI\left(x,t\right)-NDV{I}_{i,\min }\right)\times \left(FPA{R}_{\max }-FPA{R}_{\min }\right)}{NDV{I}_{i,\max }-NDV{I}_{i,\min }}}+FPA{R}_{\min },

where NDVI (x, t) is the NDVI of pixel x in month t, NDVI_i,min and NDVI_i,max are the minimum and the maximum NDVI of class i respectively, and FPAR_min and FPAR_max are constant, equal to 0.001 and 0.95 respectively.

Light utilization rate (ϵ) calculation

The results given by Potter et al. show that the light utilization rate of vegetation reaches a maximum under ideal conditions ( Potter et al., 1993 ; Field et al., 1995), and that maximum is affected by temperature (T_ϵ1 and T_ϵ2) and soil moisture (W_ϵ(x, t)) in the field:

(10) \mathit{ϵ}\left(x,t\right)=T_{ϵ}\left(x,t\right)\times T_{ϵ}\left(x,t\right)\times W_{\mathit{ϵ}}\left(x,t\right)\times {\mathit{ϵ}}^{*},

where ϵ^* is the largest light use efficiency under ideal conditions, T_ϵ1 and T_ϵ2 are described as:

(11) T_{ϵ1}\left(x,t\right)=0.8+0.02T_{\mathrm{o}\mathrm{p}\mathrm{t}}\left(x,t\right)-0.0005{\left[T_{\mathrm{o}\mathrm{p}\mathrm{t}}\left(x,t\right)\right]}^2.

(12) T_{\mathit{ϵ}2}\left(x,t\right)={\displaystyle \frac{1.184}{\left\{1+e^{0.2\times \left(T_{\mathrm{o}\mathrm{p}\mathrm{t}}\left(x,t\right)-10-T\left(x,t\right)\right)}\right\}\times \left\{1+e^{0.3\times \left(-T_{\mathrm{o}\mathrm{p}\mathrm{t}}\left(x,t\right)-10+T\left(x,t\right)\right)}\right\}}},

where T_opt and T are the average temperatures with the largest NDVI in a month and the average temperatures in a month respectively.

W_ϵ(x, t) ranges from 0.5 (extremely arid) to 1 (extremely humid) ( Piao et al., 2001), and is calculated as:

(13) W_{\mathit{ϵ}}\left(x,t\right)=\mathrm{0.5}+\mathrm{0.5}\times EET\left(x,t\right)/PET\left(x,t\right),

where EET(x, t) is regional actual evapotranspiration, which can be calculated through the following model established by Zhou and Zhang ( Zhou and Zhang, 1996).

(14) EET\left(x,t\right)={\displaystyle \frac{P\left(x,t\right)\times R_n\left(x,t\right)\times \left[P{\left(x,t\right)}^{\mathrm{2}}+R_n{\left(x,t\right)}^{\mathrm{2}}+P\left(x,t\right)\times R_n\left(x,t\right)\right]}{\left[P\left(x,t\right)+R_n\left(x,t\right)\right]\times \left[P{\left(x,t\right)}^{\mathrm{2}}+R_n{\left(x,t\right)}^{\mathrm{2}}\right]}},

where P(x, t) is the rainfall of pixel x in month t, and R_n(x, t) is the net radiation of pixel x in month t.

Similarly, PET₀(x, t) is the regional potential evapotranspiration, which can be calculated according to the vegetation-climatic model established by Thornthwaite ( Zhang, 1989).

(15) PE{T}_{\mathrm{0}}\left(x,t\right)=\{\begin{array}{ll}\mathrm{16}C{\left({\displaystyle \frac{\mathrm{10}{T}_{ef}}{I}}\right)}^a\hfill & \mathrm{0}\le T\le {\mathrm{26.5}}^{}\mathrm{°}\mathrm{C}\hfill \\ C\left(-\mathrm{415.85}+\mathrm{32.24}{T}_{ef}-\mathrm{0.43}{T}_{ef}^{\mathrm{2}}\right)\hfill & T>{\mathrm{26.5}}^{}\mathrm{°}\mathrm{C}\hfill \end{array},

(16) a=\mathrm{0.49}+\mathrm{0.0179}I-\mathrm{0.0000771}{I}^{\mathrm{2}}+\mathrm{0.000000675}{I}^{\mathrm{3}},

where C=N/360, I is thermal index, and T_ef is effective temperature.

