Abstract

Net primary productivity (NPP) is an important component of the carbon cycle and a key indicator of ecosystem performance. The aim of this study is to construct a more accurate regional vegetation NPP estimation model and explore the relationship between NPP and climatic factors (air temperature, rainfall, sunshine hours, relative humidity, air pressure, global radiation, and surface net radiation). As a key variable in NPP modeling, photosynthetically active radiation (PAR) was obtained by finding a linear relationship between PAR and horizontal direct radiation, scattered radiation, and net radiation with high accuracy. The fraction of absorbed photosynthetically active radiation (FPAR) was estimated by enhanced vegetation index (EVI) instead of the widely used normalized difference vegetation index (NDVI). Stress factors of temperature/humidity for different types of vegetation were also considered in the simulation of light use efficiencies (LUE). The authors used EVI datasets of Moderate Resolution Imaging Spectroradiometer (MODIS) from 2001 to 2011 and geographic information techniques to reveal NPP variations in Wuhan. Time lagged serial correlation analysis was employed to study the delayed and continuous effects of climatic factors on NPP. The results showed that the authors’ improved model can simulate vegetation NPP in Wuhan effectively, and it may be adopted or used in other regions of the world that need to be further tested. The results indicated that air temperature and air pressure contributed significantly to the interannual changes of plant NPP while rainfall and global radiation were major climatic factors influencing seasonal NPP variations. A significant positive 32-day lagged correlation was observed between seasonal variation of NPP and rainfall (P < 0.01); the influence of changing climate on NPP lasted for 64 days. The impact of air pressure, global radiation, and net radiation on NPP persisted for 48 days, while the effects of sunshine hours and air temperature on NPP only lasted for 16 and 32 days, respectively.

1. Introduction

Net primary productivity (NPP), the difference between photosynthesis or gross primary productivity and autotrophic respiration, is an important variable in biological and chemical carbon cycle and a key ecological indicator of the sustainable development of terrestrial ecosystem. Therefore, study on NPP is one of the key focuses on global carbon balance and climate change in scientific community (Potter et al. 2003; Tao and Zhang 2010). Many experiments have been carried out since the late 1980s, and a series of achievements has also been made. However, the majority of the literature has been focused on the global or national scale, and NPP is highly variable in space and time because of the differences or inconsistencies in natural and anthropogenic factors (Loescher et al. 2003; Matsushita et al. 2004; Peng et al. 2008). A clear and concise description for the correlation between regional NPP and climate change is still crucial for the research on the relationship between climate change and carbon cycling processes.

NPP is sensitive to many controls, and climate plays a major role in NPP spatial and temporal variation (Zhao et al. 2006). Climatic factors affect vegetation NPP in a variety of ways during photosynthesis: for example, lower rainfall or cloudiness may decrease NPP by lowering soil moisture in dry regions, while increased cloudiness may also decrease NPP by reducing the availability of photosynthetically active radiation (PAR) in moist regions (Beer et al. 2010). Therefore, climate-based ecosystem models have been developed as a means of quantifying spatiotemporal variations in NPP, which can be basically classified into three types: (i) statistical methods, such as the Miami model (Foley 1994); (ii) parametric models, such as the Global Production Efficiency Model (GLO-PEM) (Cao et al. 2004); and (iii) process models, such as the Biome Biogeochemical Cycles Model (BIOME-BGC) (Running et al. 2004). These methods have been widely applied for estimating the annual or seasonal terrestrial NPP fluxes in different areas, and many researchers have focused on comparing the results from different models for model validation (Cramer et al. 1999; Hemming et al. 2011).

Satellite remote sensing provides consistent and systematic observations of vegetation and ecosystems and plays an increasing role in characterization of vegetation structure and seasonal dynamics of terrestrial NPP (Donmez et al. 2011). Thus, these NPP models can also be used to examine the effect of climate variability on NPP (Nayak et al. 2012). Cao et al. (2004) studied the interannual variations and trends in global terrestrial NPP from 1981 to 2000 using GLO-PEM. Mohamed et al. (Mohamed et al. 2004) explored the patterns of interannual variability of NPP in relation to potential causal factors. Fang et al. (Fang et al. 2001) also examined the relationship between interannual variability in NPP and precipitation across China with the Carnegie–Ames–Stanford Approach (CASA).

However, because of a complex climate system and vegetation conditions, these NPP models may not be appropriate in all regions and the process parameters should be optimized, such as PAR fraction (Gong et al. 2012). Moreover, it is clear that the normalized difference vegetation index (NDVI) suffers several limitations, including sensitivity to atmospheric conditions, sensitivity to soil background, and saturation of NDVI values in multilayered and closed canopies, so the NDVI-based vegetation NPP estimation should also be modified (Xiao et al. 2006; Zhang et al. 2009). In addition, we still lack a clear understanding of the quantitative relationship between climate change and NPP in central China, although a great deal of related research work has been launched in China. The purpose of this paper is to map the interannual and seasonal characterization of NPP and climate variables using our improved NPP estimation method that optimized the parameter of PAR, fraction of absorbed photosynthetically active radiation (FPAR), and light use efficiencies (LUE) and extend the lagged cross-correlation analysis method to study the delayed and continuous effects of seasonal climate on NPP for different kinds of vegetation with a temporal resolution of 16 days in Wuhan.