Determination of vegetation types

Four kinds of woodland types were present within the study area, including deciduous coniferous forest, deciduous broadleaf forest, mixed forest, and evergreen coniferous forest. It should be noted that all four of these woodland types were classified as a single group in the classification experiment. Based on Zhu’s previous simulation of the maximal light utilization rate of different vegetation types in China ( Zhu et al., 2007), we select the average of the four maximal light utilization rates corresponding to the four types of woodland to represent the maximal light utilization rate of woodland overall (0.5103 gC·MJ^‒1). The maximal light utilization rate of lawn and farmland is equal to 0.542 gC·MJ^‒1. Other types of cover, including architecture, water area and badlands, were not addressed in the previous study, so a fixed value (0.389 gC·MJ^‒1) was applied to those classifications in the CASA model for this experiment.

Results and discussion

Comparison of NPP on two scales among different years

We first compared FPAR absorbed by canopy, light utilization rate of vegetation (ϵ), and estimated NPP determined for different spatial scales in different years. As described in Table 1, the June NPP estimates differ in the same year when calculated at different spatial scales, decreasing at greater scales. Both parameters used to determine NPP (FPAR and ϵ) were also influenced by the same scale effect. FPAR showed high variation based on the use of different sources of NDVI data at given time points, while the light utilization rate of vegetation was slightly influenced. As a whole, we can see from Fig. 3 that the average NPP of terrestrial vegetation in Xuzhou generally decreased from 2001‒2010 when measured at either of the two spatial scales. The exception to this trend was that the average NPP of terrestrial vegetation increased in 2009 relative to 2008. This may be due to climate effects, as the average rainfall of Xuzhou was 137.9 mm and solar radiation quantity was 530.167 MJ·m^‒2·month^‒1 in 2009 compared to the 116.3 mm and 417.4766 MJ·m^‒2·month^‒1, respectively, in 2008.

Comparison of the temporal and spatial variations in NPP

null

To analyze the temporal and spatial variations in NPP at two spatial scales, the whole study area was plotted into four sub-regions using equidistant segmentation, designated as having highest NPP sub-region (>105 gC·m^‒2·month^‒1), moderately high NPP sub-region (70‒105 gC·m^‒2·month^‒1), moderately low NPP sub-region (35‒70 gC·m^‒2·month^‒1), and lowest NPP sub-region (0‒35 gC·m^‒2·month^‒1). The minimum and maximum NPP for the MODIS-scale data were 0 and 140.63 gC·m^‒2·month^‒1 respectively, and were 0 and 143.92 gC·m^‒2·month^‒1, respectively, for the ETM+ scale data. NPP distributions from 2001 to 2010 simulated by CASA are shown in Figs. 3 and 4As illustrated in Table 2 and Fig. 6, the average NPP showed the same trends at both scales within each sub-region. The relationship of average NPP over time at the two scales was exactly the same for the highest and moderately high NPP sub-regions, but the lowest and moderately low NPP sub-regions showed little change across study years. Average NPP in the highest and moderately high sub-regions shifted strongly over the study period. In the highest NPP sub-region, average NPP reached its maximum in 2005 and its minimum in 2008 at both MODIS and ETM+ scales. In the lowest NPP sub-region, there was little change in average NPP from 2001 to 2005. From 2006 to 2010, MODIS scale data showed a decreasing trend in NPP, while under ETM+ scale data, maximum NPP occurred in 2001 and the minimum in 2009. In general, however, the average NPP for the same sub-region during the same time had few changes in the different spatial scales.

As shown in Table 3, the structures of the four NPP sub-regions vary between the two spatial scales. The lowest and moderately low NPP sub-regions of the study area increased at the higher scale in almost all years, while the other two sub-regions decreased. Furthermore, the area percentages of the moderately low and moderately high NPP sub-regions vary greatly at the larger scale. The highest NPP sub-region covers the smallest area of all four sub-regions, indicating that the NPP estimates in June for the entire study area were estimated fairly well by considering only the moderately low and moderately high NPP sub-regions.

Comparison of the NPP of different vegetation types

Comparisons of the NPP of farmland, woodland, and lawn showed significant decreases when measured using MODIS data. The NPP of woodland reached its maximum in 2009 and 2010. In other years, the NPP of farmland, lawn, and woodland decreased with relatively increasing magnitude when using ETM+ data, leading to higher total woodland NPP than for total lawn NPP. Overall, scale had a significant effect on the NPP of different vegetation types, especially for woodland. A previous study has shown that the NPP of woodland is higher than that of farmland and lawn because of its greater carbon sequestration capacity (Harcombe et al., 1993). However, few areas of woodland were included in the present study area, but were distributed sporadically in the urban area or building zones, resulting in a heterogeneous spatial structure. This type of pattern can cause individual pixels of RS data to encompass multiple categories of data (i.e., “mixed pixels”). Therefore, woodland NPP likely had a lower estimated value due to the presence of mixed pixels. Based on the NPP estimates at different spatial scales, it appears that urban woodland NPP had a high sensitivity to scale effects, showing an increase when using larger-scale data to investigate the study area.

Effects of urbanization on NPP

NPP distribution from 2001 to 2010 simulated by CASA is shown in Fig. 3 and Fig. 4. During this study period, mean NPP decreased from 51.54 to 42.63 gC·m^‒2·month^‒1 using MODIS data. Many factors can influence NPP, such as climate (solar radiation, temperature, and precipitation), land use, land cover, and human disturbance. Table 4 shows that no obvious changes in forest cover occurred from 2001 to 2007, suggesting that the most significant cause for total NPP reduction was urbanization due to local economic and population growth. The GDP of Xuzhou has increased by about 17% annually during the study period, while the population has increased by about 70 thousand people. The reduction in mean NPP was therefore mainly caused by increasing human disturbance and land use change. Most of the forest was destroyed and transformed to bare land, residential area, cropland, etc. during the study period. For example, examining the classification map of Landsat data for the study period showed that a new city had been constructed to the southeast of Xuzhou, replacing the forest area. The amount of forestland destroyed increased in parallel with increased population and economic growth. With greater attention being given to environmental conservation by the local government, forest planting began in areas surrounding Xuzhou in 2008. According to the forest inventory data and field survey, a sharp increase in forest coverage occured from 2008 to 2010. However, the young reestablished forest has a lower Leaf Area Index (LAI) than the removed mature forest. Nevertheless, this appears to be the main indication that NPP began increasing in the study area near the end of the study period.

Conclusions

The NPP estimation model established in this study was used to estimate NPP in the city of Xuzhou in June of each year from 2001 to 2010. We analyzed temporal and spatial variations in the estimates, and found that 1) the NPP of the study area decreased when estimates were based on a greater spatial scale, 2) NPP of vegetation in the study area showed a consistent, decreasing trend at both spatial scales over time, and 3) this trend reflects changes in climate and environment. In particular, urban development was an important factor in overall regional NPP change. During the past decade, land use and land cover have changed in the Xuzhou area due to urbanization. Human settlements have displaced large areas of forest, decreasing the overall areas of forest and cropland in the study area. Based on the spatial structure of NPP, the proportion of the study area accounted for by the lowest NPP sub-region increased when larger scale data were used for estimation, implying that the heterogeneous nature of urban areas creates mixed pixels in spatial data. These pixels mask the presence of small areas of vegetation as the scale increases. In this study, spatial scale had a large effect on NPP estimates for different vegetation types, especially woodland; woodland NPP was directly impacted by mixed pixels in the urbanized study area, leading to decreases in woodland NPP estimates when larger spatial scale data was used.