2. Study area and data

2.1. Study area

Geographically situated between 29°58′–31°22′N and 113°41′–115°05′E, Wuhan lies in central China (Hubei Province) with the Yangtze River and Han River flowing across the city, which was made up by three parts: Wuchang, Hankou, and Hanyang (Figure 1) (Chen and Masser 2003). The typical subtropical monsoon climate here is influenced by its geographic location close to Yangtze River, which is characterized by well-marked seasons with cold and dry winter, abundant rainfall, and sunshine in summer. It is generally higher than 30°C in summer with extreme maximum temperature of 44.5°C; thus, Wuhan was called as one of China’s four furnaces (Hao 2006). The statistics showed that annual grass vegetation (AGV) was the dominant land-cover type (about 70% of study area), and the percentage of buildup was about 6% in 2010 (Zhu and Lin 2008; Gong et al. 2012). It is a transition area between subtropical evergreen broadleaf vegetation and temperate deciduous broadleaf vegetation with Masson’s pine, fir, and oak distributing in this area (Figure 2), and evergreen broadleaf vegetation (EBV), deciduous broadleaf vegetation (DBV), and annual broadleaf vegetation (ABV) took up 3%, 3%, and 4% of the whole area, respectively (Cen et al. 2008; Kabba and Li 2011). With increasing requirements on improving the ecological environment, we studied the NPP variations and their relationship with climate to get an overall understanding of the vegetation ecosystem in Wuhan.

Figure 1.

Location of the study area: (a) location of Hubei Province in China, (b) location of Wuhan in Hubei Province, and (c) the geographical distribution pattern in Wuhan.

Figure 1.

Location of the study area: (a) location of Hubei Province in China, (b) location of Wuhan in Hubei Province, and (c) the geographical distribution pattern in Wuhan.

Figure 2.

Spatial distribution of vegetation covers in Wuhan (ENV, EBV, DNV, DBV, ABV, AGV, and GV).

Figure 2.

Spatial distribution of vegetation covers in Wuhan (ENV, EBV, DNV, DBV, ABV, AGV, and GV).

2.2. Data and preprocessing

The main daily meteorological datasets from six measurement sites (Figure 2) used in this study were obtained from China Meteorological Data Sharing Service System, including air temperature, rainfall, sunshine hours, relative humidity, air pressure, horizontal direct radiation, global radiation, and surface net radiation. PAR, global radiation, direct solar radiation, scattered radiation data, etc., from 2009 to 2012 were measured by the Kipp & Zonen radiation measurement system that has been calibrated strictly in September 2010. Each climatic variable was checked carefully and averaged every 16 days.

The Moderate Resolution Imaging Spectroradiometer (MODIS) sensor on board the Terra and Aqua satellites has 36 spectral bands; 7 spectral bands are primarily designed for study of vegetation and land surface: blue (459–479 nm), green (545–565 nm), red (620–670 nm), near infrared (841–875 nm and 1230–1250 nm), and shortwave infrared (1628–1652 nm and 2105–2155 nm) (Gobron et al. 2000). MODIS 16-day composite data at 250-m spatial resolution (MOD13Q1) from 2001 to 2011 were downloaded from the National Aeronautics and Space Administration (NASA) website (http://modis.gsfc.nasa.gov/); each composite MOD13Q1 pixel was temporally filtered to the closest nadir view angle with the least cloud or cloud shadow and the lowest aerosol loading (Zhao et al. 2005). The MOD12Q1 land-cover product was used to classify the vegetation types with four land-cover classification schemes and then assigned the maximum LUE (ɛmax) for different vegetation types according to a newly simulated result in China (Zhu et al. 2006), which have been used widely in China (Zhou et al. 2010). All the climatic and remote sensing data were interpolated and resampled to get the same spatial resolution with MOD13Q1 by universal kriging and image resampling methods (Nencini et al. 2007; Falivene et al. 2010). Using Environment for Visualizing Images (ENVI) and ArcGIS software, a 16-day composite enhanced vegetation index (EVI) was acquired after converting all the MODIS data to geographic coordinate system and World Geodetic System 1984 (WGS84) data, which were used to further calculate FPAR.

3. Methodology

3.1. Modeling and mapping NPP

Accurate estimation of NPP is critical for understanding the carbon dynamics within the atmosphere–vegetation–soil continuum and the response of the terrestrial ecosystem to climate (Wang and Zhao 2008). It is considered that 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 (Liu et al. 2001). Vegetation NPP may predominantly be affected by two variables of absorption photosynthetic active radiation (APAR) and light use efficiency ɛ (Nemani et al. 2003),

 
formula

where PAR is incident photosynthetically active radiation in a time period (day or month), FPAR is the fraction of PAR absorbed by vegetation canopy, and ɛ() represents the actual radiation conversion efficiency. We can see from above that PAR, FPAR, and ɛ() are most crucial parameters in NPP estimation, and we try to improve the accuracy of each variable in regional scale in following parts.

3.1.1. Estimation of PAR

PAR refers to the solar radiation spectrum in the wavelength range of 400–700 nm that plants use for the photosynthetic process, which is a key variable required by almost all terrestrial ecosystem models (Janjai and Wattan 2011). At present, a worldwide routine network of PAR measurements is not yet established, PAR was estimated by the relationship between global radiation and PAR, and the widely used PAR fraction was about 0.5 (PAR is half of global radiation) (Potter et al. 2003; Piao et al. 2006). In fact, PAR used by vegetation canopy is lower than half of the global radiation, and PAR fraction varies with space and time, which is not suitable for regional NPP simulation (Gitelson et al. 2012). So we developed a new method for regional PAR calculation by finding a relationship between PAR and horizontal direct radiation, scattered radiation, and surface net radiation. It is known that PAR changes greatly under different weather conditions, so we divided the daily measured datasets from September 2009 to July 2012 into two parts with daily global radiation 120 W m−2 being the cutoff point to establish a more reliable and accurate PAR estimation model. Linear regression between modeled and observed PAR every day from September 2009 to July 2012 in Wuhan (Figure 3) with the relative error only being 4.6 W m−2, and the correlation coefficient was obtained; we also compared our simulated PAR with measured data every month in 2010, which can prove the accuracy of the new model (Figure 4),

 
formula

where Qg is daily global radiation; QD is daily scattered radiation; QH is horizontal direct radiation that can be inferred from QH = QR × sinh (QR represents the measured direct solar radiation and h is sun elevation angle); and QN is surface net radiation, which also can be obtained by the formula QN = Qg + QQQs (Q, Q, and Qs represent downward longwave radiation, upward longwave radiation, and reflected shortwave radiation, respectively).

Figure 3.

The linear correlation of modeled PAR and measured data.

Figure 3.

The linear correlation of modeled PAR and measured data.

Figure 4.

The monthly comparison chart of modeled and observed PAR in 2010.

Figure 4.

The monthly comparison chart of modeled and observed PAR in 2010.

3.1.2. Estimation of FPAR

FPAR measures the proportion of PAR that is absorbed by vegetation canopies in the spectrum of 400–700 nm. It is a key parameter for global vegetation (GV) and crop yield modeling (Lobell et al. 2002). For decades, numerous studies have focused on estimating FPAR by leaf area index that is closely correlated with satellite-derived vegetation indices: for example, NDVI, which is calculated as a normalized ratio between red band ρred and near-infrared band ρnir (Wardlow and Egbert 2008),

 
formula

In remote sensing analysis, FPAR is usually estimated as a linear or nonlinear function of NDVI (Field et al. 1998; Kalfas et al. 2011): for example, researchers discovered an NDVI–FPAR relationship from NASA MOD15 (Myneni et al. 2002),

 
formula

These NDVI-based NPP models have been applied at regional and global scale using Advanced Very High Resolution Radiometer (AVHRR) or MODIS products. However, it is well known that NDVI has several limitations, including saturation in a multilayer closed canopy and sensitivity to both atmospheric aerosols and the soil background (Huete et al. 2002). In addition, vegetation canopies are composed of photosynthetically active vegetation (PAV; mostly green leaves) and nonphotosynthetically active vegetation (NPV; mostly senescent foliage and branches), and NPV has a significant effect on FPAR at the vegetation canopy level. Moreover, individual green leaves also have some proportion of NPV at the leaf level, depending on leaf age, species, and leaf morphology (Asner et al. 1998; Hanan et al. 2002). Thus, vegetation FPAR should be

 
formula

FPARPAV represents the fraction of PAR absorbed by PAV that is only used for vegetation photosynthesis, which is a key point in FPAR estimation (Xiao et al. 2005). To overcome above shortcomings and obtain FPARPAV accurately, EVI was developed (Huete et al. 1997), which was defined as

 
formula

where G = 2.5, C1 =6, C2 =7.5, and L = 1; the blue band was mainly used for atmospheric correction in aerosols and variable soil and canopy background. Note that EVI is linearly correlated with the green leaf area index and sensitive to canopy variations (Boegh et al. 2002). Xiao et al. (Xiao et al. 2004) developed a linear function between FPARPAV and EVI for modeling GPP and NPP, which has been successfully implemented in different areas, such as the Amazon, North America, and East Asia (Xiao et al. 2005; Wu et al. 2008),

 
formula

where the coefficient α is 1.0 in Xiao’s first version of the Vegetation Photosynthesis Model (VPM), representing the simplest case of parameterization. VPM clearly divided FPAR into FPARPAV and FPARNPV, which was a critical issue in satellite-based vegetation modeling (Xiao et al. 2004). In this paper, we adopted this method to estimate vegetation FPAR for studying the relationship between NPP and climate change in regional scale.

3.1.3. Estimation of actual light-use efficiency ɛ(x, t)

LUE is another important indicator of plant photosynthesis and a key parameter for remote sensing–based models for monitoring vegetation productivity, which is affected by temperature, water, and leaf phenology (Huemmrich et al. 2010). Many studies have focused on simulating accurately of the actual vegetation light use efficiency in global and regional scale by remote sensing or in situ measurement (Hilker et al. 2008),

 
formula

where T1() and T2() are the temperature stress factors, representing the effects of low temperature and high temperature on maximum light use efficiency, respectively; W() is moisture stress influencing coefficient; and ɛmax is the maximum light use efficiency (Christopher et al. 1995).

The maximum light use efficiency ɛmax is controlled by vegetation types; spatial resolution; and the uniformity of vegetation coverage, which is hard to achieve. A fixed ɛmax value is often used in NPP modeling: for example, 0.389 gC MJ−1 was provided in CASA model, which was not precise enough in the study of vegetation monitoring and has been in dispute in scientific community (Zhu et al. 2006). Although Running et al. (Running et al. 2004) considered that there should be different values of ɛmax for different vegetation types and also simulated ɛmax for 10 kinds of vegetation by BIOME-BGC, it is not proper to adopt the values that were based in North America and not in the global world or China. Zhu et al. (Zhu et al. 2006) developed ɛmax of typical vegetation in China with remote sensing, meteorological data, and measured NPP; ɛmax obtained from their research is fairly reliable for NPP modeling in central China (Zhu et al. 2007), and we took their simulating results in this study (Table 1).

Table 1.

The maximum light utilization efficiency for different vegetation types.

The maximum light utilization efficiency for different vegetation types.
The maximum light utilization efficiency for different vegetation types.

3.2. Statistical regression and time lag serial correlation analysis

We extracted each climate variable and NPP of six vegetation types from 2001 to 2011 within the study area using the above improved method. The correlation between NPP and air temperature, rainfall, sunshine hours, relative humidity, air pressure, global radiation, and net radiation was analyzed.