References

[1] Del Grosso S, Parton W, Stohlgren T, Zheng D, Bachelet D, Prince S, Hibbard K, Olson R (2008). Global potential net primary production predicted from vegetation class, precipitation, and temperature. Ecology, 89(8): 2117–2126

[2] Dong D, Ni J (2011). Modeling changes of net primary productivity of karst vegetation in southwestern China using the CASA model. Acta Phytoecol Sin, 7(31): 1855–1866

[3] Duffie J A, Beckman W A (2006). Solar engineering of thermal processes. New York: Wiley

[4] Fang J (2000). Global Ecology. Beijing: Higher Education Press

[5] Feng X, Liu G, Chen J M, Chen M, Liu J, Ju W M, Sun R, Zhou W (2006). Net primary productivity of China’s terrestrial ecosystems from a process model driven by remote sensing. Journal of environmental management, 85(3): 563–573

[6] Field C B, Randerson J T, Malmström C M (1995). Global net primary production: combining ecology and remote sensing. Remote Sens Environ, 51(1): 74–88

[7] Franklin S E, Lavigne M B, Deuling M J, Wulder M A, Hunt Jr E R (1997). Estimation of forest leaf area index using remote sensing and GIS data for modelling ner primary production. International Journal of Remote Sensing, 18(16): 3459–3471

[8] Hazarika M K, Yasuoka Y, Ito A, Dye D (2005). Estimation of net primary productivity by integrating remote sensing data with an ecosystem model. Remote Sens Environ, 94(3): 298–310

[9] Hejun Y(2006). Study on the formula of the solar declination and time difference in meteorology. Meteorological, Hydrological and Marine Instruments, 3(13): 011

[10] Huang F, Qi X, Xu S (2011). Monitoring NPP changes in Changbai Mountain area, China with MODIS images. Geoinformatics, 2011 19th International Conference on.Shanghai, 1–4

[11] Imhoff M L, Bounoua L, DeFries R, Lawrence W T, Stutzer D, Tucker C J, Ricketts T (2004). The consequences of urban land transformation on net primary productivity in the United States. Remote Sens Environ, 89(4): 434–443

[12] Li Y P, Ji J J (2001). Model estimates of global carbon flux between vegetation and the atmosphere. Adv Atmos Sci, 5(18): 807–818

[13] Liu J, Chen J, Cihlar J, Chen W (1999). Net primary productivity distribution in the BOREAS region from a process model using satellite and surface data. J Geophys Res, 104(D22): 27735–27754

[14] Liu Y, Ju W, He H, Wang S, Sun R, Zhang Y (2013). Changes of net primary productivity in China during recent 11 years detected using an ecological model driven by MODIS data. Front Earth Sci, 7(1): 112–127

[15] Lu D, Xu X, Tian H, Moran E, Zhao M, Running S (2010). The effects of urbanization on net primary productivity in southeastern China. Environ Manage, 46(3): 404–410

[16] Melgani F, Bruzzone L (2004). Classification of hyperspectral remote sensing images with support vector machines. IEEE Transactions on Geoscience and Remote Sensing, 42(8): 1778–1790

[17] Neitsch S L, Arnold J G, Kiniry J R, 2002. Soil and water assessment tool theoretical documentation version 2000. Grassland, Soil & Water Research Laboratory, TEMPLE, TEXAS

[18] Piao J Y, Fang S L, Guo Q H (2001). Application of CASA model to the estimation of Chinese terrestrial net primary productivity. Acta Phytoecol Sin, 25(5): 603–608

[19] Potter C S, Randerson J T, Field C B, Matson P A, Vitousek P M, Mooney H A, Klooster S A (1993). Terrestrial ecosystem production: a process model based on global satellite and surface data. Global Biogeochem Cycles, 7(4): 811–841