We adopted an assumption that, if correlation between the coefficient of variation (CV) of NPP and climate factor is significant, the seasonal or interannual fluctuation of NPP is attributed to the temporal variability in these climatic parameters (Fang et al. 2001; Peng et al. 2010). In addition, the response of NPP to climate is not instantaneous and sometimes exerts delayed effects. Time lagged serial correlation analysis incorporates both immediate physiological alterations and delayed biogeochemical adjustments of vegetation because of variable climates (Mohamed et al. 2004; Steele et al. 2005), which was defined as follows:

 
formula

where Sk(x, y) is the sample covariance and Sx and Sy+k represent the standard deviations that are calculated by the following equations:

 
formula

where n is the sample number of xi and yi, k is the number of time lag, and kn/4 and k = n −10. In this study, we focused on the delayed and continuous effects climatic factors on seasonal variations of NPP with a 16-day interval during the year, so n is 23 (365/16) and k = 0, 1, 2, 3, 4, and 5. When k is 0, it means there is no time lag, which expresses immediate NPP response to climate variation. The critical values of the significant positive correlation at P = 0.01 are 0.505, 0.515, 0.525, 0.535, 0.545, and 0.56 when k is 0, 1, 2, 3, 4 and 5, respectively. If the value of the correlation coefficient Rk is greater than the corresponding critical value and Rk is the maximal correlation coefficient at different time lags, then the time lag equals k times 16 days.

The effect of climate change on NPP may persist for some time. In this study, the duration of climate influence on NPP was computed by accumulating values of the time for significant correlation at P = 0.01 between NPP and climatic factors at different time lags. If Rk and Rk+1 are both greater than the critical values of significant correlation coefficients at P = 0.01, the duration of climate influence is 16 days. If Rk, Rk+1, and Rk+2 are all larger than the critical values of significant correlation coefficients (P = 0.01), then the duration time is 32 days; other duration lengths can be achieved by analogy with this method.

4. Results and analysis

4.1. Correlations between interannual variation of NPP and climatic factors

The 11-yr datasets of MODIS and climatic parameters were used to explore the patterns of interannual variability of NPP in relation to potential causal factors in Wuhan. We proved the accuracy of our estimated NPP by comparing with simulated results from different models, as well as measured values. As is shown in Table 2, NPP values of different vegetation types in this study are not only surprisingly approximate to that of Zhang et al. (Zhang et al. 2011) and Gong et al. (Gong et al. 2012), who were focusing on the same study area, but also lie in the range of the measured values and results from other models. Meanwhile, average relative errors (AREs) between modeled and measured NPP values dropped to same extent for most types of vegetation: for example, the ARE decreased from 5% in Zhu et al. (Zhu et al. 2007) to 3.6% in this study for EBV. It is also worth noting that Liu (Liu 2001) also studied the temporal and spatial variation of NPP using multiple models such as GLO-PEM; Carbon Exchange between Vegetation, Soil, and Atmosphere (CEVSA); and CASA in central China. In addition, it is reasonable to have slightly lower NPP values for deduction of the effect FPARNPV on NPP, which indicated that NPP values of different vegetation types were reliable in this study.

Table 2.

Annual average NPP values from different simulated models (gC−2 yr−1).

Annual average NPP values from different simulated models (gC−2 yr−1).
Annual average NPP values from different simulated models (gC−2 yr−1).

As can be seen from (Figure 5), annual NPP values for different vegetation types differed greatly, and EBV showed the highest yearly NPP with an average value of 1004 gC m−2 yr−1 during the last 11 years. DBV, evergreen needleleaf vegetation (ENV), and deciduous needleleaf vegetation (DNV) displayed slightly lower NPP with annual mean values higher than 540 gC m−2 yr−1. NPP of AGV and ABV were almost lower than 500 gC m−2 yr−1 from 2001 to 2011 because of plants physiological and morphological characteristics leading to their lower light utilization efficiencies. Overall, the mean annual vegetation NPP was about 501 gC m−2 yr−1 for large areas of AGV and ABV distributed in the study area. There was a similar variation trend for most of the vegetation types here: NPP obviously grew from 2001 to 2008 and then increased slightly from 2009 to 2011, which was attributed to the combined effect of precipitation, temperature, and sunshine for vegetation growth: for example, significantly more rainfall and a higher temperature in winter and spring were discovered in 2002 after drought in 2000 and 2001, which certainly led to an increase in NPP values; a hot dry summer and a continuous low temperature in winter was observed in 2009, which inevitably brought about low NPP values for most of the plants in Wuhan.

Figure 5.

The yearly NPP values of different vegetation types from 2001 to 2011.

Figure 5.

The yearly NPP values of different vegetation types from 2001 to 2011.

To find the main climatic parameters that influenced the interannual variations in NPP of different vegetation, we studied the correlations between the coefficient of variations (CVs) of NPP and climate factors. As shown in Figure 6, the CVs of NPP for ENV, DNV, ABV, and AGV almost displayed significant negative correlations with that of air temperature and air pressure (r < −0.5); CVs of NPP for most plants showed positive correlations with CV of relative humidity. In addition, negative correlations (significant) were also found between CVs of rainfall, sunshine, global radiation, and NPP for DBV and ABV. Above negative (positive) correlations reflected the impacts from increase (decrease) of water loss on vegetation production.

Figure 6.

Correlations between CVs of annual NPP for different vegetation and main climatic factors.

Figure 6.

Correlations between CVs of annual NPP for different vegetation and main climatic factors.

From above, we can learn that the interannual variations of air temperature, air pressure, rainfall, global radiation, and sunshine all contributed to the interannual fluctuation of NPP to a certain extent. In addition, air temperature and air pressure were the predominant climatic factors that determined the interannual variation of NPP in the study area, which was perfectly consistent with Mohamed et al. (Mohamed et al. 2004). Climatic factors played different roles in NPP of different types of vegetation: for example, NPP of DBV or AGV decreased with higher air temperature but increased with enhancing humidity.

4.2. Correlations between seasonal variation of NPP and climatic factors

Climatic variables generally changed greatly from spring to winter in Wuhan (Figure 7); for example, the lowest air temperature appeared at the beginning of the year with about 3.9°C in January, and the highest temperature was generally seen in mid-August with 30°C. There was also an obvious seasonal vegetation NPP variation pattern, which is depicted in Figure 8: low in winter and spring and high in summer. EBV showed higher NPP all year than other kinds of vegetation, with the smallest value of 6 gC m−2 month−1 in late December and the maximum value of 115.2 gC m−2 month−1 in early July. ABV appeared the minimum NPP throughout the year, with the highest value of 48.7 gC m−2 month−1 appearing in early July. As stated above, the difference in NPP values for different types of vegetation lie in the combined effects of plant ecological mechanisms and climate: for example, increase in plant growth under warm temperatures results in an increase of biomass input to the soil pool, whereas the associated precipitation increases soil moisture and facilitates the transformation of organic matter into readily available inorganic nutrients resulting in the difference of plants productivity (Mohamed et al. 2004; Peng et al. 2008).

Figure 7.

Seasonal variations of main climatic parameters in Wuhan.

Figure 7.

Seasonal variations of main climatic parameters in Wuhan.

Figure 8.

Seasonal variations of NPP for different vegetation types.

Figure 8.

Seasonal variations of NPP for different vegetation types.

Linear regressions of main climatic factors and NPP for six vegetation types at different time lags were conducted, and the maximal correlation coefficients were obtained and considered as the correlation coefficients between one climatic parameter and NPP for each vegetation type. As was shown in Figure 9, CVs of seasonal NPP at 16-day interval exhibited significant positive correlation with that of global radiation and rainfall (P = 0.01) for most types of vegetation. Meanwhile, the CV of NPP for AGV also showed significant positive correlation with that of net radiation. However, no significant correlations were found between CV of NPP with that of air temperature or sunshine.

Figure 9.

Correlations between CVs of seasonal NPP for different vegetation and climatic factors.

Figure 9.

Correlations between CVs of seasonal NPP for different vegetation and climatic factors.

The correlations of mean and CVs of NPP for six vegetation types and those of climatic parameters indicated that NPP showed significant correlation with most of climatic factors in Wuhan at P = 0.01, while global radiation and rainfall were the major elements that determine the seasonal fluctuation of NPP. Not only were the correlations between NPP and global radiation and rainfall significant but the correlations between CVs of NPP for most vegetation types and global radiation and rainfall were significant as well. This revealed that the seasonal variation of NPP was mainly attributed to global radiation and rainfall in Wuhan, and the relatively small fluctuation in global radiation and rainfall considerably determined the seasonal variation of NPP. Significant positive correlation was also found between CVs of net radiation (direct energy source of plants) and NPP for AGV (the dominant plants growing here). Meanwhile, the correlation between mean values of NPP and those of sunshine and air temperature were statistically significant at P = 0.01, but the CVs of NPP and climatic parameters showed no significant correlations; this phenomenon may be attributed to relative stability and less sunshine or temperature resulting in decreasing productivity of NPP, which indicated that, although sunshine and air temperature did not seem to contribute to the seasonal fluctuation of NPP, they remained essential predictors of the NPP value.

4.3. Delayed and continuous effects of climatic parameters on NPP

In this paper, time lagged serial correlation analysis was adopted to study the delayed and continuous effects climatic parameters on seasonal fluctuations of NPP for six vegetation types (Figure 10). It was discovered that NPP of all plants exhibited significant positive correlation at 32-day lag with rainfall (P = 0.01), and the impacts would persist for 64 days in most of the study area. Significant positive correlation at zero lag with 16-day duration was observed between NPP for six vegetation types and seasonal variation of sunshine. Besides, NPP showed significant positive zero lagged correlation with air temperature, and the duration length of air temperature influenced on NPP was 32 days. Furthermore, significant negative correlation was found between air pressure and NPP, and the continuous effects of air pressure would last for 48 days. In addition, significant positive correlation at zero lag with 48-day duration was also observed between NPP and global radiation and net radiation in the study area.

Figure 10.

Lagged correlations of seasonal NPP for different vegetation types vs lagged main climatic parameters in Wuhan.

Figure 10.

Lagged correlations of seasonal NPP for different vegetation types vs lagged main climatic parameters in Wuhan.

According to what is described above, each climatic parameter played a different role in plant growth and NPP formation. Rainfall infiltrates the soil and then is absorbed by the roots of plants and subsequently stored by the canopy and stems, resulting in delayed vegetation growing, which may explain the significant positive correlation at 32-day lag with 64-day duration. Because of the combined effects between the mechanism of vegetation growth and climate, the response of vegetation to sunshine is instantaneous and could not persist for a long time, which should be reason for the significant positive correlation at zero lag with 16-day duration between NPP and sunshine. As we know, solar radiation is the fundamental source of energy for plants on Earth, and plants grow faster at relative higher temperature because of more biomass entering the soil pool and transforming the organic matter into readily available inorganic nutrients under water infiltration. This is the main reason for significant positive zero lagged correlation with global radiation, net radiation, and air temperature situating at the subtropical monsoon region with appropriate air temperature and sufficient rainfall supply. The impact of solar radiation or temperature on plants would last for 32 or 48 days in Wuhan. Meanwhile, the air pressure decreased with increasing temperature generally, which would account for the negative correlation between NPP and air pressure.

5. Discussion and conclusions

In this paper, an improved regional vegetation NPP model was constructed: PAR was estimated accurately by a linear function of horizontal direct radiation, scattered radiation, and surface net radiation; FPAR was simulated by EVI that was sensitive to canopy properties and removed the effect of NPV, which better described the fraction of PAR that is absorbed by vegetation canopies; and light use efficiency was enhanced by adopting a newly developed maximum light use efficiency ɛmax that considered the difference in different types of vegetation. We studied the seasonal and interannual variation of NPP for different types of vegetation in Wuhan from 2001 to 2011 with this improved method. Meanwhile, the role of climatic factors in seasonal variation of NPP was also studied; it was observed that global radiation and rainfall were pivotal parameters that resulted in the seasonal fluctuations of NPP, and a relatively small change in air temperature and air pressure could considerably influence the variation of NPP in Wuhan. At last, we discussed the delayed and continuous effect of climate variability on NPP for different kinds of vegetation to reveal and assess the inconsistency or diversity that climate acted on NPP using time lagged serial correlation analysis method and found that rainfall was the main climatic variable that affected the growth of plants in the study area.

To sum up, we proposed an accurate and operational model to map vegetation NPP in regional scale. At radiation measurement sites distributed all over the world, it is reasonable to say that the accuracy of NPP estimation could be significantly improved, and the model can also be tested and modified when applied to other regions. Moreover, a clear understanding of quantitative relationship between NPP and climatic factors may be obtained to learn more about carbon cycling processes further. At the same time, there are still some problems to be addressed in a later study; for example, more environmental factors should be considered in the process of photosynthesis to acquire a more precise light use efficiency and more attention should be paid to the spatial and temporal resolution of NPP to accurately capture the phenology of the growing season and the effect of climate on NPP.

Acknowledgments

We thank the China Meteorological Administration for providing the meteorological data. This research is supported by the Ministry of Science and Technology of China (973 Program: 2009CB723905 and 2011CB707106) and NSFC (Program: 10978003 and 41127901). We also express our sincere gratitude to all members of the lidar group at Wuhan University.

REFERENCES

REFERENCES
Asner
,
G. P.
,
C. A.
Wessman
, and
S.
Archer
,
1998
:
Scale dependence of absorption of photosynthetically active radiation in terrestrial ecosystems
.
Ecol. Appl.
,
8
,
1003
1021
.
Beer
,
C.
,
M.
Reichstein
,
E.
Tomelleri
,
P.
Ciais
, and
M.
Jung
,
2010
:
Terrestrial gross carbon dioxide uptake: Global distribution and covariation with climate
.
Science
,
329
,
834
835
.
Boegh
,
E.
,
H.
Soegaard
,
N.
Broge
,
C. B.
Hasager
,
N. O.
Jensen
,
K.
Schelde
, and
A.
Thomsen
,
2002
:
Airborne multispectral data for quantifying leaf area index, nitrogen concentration, and photosynthetic efficiency in agriculture
.
Remote Sens. Environ.
,
81
,
179
193
.
Cao
,
M. K.
,
D.
Stephen
, and
S.
Prince
,
2004
:
Remotely sensed interannual variations and trends in terrestrial net primary productivity from 1981–2000
.
Ecosystems
,
73
,
233
242
.
Cen
,
Y.
,
P. C.
Zhang
, and
L.
Shi
,
2008
: Spatial-temporal characteristics of LUCC in Wuhan area using satellite data. International Conference on Earth Observation Data Processing and Analysis, D. Li, J. Gong, and H. Wu, Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 7285), 728555, 10.1117/12.815940.
Chen
,
J.
, and
I.
Masser
,
2003
:
Urban growth pattern modeling: A case study of Wuhan city, PR China
.
Landscape Urban Plann.
,
62
,
199
217
.
Christopher
,
B. F.
,
T. R.
James
, and
M.
Carolyn
,
1995
:
Global net primary production: Combining ecology and remote sensing
.
Remote Sens. Environ.
,
51
,
74
88
.
Cramer
,
W.
,
D. W.
Kicklighter
,
A.
Bondeau
,
B.
Moore
,
G.
Churkina
,
B.
Nemry
,
A.
Ruimy
, and
A.
Schloss
,
1999
:
Comparing global models of terrestrial net primary productivity: Overview and key results
.
Global Change Biol.
,
5
,
1
15
.
Donmez
,
C.
,
S.
Berberoglu
, and
P.
Curran
,
2011
:
Modelling the current and future spatial distribution of NPP in a Mediterranean watershed
.
Int. J. Appl. Earth Obs. Geoinf.
,
13
,
336
345
.
Falivene
,
O.
,
L.
Cabrera
, and
A.
Saze
,
2010
:
Interpolation algorithm ranking using cross-validation and the role of smoothing effect. A coal zone example
.
Comput. Geosci.
,
36
,
512
519
.
Fang
,
J. Y.
,
S. L.
Piao
,
Z.
Tang
,
C.
Peng
, and
W.
Ji
,
2001
:
Interannual variability in net primary production and precipitation
.
Science
,
293
,
1723a
.
Field
,
C. B.
,
M. J.
Behrenfeld
,
J. T.
Randerson
, and
P.
Falkowski
,
1998
:
Primary production of the biosphere: Integrating terrestrial and oceanic components
.
Science
,
281
,
237
240
.
Foley
,
J. A.
,
1994
: Net primary productivity in the terrestrial biosphere: The application of a global model. J. Geophys. Res.,99 (D10), 20 773–20 783.
Gitelson
,
A. A.
,
Y.
Peng
,
J. G.
Masek
,
D. C.
Rundquist
,
S.
Verma
,
J. L.
Hatfield
, and
T.
Meyers
,
2012
:
Remote sensed estimation of crop gross primary production with Landsat data
.
Remote Sens. Environ.
,
121
,
404
414
.
Gobron
,
N.
,
B.
Pinty
,
M. M.
Verstrate
, and
J.
Widlowski
,
2000
:
Advanced vegetation indices optimized for up-coming sensors: Design, performance, and applications
.
IEEE Trans. Geosci. Remote Sens.
,
386
,
2489
2505
.
Gong
,
W.
,
L. C.
Wang
,
A. W.
Lin
, and
M.
Zhang
,
2012
:
Evaluating the monthly and interannual variation of net primary production in response to climate in Wuhan during 2001 to 2010
.
Geosci. J.
,
16
,
347
355
.
Hanan
,
N. P.
,
G.
Burba
,
S.
Verma
,
J. A.
Berry
,
A.
Suyker
, and
E.
Walter
,
2002
:
Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR absorption
.
Global Change Biol.
,
8
,
563
574
.
Hao
,
L. S.
,
2006
: Comparison of climate change in cities of Harbin, Shijiazhuang, Wuhan and Guangzhou. Adv. Climate Change Res.,2, 193–196.
Hemming
,
D.
,
R.
Betts
, and
M.
Collins
,
2011
:
Sensitivity and uncertainty of modelled terrestrial net primary productivity to doubled CO2 and associated climate change for a relatively large perturbed physics ensemble
.
Agric. For. Meteor.
,
170
, 79–88,
doi:10.1016/j.agrformet.2011.10.016
.
Hilker
,
T.
,
N. C.
Coops
,
M. A.
Wulder
,
T. A.
Black
, and
R.
Guy
,
2008
:
The use of remote sensing in light use efficiency based models of gross primary production: A review of current status and future requirements
.
Sci. Total Environ.
,
404
,
411
423
.
Huemmrich
,
K. F.
,
J. A.
Gamon
,
C. E.
Tweedie
,
S. F.
Oberbauer
,
G.
Kinoshita
,
S.
Houston
,
Y.
Webber
, and
W.
Oechel
,
2010
:
Remote sensing of tundra gross ecosystem productivity and light use efficiency under varying temperature and moisture conditions
.
Remote Sens. Environ.
,
114
,
481
489
.
Huete
,
A. R.
,
H. Q.
Liu
,
K.
Batchily
, and
W.
Van
,
1997
:
A comparison of vegetation indices global set of TM images for EOS-MODIS
.
Remote Sens. Environ.
,
59
,
440
451
.
Huete
,
A. R.
,
K.
Didan
,
T.
Miura
,
E. P.
Rodriguez
,
X.
Gao
, and
L.
Ferreira
,
2002
:
Overview of the radiometric and biophysical performance of the MODIS vegetation indices
.
Remote Sens. Environ.
,
83
,
195
213
.
Janjai
,
S.
, and
R.
Wattan
,
2011
:
Development of a model for the estimation of photosynthetically active radiation from geostationary satellite data in a tropical environment
.
Remote Sens. Environ.
,
115
,
1680
1693
.
Kabba
,
V.
, and
J. F.
Li
,
2011
:
Analysis of land use and land cover changes, and their ecological implications in Wuhan, China
.
J. Geogr. Geol.
,
3
,
103
118
.
Kalfas
,
J. L.
,
X. M.
Xiao
,
D. X.
Vanegas
,
S.
Verma
, and
A.
Suyker
,
2011
:
Modeling gross primary production of irrigated and rain-fed maize using MODIS imagery and CO2 flux tower data
.
Agric. For. Meteor.
,
151
,
1514
1528
.
Liu
,
M. L.
,
2001
: Research of land use/land cover change and terrestrial ecosystem vegetation carbon and production cover China (in Chinese). Ph.D.dissertation, Institute of Remote Sensing Applications, Chinese Academy of Sciences, 178 pp.
Lobell
,
D. B.
,
J. A.
Hicke
,
G. P.
Asner
,
C. B.
Field
,
C. J.
Tucker
, and
S.
Los
,
2002
:
Satellite estimates of productivity and light use efficiency in United States agriculture, 1982–98
.
Global Change Biol.
,
8
,
722
735
.
Loescher
,
H. W.
,
S. F.
Oberbauer
,
H. L.
Gholz
, and
D.
Clark
,
2003
:
Environmental controls on net ecosystem-level carbon exchange and productivity in a Central American tropical wet forest
.
Global Change Biol.
,
9
,
396
412
.
Matsushita
,
B.
,
M.
Xu
,
J.
Chen
,
S.
Kameyama
, and
M.
Tamurd
,
2004
:
Estimation of regional net primary productivity (NPP) using a process-based ecosystem model: How important is the accuracy of climate data?
Ecol. Modell.
,
178
,
371
388
.
Mohamed
,
M. A. A.
,
I. S.
Babiker
,
Z. M.
Chen
,
K.
Ikeda
,
K.
Ohta
, and
K.
Kato
,
2004
:
The role of climate variability in the inter-annual variation of terrestrial net primary production (NPP)
.
Sci. Total Environ.
,
332
,
123
137
.
Myneni
,
R. B.
,
S.
Hoffman
,
Y.
Knyazikhin
,
J. L.
Privette
,
R. R.
Nemani
, and
S.
Running
,
2002
:
Global products of vegetation leaf area and fraction absorbed PAR from year one of MODIS data
.
Remote Sens. Environ.
,
82
,
214
231
.
Nayak
,
R. K.
,
N. R.
Patel
, and
V. K.
Dadhwal
,
2012
:
Inter-annual variability and climate control of terrestrial net primary productivity over India
.
Int. J. Climatol.
, 33, 132–142,
doi:10.1002/joc.3414
.
Nemani
,
R. R.
,
C. D.
Keeling
,
H.
Hashimoto
,
W. M.
Jolly
,
S. C.
Piper
, and
C.
Tucker
,
2003
:
Climate-driven increases in global terrestrial net primary production from 1982 to 1999
.
Science
,
300
,
1560
1563
.
Nencini
,
F.
,
A.
Garzelli
,
S.
Baronti
, and
L.
Alparon
,
2007
:
Remote sensing image fusion using the curvelet transform
.
Inf. Fusion
,
8
,
143
156
.
Peng
,
D.-L.
,
J.-F.
Huang
,
C.-X.
Cai
,
R.
Deng
, and
J.-F.
Xu
,
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.
,
50
,
1580
1588
.
Peng
,
D.-L.
, and
Coauthors
,
2010
:
Spatial and seasonal characterization of net primary productivity and climate variables in southeastern China using MODIS data
.
J. Zhejiang Univ. Sci.
,
11B
,
275
285
.
Piao
,
S. L.
,
J. Y.
Fang
, and
J.
He
,
2006
:
Variations in vegetation net primary production in the Qinghai-Xizang Plateau, China, from 1982 to 1999
.
Climatic Change
,
74
,
253
267
.
Potter
,
C. S.
,
S. A.
Klooster
, and
R.
Myneni
,
2003
:
Continental scale comparisons of terrestrial carbon sinks estimated from satellite data and ecosystem modeling 1982–1998
.
Global Planet. Change
,
393
,
201
213
.
Running
,
S. W.
,
R.
Ramakrishna
,
F.
Nemani
,
A.
Heinsch
,
M. S.
Zhao
,
M.
Reeves
, and
H.
Hashimoto
,
2004
:
A continuous satellite-derived measure of global terrestrial primary production
.
Bioscience
,
54
,
547
560
.
Steele
,
B. M.
,
S. K.
Reddy
, and
R.
Nemani
,
2005
:
A regression strategy for analyzing environmental data generated by spatio-temporal processes
.
Ecol. Modell.
,
1812
,
93
108
.
Tao
,
F. L.
, and
Z.
Zhang
,
2010
:
Dynamic responses of terrestrial ecosystems structure and function to climate change in China
.
J. Geophys. Res.
,
115
, G03003, doi:10.1029/2009JG001062.
Wang
,
Q.
,
P.
Zhao
,
H.
Ren
, and
Y.
Kakubari
,
2008
:
Spatiotemporal dynamics of forest net primary production in China over the past two decades
.
Global Planet. Change
,
61
,
267
274
.
Wang
,
Y. Y.
,
M. C.
Yang
,
Y. Z.
Pan
, and
W.
Zhu
,
2005
: Valuation of the organic matter produced by terrestrial vegetation ecosystems in China in 1995 based on remote sensing (in Chinese). Ecol. Environ.,144, 455–459.
Wardlow
,
B. D.
, and
S. L.
Egbert
,
2008
:
Large-area crop mapping using time-series MODIS 250 m NDVI data: An assessment for the U.S. central Great Plains
.
Remote Sens. Environ.
,
112
,
1096
1116
.
Wu
,
W. X.
,
S. Q.
Wang
,
X. M.
Xiao
,
G. R.
Yu
,
Y. L.
Fu
, and
Y.
Hao
,
2008
:
Modeling gross primary production of a temperate grassland ecosystem in Inner Mongolia, China, using MODIS imagery and climate data
.
Sci. China
,
51D
,
1501
1512
.
Xiao
,
X. M.
,
D.
Hollinger
,
J.
Aber
,
M.
Goltz
,
E. A.
Davidson
,
Q. Y.
Zhang
, and
B.
Moore
,
2004
:
Satellite-based Modeling of gross primary production in an evergreen needleleaf forest
.
Remote Sens. Environ.
,
89
,
519
534
.
Xiao
,
X. M.
,
Q. Y.
Zhang
,
S.
Saleska
,
L.
Hutyra
,
P.
Camargo
,
S.
Wofsy
,
S.
Frolking
,
S.
Boles
,
M.
Keller
, and
B.
Moore
,
2005
:
Satellite-based modeling of gross primary production in a seasonally moist tropical evergreen forest
.
Remote Sens. Environ.
,
94
,
105
122
.
Xiao
,
X. M.
,
S.
Boles
,
S.
Frolking
,
C. S.
Li
,
J. Y.
Babu
,
W.
Salas
, and
B.
Moore
,
2006
:
Mapping paddy rice agriculture in South and Southeast Asia using multi-temporal MODIS images
.
Remote Sens. Environ.
,
100
,
95
113
.
Zhang
,
J. L.
,
Y. L.
Hu
,
X. M.
Xiao
,
P. S.
Chen
,
S. J.
Han
,
G. Z.
Song
, and
G.
Yu
,
2009
:
Satellite-based estimation of evapotranspiration of an old-growth temperate mixed forest
.
Agric. For. Meteor.
,
149
,
976
984
.
Zhang
,
N.
,
F. Y.
Mao
, and
W.
Gong
,
2011
:
Estimation of net productivity of vegetation of Wuhan in 2009
.
Geomatics Inf. Sci. Wuhan Univ.
,
36
,
1447
1451
.
Zhao
,
M. S.
,
F. A.
Heinsch
,
R. R.
Nemani
, and
S.
Running
,
2005
:
Improvements of the MODIS terrestrial gross and net primary production global data set
.
Remote Sens. Environ.
,
95
,
164
176
.
Zhao
,
M. S.
,
R. R.
Nemani
, and
S.
Running
,
2006
:
Sensitivity of Moderate Resolution Imaging Spectroradiometer (MODIS) terrestrial primary production to the accuracy of meteorological reanalyses
.
J. Geophys. Res.
,
111
, G01002, doi:10.1029/2004JG000004.
Zhou
,
T.
,
P. J.
Shi
,
G. S.
Jia
,
X. J.
Li
, and
Y. Q.
Luo
,
2010
:
Spatial patterns of ecosystem carbon residence time in Chinese forests
.
Sci. China
,
53
,
1229
1240
.
Zhu
,
H. Y.
, and
A. W.
Lin
,
2008
: Research on land use/cover change of Wuhan based on object oriented image interpretation method. Geoinformatics 2008 and Joint Conference on GIS: The Built Environment and Its Dynamics, L. Liu et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 7144), 71440A, 10.1117/12.812700.
Zhu
,
W. Q.
,
Y. Z.
Pan
, and
H.
He
,
2006
:
Simulation of maximum light use efficiency for some typical vegetation types in China
.
Chin. Sci. Bull.
,
51
,
457
463
.
Zhu
,
W. Q.
,
Y. Z.
Pan
, and
J.
Zhang
,
2007
:
Comprehensive analysis of the impact of climatic changes on Chinese terrestrial net primary productivity
.
Chin. Sci. Bull.
,
5223
,
3253
3260
.