[20] Pranuthi G, Dubey S K, Tripathi S K (2012). Comparison of different models for estimation of net primary productivity in India. Journal of Agrometeorology, 2(14): 105–110

[21] Prieto-Blanco A, North P R, Barnsley M J, Fox N (2009a). Satellite-driven modelling of Net Primary Productivity (NPP): theoretical analysis. Remote Sens Environ, 113(1): 137–147

[22] Prieto-Blanco A, North P R J, Barnsley M J, Fox N (2009b). Satellite-driven modelling of Net Primary Productivity (NPP): Theoretical analysis. Remote Sens Environ, 113(1): 137–147

[23] Ruimy A, Saugier B, Dedieu G (1994). Methodology for the estimation of terrestrial net primary production from remotely sensed data. J Geophys Res, 99(D3): 5263–5283

[24] Sasai T, Saigusa N, Nasahara K N, Ito A, Hashimoto H, Nemani R, Hirata R, Ichii K, Takagi K, Saitoh T M, Ohta T, Murakami K, Yamaguchi Y, Oikawa T (2011). Satellite-driven estimation of terrestrial carbon flux over Far East Asia with 1-km grid resolution. Remote Sens Environ, 115(7): 1758–1771

[25] Singh R, Rovshan S, Goroshi S, Panigrahy S, Parihar J (2011). Spatial and Temporal Variability of Net Primary Productivity (NPP) over Terrestrial Biosphere of India Using NOAA-AVHRR Based GloPEM Model. Journal of the Indian Society of Remote Sensing, 39(3): 345–353

[26] Sun R, Zhu Q (1999). Net primary productivity of terrestrial vegetation: a review on related researches. Chinese Journal of Applied Ecology, 6(10): 757–760

[27] Schuol J, Abbaspour K C (2007). Using monthly weather statistics to generate daily data in a SWAT model application to West Africa. Ecological modeling, 201(3): 301–311

[28] Tan K, Du P J (2008). Hyperspectral remote sensing image classification based on support vector machine. Journal of Infrared and Millimeter Waves, 27(2): 123–128

[29] Tian G, Qiao Z(2014). Assessing the impact of the urbanization process on net primary productivity in China in 1989–2000. Environmental Pollution, 184: 320–326

[30] Turner D P, Ritts W D, Cohen W B, Gower S T, Running S W, Zhao M, Costa M H, Kirschbaum A A, Ham J M, Saleska S R, Ahl D E (2006). Evaluation of MODIS NPP and GPP products across multiple biomes. Remote Sens Environ, 102(3): 282–292

[31] Xu C, Liu M, An S, Chen J, Yan P (2007). Assessing the impact of urbanization on regional net primary productivity in Jiangyin County, China. J Environ Manage, 85(3): 597–606

[32] Yuan Q Z, Wu S H, Zhao D S, Dai E F, Chen L, Zhang L (2014). Modeling net primary productivity of the terrestrial ecosystem in China from 1961 to 2005. J Geogr Sci, 24(1): 3–17

[33] Zhang X S (1989). The potential evapotranspiration (PE) index for vegetation and vegetation-climatic classification (2)-an introduction of main methods and pep program. Acta Phytoecol Sin, 3(13): 197–207

[34] Zhao J, Yan X, Zhu Y (2007). Advance in research on net primary productivity of terrestrial vegetation. J Desert Res, 5(27): 780–786

[35] Zhao M S, Heinsch F A, Nemani R R, Running S W (2005). Improvements of the MODIS terrestrial gross and net primary production global data set. Remote Sens Environ, 95(2): 164–176

[36] Zhou G S, Zhang X S (1996). Study on climate-vegetation classification for global change in China. Acta Bot Sin, 1(38): 8–17

[37] Zhu W Q, Pan Y Z, Zhang J S (2007). Estimation of net primary productivity of Chinese terrestrial vegetation based on remote sensing. Journal of Plant Ecology, 3(31): 413–424

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg