Inter-annual variability and climate control of terrestrial net primary productivity over India

Authors


Abstract

Using satellite observations of Normalized Difference Vegetation Index together with climate data from other sources in a terrestrial biosphere model, inter-annual variability of Net Primary Productivity (NPP) over India during 1981–2006 was studied. It is revealed that the variability is large over mixed shrub and grassland (MGL), moderate over cropland and small over the forest regions. Inter-annual variability of NPP exhibits strong positive coherence with the variability of precipitation, and weak coherence with the variability of temperature and solar radiation. Estimated linear growth rate of annual NPP is 0.005 Pg C Yr−2 which is equivalent to 8.5% over the country during past 25 years. This increase is primarily due to the enhancement of productivity over agricultural lands in the country. NPP has increased over most parts of the country during the early 15-year period (1981–1995) resulting in a 10% growth rate of national NPP budget. On the other hand, the NPP growth rate has been reduced to 2.5% during later 15 years period (1991–2005) owing to large decline of NPP over the Indo-Gangetic plains. Climate had a strong control on NPP growth rate during both the periods. Copyright © 2012 Royal Meteorological Society

1. Introduction

Net primary productivity (NPP) is the fundamental processes in biosphere functioning and is defined as the net accumulation of dry matter by green plants per unit time and space. NPP provides the energy and matter that drives most biotic processes on Earth. Long-term sustenance of NPP over space can also contribute to enrich our planet both biologically and environmentally. Climate controls on NPP fluxes are an issue of vital relevance to human society, mainly because of concerns about the extent to which NPP in managed ecosystem can provide adequate food and fibre for a growing population (Potter et al., 2003).

Terrestrial NPP represents the annual carbon storage from the atmosphere to the biosphere and plays a crucial role in limiting the increasing rate of atmospheric CO2. Monitoring regional carbon storage in the form of NPP is, therefore, indispensable for improving the state of the biosphere's health and system for carbon credit trading (Bonan, 1995; Hunt et al., 1996; Chen et al., 2000). Several studies revealed that the NPP over different parts of the globe have been increasing owing to different reasons (Hicke et al., 2002; Nemani et al., 2003; Cao et al., 2005). Nemani et al. (2003) showed that the NPP of the globe has increased by 6% during past two decades (1982–1999) which could be associated with large positive contributions from the tropical ecosystems. Studies by Cao et al. (2005) revealed that the global NPP has increased by 3.5% between the 1980s and 1990s which was associated with regional variations ranging from a 1.9% increase over South America to 6% over the Africa continent.

Inter-annual variability in carbon storage can be attributed to climate variability, CO2 fertilisation effect and human intervention such as land use changes (Galloway and Melillo, 1998; Fu and Wen, 1999; Gao et al., 2000; Schimel et al., 2001;). Several studies have emphasized that climate has a dominant control over terrestrial primary productivity across the world's ecosystems (Tian et al., 1998, 2003; Potter et al., 2003). For example, the NPP of the Amazon evergreen rain forest exhibits large inter-annual variability driven by climate change: during the El Nino years, which bring hot, dry weather to much of the Amazon region, the ecosystem acts as a source, and in other years it acts as a sink of carbon to the atmosphere (Tian et al., 1998). Studies by Fang et al. (2001) suggested that the inter-annual variability of NPP is strongly related with the inter-annual variability of precipitation over China, while the study by Wen-quan et al. (2006) suggested that the linear increasing trends of 1.17% in NPP over the northeast China transect is due mainly to the increasing temperature.

India is a large country with 329 million hectares of geographical area situated in the tropics between 8 and 38°N latitude and 66 and 100°E longitude. The climate of the country varies from monsoonal in the south to temperate in the north. The country has a diverse vegetation cover. Of the total geographic area of the country (GAC), forest and agricultural land accounts about 21 and 55%, respectively. Many studies are being carried out to estimate the terrestrial NPP and analyse its spatio-temporal variability over India (Hingane, 1991; Dadhwal and Nayak, 1993; Chhbra and Dadhwal, 2004; Nayak et al., 2009). However, most of these studies could not explain the broad spectrum of NPP seasonal variability over the country; they led to different estimates of seasonal and annual NPP budgets over the country owing to the following limitations: (1) studies are carried out for different years; (2) different methodology and datasets are being used, (3) no attempts are made to describe inter-annual variability of NPP over India.

Trends in fraction of absorbed photosynthetically active radiation (fAPAR) over the Indian landmass has also been investigated in the past (Pandya et al., 2004), but it could provide information only on changes in vegetation signal and not absolute magnitude of net primary productivity. Simulation of Net Carbon Exchange (NCE) over monsoon Asia by climate-driven Terrestrial Ecosystem Model (TEM) revealed that carbon accumulation over the Indian landmass is mainly resulted from land-use changes (e.g. cropland establishment) during the 1980s (Tian et al., 2003). This study also concludes that precipitation has dominant control on inter-annual variability in NPP and NCE over land ecosystems in India and other parts of southeast Asia. However, model simulations based on purely climate data and at coarse grid may not justify spatio-temporal dynamics of NPP at regional scale. On the other hand, efficiency-based models such as the Carnegie–Ames–Stanford Approach (CASA; Potter et al., 2003) is amenable to satellite measure of actual greenness as input for estimating NPP at finer grid scale. Furthermore, a direct estimation and comprehensive analysis on inter-annual variability of net primary productivity over the country has not been adequately investigated. Therefore, in this paper, high spatial resolution of satellite observations of Normalized Vegetation Index (NDVI) and other climatic data sources are used in the CASA to estimate the NPP over the Indian subcontinent during 1981–2006. The spatial-temporal distribution of NPP over the subcontinent and its response to climate change is also analysed.

2. Materials and methods

2.1. Model

The CASA, a simple terrestrial biosphere model based on light-use efficiency (LUE) simulates terrestrial NPP by estimating optimal metabolic rates of carbon fixation processes under the limiting effect of time while varying inputs of satellite greenness index (NDVI), surface solar radiation (SOLR), air temperature (TEMP), precipitation (PPT), plus soil and vegetation cover maps (Potter et al., 2003). What follows is the governing equation of NPP used in the model.

equation image(1)

Where PAR is the photosynthetically active radiation (MJ) within 400–700 nm wavelengths, fAPAR is the fraction of Absorbed Photosyntheically Active Radiation, and ε* is the maximum light use efficiency. T1, T2, and Ws are stress scalars that reduce ε*. The T1 and T2 represent monthly deviations from site-specific optimal temperature and from 20 °C, respectively, and Ws refer to monthly scale relative soil moisture deficit based on difference between actual and potential evapotranspiration determined by soil water balance.

The model has been implemented by several workers to estimate regional or continental patterns of NPP and the CO2 sink over North America, Eurasia, South America, Africa, and Australia (Field et al., 1995; Thompson et al., 1996; Potter et al., 2003). Recently, it has also been implemented for the Indian subcontinent (Nayak et al., 2009). Implementation of this model for regional applications requires proper input datasets and values of several parameters to be tuned to account for the effect of regional heterogeneity. Among the several parameters, the value of maximum LUE (ε*) is critical in estimating regional NPP. Generally, in the global simulation studies, the value of maximum ε* is often taken as a constant between the range of 0.39 and 0.43 g C MJ−1 (Potter et al., 1993, 2003; Field et al.,1995). Use of such constant values for regional simulation of NPP could lead to large errors in national NPP estimates. Therefore, in the present study, the value of ε* is considered to be varied across different vegetation types similar to our previous study (Table II of Nayak et al., 2009) and simulation of monthly NPP at a 2-min spatial resolution for the period 1981–2006 are carried out.

2.2. Model datasets

The time-varying input data for the model run includes: NDVI, TEMP, SOLR, and PPT. The monthly NDVI data used in the study are based on Global Inventory Modeling and Mapping Studies (GIMMS) databases at an 8-km spatial resolution (Tucker et al., 2005). The data of climatic parameters used in the study are based on Climate Research Unit (CRU) at the University of East Anglia (UEA) (www.cru.uea.ac.uk/cru/data/hrg). The CRU monthly climatic databases of temperature, precipitation, and cloudiness during past 100 years are available at 0.50 grid and derived by interpolating measurements acquired from weather stations across the globe. The CRU cloud fraction data are used to obtain surface solar radiation over the country at monthly time scale by using the methodology described in Mani (1980). Extra-terrestrial radiation used in the method is obtained from the Surface Radiation Budget (SRB) databases (http://gewex-srb.larc.nasa.gov/). Monthly climatic drivers and NDVI so obtained are resampled at 2-min spatial resolutions in order to match the model grid size.

The land cover map of the country used in the present study is based on the land cover map of Southeast Asia derived at a 1-km spatial resolution by using multi-date SPOT-VEGETATION data for the global land cover-mapping project (Agrawal et al., 2003). In the present study, the land cover attribute data are then resampled at a 2-min resolution using ENVI pixel aggregate function. The soil attribute map of the country used in the present study is based on the Food and Agriculture Organization (FAO) of UNESCO's world soil map (Reynolds et al., 1999). The map provides the data of various compositions of the soil of the upper 30-cm depth: sand, silt, and clay fraction at a 5-min resolution. The data are interpolated over the subcontinent at the model resolution and then a soil attribute map, as per the seven classes defined in the CASA model, is prepared.

2.3. Method of estimation of linear growth rate of NPP and climatic contributions

In the present study, estimation of annual NPP from the monthly NPP was carried out by integrating the monthly value between the period from June of the given year to May of the next year instead of a calendar year, following the Pandya et al. (2004) method: to highlight the effect of southwest (SW) monsoon on NPP and to make the data compatible with the agricultural year. Hereafter, the annual NPP for the agricultural years is referred to as annual NPP. The linear growth rates of NPP at each pixel over the study region are estimated by fitting a straight line between NPP and agricultural years using the least square procedure. Hereafter, the linear growth rate of NPP could be referred to as NPP trend.

In order to assess the role of climate on the NPP trend during the study period, we have adopted the method of calculating the total differential of a dependent quantity. As NPP is the function of climate (solar radiation, precipitation, and air temperature) and other variables, the total differential of NPP can be expressed as

equation image

In the above expression, the term on the left represents NPP trend. The terms on the right hand side represent contribution of various climatic factors Pi (such as precipitation, air temperature, and solar radiation). The last term R′ on the right hand side represents the residual term that includes the contributions from agricultural development and other human activities, and so on. Each of the climatic contribution has two factors: one is the sensitivity term equation image and other is the trend equation image. Sensitivity terms are computed as the slope of the linear regression line between NPP and Pi. The trend terms are computed by fitting the straight line between Pi and years using the least square procedure.

3. Results

3.1. Spatial pattern of annual NPP climatology and its inter-annual variability

The spatial pattern of annual NPP climatology and associated coefficient of variation (CV) based on model simulations of the past 26 years (1981–2006) over India are presented in Figure 1. As shown in the figure, the estimated NPP exhibits large spatial variation across the country. Very large NPP (>900 g C m−2 yr−1) together with a small variance (CV < 15%) are estimated over the evergreen and deciduous forest regions of the northeast states, eastern, and southwestern peninsular region of the country, and partly over the Indo-Gangetic plains. NPP over the Indo-Gangetic plains and over the coastal states in eastern and southwestern India are estimated in the range of 600–900 g C m−2 yr−1 together with a small variance (CV < 5%). Moderate values of NPP in the range of 200–600 g C m−2 yr−1 associated with relatively large inter-annual variability (CV in the range of 15–30%) are estimated over the mixed shrub and grassland (MGL) of central India. Low mean NPP in the range of 100–200 g C m−2 yr−1 is estimated in the desert tracts of northwestern India and the northern portion of the Himalaya Mountain ranges. The high variability (CV > 30%) is also found in desert tracts of western India where rainfall amounts are low and erratic. In this area, the vegetation signal shows pulse during dry and wet years. Mean total NPP of the country estimated to be 1.42 Pg C at the rate of 520 g C m−2.

Figure 1.

Mean and percentage of variation of annual NPP over India during 1981–2006. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Modelled NPP exhibits large spatial variation across different vegetation covers of the country. Statistics of annual NPP over major land cover types of the country based on simulated result and that of observed data compiled from global NPP point databases (Olson et al., 2001; Scurlock and Olson, 2002) (Table I), suggest that there exists good agreement between the two estimates. According to simulated results, the mixed dry irrigated cropland (ICL) contributed 35% of national annual NPP budget (1.42 Pg C) at the rate of 450 g C m−2 which is approximately 20% higher than the ground-based measurement. Mixed irrigated cropland (MICL) contributed 21% of total national NPP at the rate of 600 g C m−2 which is 15% less than the observation. Deciduous broadleaf forest (DBF) has contributed 19% of national NPP budget at the rate of 660 g C m−2 which is 12% less than the observed mean. The evergreen broadleaf forest (EBF) had the highest annual mean NPP (1057 g C m−2), but its contribution to total NPP is hardly 9.5% due to less coverage of area. The modelled mean annual NPP of this land cover is found to be close to the observation. The MGL contributed 9% and grassland (GL) contributed only 1% to national NPP budget. The modelled mean NPP of the GL (191 g C m−2) is comparable to the observation, while modelled mean NPP for MGL is almost 30% less than the observations. Other land covers having limited geographical area coverage have contributed a very small fraction to the NPP budget of the country and, therefore, are ignored here.

Table 1. Statistics of annual NPP climatology over major land cover types in India
Land use/land cover (% land coverage of the country)Mean (and std) (gC/m2/year)Total NPP in Pg C/year (% CV)% contributionObserved mean NPP1 (std)
  • 1

    NPP measurements over India compiled from Global point databases (Olson et al., 2001; Scurlock and Olson 2002)

Mixed dry land and irrigated crop land (37)450 (246)0.50 (13.8)35370 (86)
Irrigated cropland and pasture (16.5)595 (245)0.30 (10.5)21741 (396)
Broadleaf deciduous forest (14.5)658 (286)0.27 (10.4)19754 (391)
Mixed shrub and grassland (11)400 (251)0.13 (17.8)9654 (325)
Broadleaf evergreen tree (5)1057 (378)0.14 (8.5)9.5947 (319)
Grassland (4)191 (197)0.01(13)0.7190 (60)
All LU/LC (100)5291.42
Table 2. Correlation coefficient between simulated NPP and climatic parameters over major land cover types of the country: Rppt is for precipitation, RTair is for air temperature, and RSolar is for surface solar radiation
Land cover typeNo. of observationsRpptRTairRSolar
Irrigated cropland and pasture71000.49− 0.070.09
Mixed dry land and irrigated cropland215640.65− 0.12− 0.08
Mixed shrub and grassland32900.5100.1
Grassland20340.69− 0.18− 0.43
Broadleaf deciduous forest71860.220.020.02
Broadleaf evergreen forest24670.290.130.09

3.2. Inter-annual variability of national total NPP

Inter-annual variability and trend of NPP holds prime importance in analysis of sink capacity of terrestrial ecosystems. In order to analyse the inter-annual variability of NPP in greater detail, integration of annual NPP over the country and over the regions dominated by crop and forest land are computed. The time series of annual NPP during past 25 agricultural years (1981–2006) and its linear growth rate over the country and over the major land cover types are presented in Figure 2. Total NPP over the country is estimated in the range of 1.27–1.50 Pg C yr−1 during the 25-years period with a mean value of 1.42 Pg C yr−1 and standard deviation of 0.06 Pg C yr−1. The lowest NPP is noticed in 1983 while the highest NPP is in 1997. During the early 6 years of the study period (1981–1986), NPP exhibits biennial oscillations and will be described further in the subsequent section.

Figure 2.

(a) Annual total NPP (solid line) and mean precipitation (ppt: dotted line) over the country. The trend line is drawn corresponding to annual total NPP. (b) The same as in (a) but for the croplands; (c) the same as in (a) but for the forest region over the country

Estimated linear growth rate suggest that total NPP over the country increases linearly from 1.36 Pg C yr−1 in 1981 to 1.475 Pg C yr−1 in 2005 at the rate of 0.005 Pg C yr−2. This increasing rate is equivalent to 8.5% (0.115 Pg C) over past 25 years. This positive NPP growth rate over the country is primarily due to enhancement of productivity over the cropland (at the rate of 0.0036 Pg C yr−2) with lesser extent due to enhancement of productivity over the forest region (at the rate of 0.0007 Pg C yr−2). In terms of percentage, the cropland is responsible for 70% change in annual NPP trend while forest region is only contributed 7% to the national NPP trend. Inter-annual trends of kharif crop season (June–October) NPP and rabi crop season (November–May of the next year) NPP are also analysed. The results suggest that both the NPP trends are positive and significant similar to the trend of annual NPP. However, the kharif crop season NPP trend (0.0023 Pg C yr−2) is relatively smaller than the rabi crop season NPP trend (0.0027 Pg C yr−2). The results further suggest that cropland contributed almost 70% change in both the seasonal NPP trend, similar to the annual NPP trend.

3.3. Inter-annual variability of NPP in relation to climate

A number of studies suggested that the inter-annual variability of terrestrial NPP over different parts of the globe have been related strongly to the inter-annual variability of climatic parameters (Nemani et al., 2003; Tian et al., 2003). In order to examine the relationship of NPP with climatic drivers (precipitation, air temperature, solar radiation) over India, CVs of the climatic parameters are estimated and compared with the CV of NPP. The comparison suggests that the locations of large variability of precipitation are mostly coinciding with the location of large values of CV of NPP: desert tracts of northwestern India, the northern portion of the Himalayas, MGL of central India and the south-central plateau. On the other hand, air temperature and solar radiation data exhibit very weak inter-annual variability. The scatterplots as shown in Figure 3 depict the relationship between the CVs of annual NPP and that of precipitation for the five major land cover types (viz. ICL, MIC, MGL, GL, DBF, EBF) over India. We found strong and significant relations between the CVs of NPP and those of precipitation for all land cover types. Estimated correlation coefficients are 0.54 for ICL, 0.68 for MIC, 0.59 for MGL, 0.71 for GL, 0.27 for DBF, and 0.36 for EBF. The comparison between the annual total NPP and mean precipitation over the country and over the regions dominated by cropland and forest corresponding to the agricultural years (Figure 2) suggest that maxima and minima of NPP are mostly coinciding with the maxima and minima of precipitation, respectively. The agricultural years 1991, 1994, 1997, and 2002 witnessed reduced precipitation which causes large reduction in NPP over India. The biennial oscillations feature presence in the national total NPP and mean precipitation during early 6 years of the study period (1981–1986) suggest that climatic fluctuation has a stronger role on the fluctuation of NPP over India. This is the period when India did not have a proper management system to tackle the severe effects of climatic fluctuation.

Figure 3.

Scatterplots between percentage of coefficient of variation of annual NPP and that of precipitation over major land covers of the country: for (a) irrigated cropland and pasture (ICL), (b) mixed-irrigated cropland and pasture (MICL), (c) mixed shrub and grassland (MGL), (d) grassland (GL), (e) deciduous broadleaf forest (DBF), (f) evergreen broadleaf forest (EBF), and (g) for all major land cover types

Further investigation is carried out by examining the coherence between the estimated NPP and climatic parameters at each pixel level. In order to do that, the whole 25-year study period of 1981–2005 was divided into two 15-year periods: 1981–1995 and 1991–2005. Hereafter, these two periods are referred to as early and later periods of the study period. The reason for considering these two periods is based on the idea of highlighting the effect of management practice in the agricultural system over the country. The early 15-year period witnessed the progress and early phase of green revolution, while the country has experienced a matured management system for the agricultural practice in the later 15-year period.

As shown in Figure 4, the spatial pattern of correlation between NPP and precipitation is positive over most part of the country during both the periods excepted over some parts of the Indo-Gangetic plains (parts of Uttar Pradesh, Jharkhand, and Madhya Pradesh), Western Ghats (Maharashtra) and the northeast states. In these regions, negative correlations are estimated in the later period (1991–2005) as compared to the positive correlation during the early period (1981–1995). The regions come under the flood-affected area, and it has been speculated that improper management of floods during the early period, and an improvement in the flood management mechanism in the later period could lead to negative and positive correlations, respectively.

Figure 4.

Upper panels: the spatial pattern of correlation coefficient between annual NPP and precipitation during 1981–1995 (left panel) and during 1991–2005 (right panel). Middle panels: the same as in the upper panels but for air temperature instead of precipitation; lower panels: the same as in the upper panels but for solar radiation instead of precipitation. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Variability of NPP has different relationship with the variability of air temperature at two different periods. Positive and negative correlation between these parameters is estimated over major parts of the country during the early and later periods, respectively. Solar radiation has a different effect on the variability of NPP across the country in two periods. Estimated positive correlation between solar radiation and NPP over the Indo-Gangetic plain is much larger in recent times (later period) as compared to the early period. On the other hand, larger negative correlations between these parameters is estimated over the western peninsular India (Gujarat, western part of Madhya Pradesh, central Maharashtra) during the recent periods, as compared to the early period.

3.4. Spatial pattern of NPP trend and its relation with climate

NPP trend is calculated at every grid cell over the study region for the three periods: early period (1981–1995), later period (1991–2005) and the complete study period (1981–2005), and is presented in the left panels of Figure 5. The total climatic contributions on the annual NPP growth rate coming from precipitation, solar radiation, and air temperature for these periods are presented in the right panels of the Figure 5.

Figure 5.

Upper panels: spatial pattern of linear growth rate of NPP computed during 1981–1995 (left panel) and corresponding climatic contribution (right panel). Middle panels: the same as in the upper panels but for the period 1991–2005; lower panels: the same as in the upper panels but for the complete 25-year study period from 1981 to 2005. This figure is available in colour online at wileyonlinelibrary.com/journal/joc

Large differences are noticed among the trends estimated for three different periods. During the early 15-year period, NPP trends are positive over most part of India except over the Western Ghats and some parts in the southeastern peninsular region. The total integrated NPP trend for the country is estimated to be 8.345 Tg C yr−1. This increasing trend is equivalent to 10% increase of national NPP budget corresponding to the base year 1981. During this period, the climate has positive contributions over the Indo-Gangetic plains and central India. Climate has a small negative contribution over Orissa, Jharkhand. Close inspection of the individual climatic contributions suggests that both solar radiation and air temperature have a strong control on NPP trend, while precipitation has minor control on the trend. on the national scale, climate has 25% positive contribution to the national NPP trend for the early period.

During the later 15-year period, large negative NPP trends are estimated over the Ingo-Gangetic plains and northeastern states. Previously negative trends over the southeastern peninsular regions are converted to the region with large positive trends. The resultant annual growth rate of national NPP budget for the country is reduced to 2.68 Tg C yr−1, which is equivalent to 2.7% of the national annual NPP corresponding to 1991 value (1.45 Pg C). This reduction in NPP budget for the country is associated with large negative climatic contributions from the Indo-Gangetic plains and coastal regions of Andhra Pradesh and Tamil Nadu. In these regions, declining strength of climatic contributions is much stronger than the actual value of negative trends of NPP. This suggests that agricultural management practice could help to prevent further decline of NPP. Over the central and south peninsular India, climate has positive influence on the NPP growth rate. In the national scale, the climate has − 1.4 Tg C yr−1 contributions to the NPP trend which is equivalent to − 50% of the NPP growth rate during this period. Close inspection of the individual climatic factors suggests that precipitation and solar radiation both play strong role in the decline of the NPP over the Indo-Gangetic plains, and enhancement of NPP over the central peninsular plateau. Air temperature has relatively small control on the NPP trend across the country.

Regarding the complete study period (1981–2005), however, spatial pattern of NPP trend mostly similar to trends of 1991–2005, the magnitude being much weaker. Positive trends are observed over the deserted tracts of northwestern India, most parts of the central peninsular plateau, and some parts of the Indo-Gangetic plains. Relatively small negative NPP trends are observed over the forest regions located in the Western Ghats and some parts in the northeastern states. The resultant trend for the country during this period is 5 Tg C yr−1 which is equivalent to 8.5% of the annual NPP corresponding to the year 1981 (1.42 Pg C). During this period, the climate has a small positive contribution (15%) to the annual NPP growth rate. Major parts of the country have experienced positive climatic contributions on the growth rate of NPP except the Indo-Gangetic plain and the northeast states. Negative climatic contributions over the Indo-Gangetic plains and northeast states are primarily contributed by variability in the precipitation.

3.5. Inter-annual variation of NPP in relation to atmospheric CO2 and ENSO

Global terrestrial biospheric uptake plays a very significant role on the control of seasonal and inter-annual variability of atmospheric CO2. To examine the role of biospheric uptake over India on the control of atmospheric CO2, comparison between anomalies of annual NPP over India, and the anomalies of atmospheric CO2 growth rate (measured at Maunaloa) are presented in Figure 6. The figure suggests that there exists an inverse relation between them during most of the study period except during 1981–1982, 1996–1998, and 2004–2006. The inverse relationship between the two parameters suggests that when the NPP growth rate over the country declined, the atmospheric CO2 growth rate was enhanced, and vice versa.

Figure 6.

Anomaly of simulated annual NPP over India is compared with atmospheric CO2 growth rate anomaly in (a), and with multivariate ENSO index (MEI) in (b). The data of atmospheric CO2 are taken from http://cdiac.ornl.gov/ftp/trends/co2/maunaloa.co2 and data of MEI are taken from http://www.esrl.noaa.gov/psd/people/klaus.wolter/MEI/mei.html

ENSO events have strong control on temperature and precipitation over different parts of the globe and, hence, influence the variability of global and regional NPP in inter-annual scale (Michael et al., 2001). The comparison between total NPP over India and Multi-variate ENSO Index (MEI) during the past 26-year study period (Figure 6, lower panel) suggest that NPP declined over India during all the three major EL Nino events (1982–1983, 1987–1988, 1997–1998, 2002–2003).

4. Concluding remarks

In this paper, the CASA, a remote-sensing-driven terrestrial biosphere model which takes account of variable LUE for different vegetation types is used to simulate spatio-temporal patterns of NPP over India during 1981–2006. The simulated NPP exhibited strong inter-annual variability across the country. The variability is largest over the low productive regions, moderate over the agricultural lands and small over the forest regions. There exists a significantly large correlation between the NPP variability and rainfall variability across the country. The correlation between NPP variability and that of air temperature and solar radiation are weak over most parts of the country. These results agreed with the conclusion made by Tian et al. (2003) that the rainfall variability has much control on the NPP variability of monsoon Asia.

Estimated long-term linear growth rate of annual NPP suggested that there is a small but significant increase of NPP over the country by 0.005 Pg C yr−2 (equivalent to 8.5%) during the past 25 years. This rate is relatively larger than that of global NPP increasing rate (i.e. 6%) reported earlier (Nemani et al., 2003) and is similar in magnitude as that of North America (Hicke et al., 2002). This increase of NPP of terrestrial India is primarily due to enhancement of productivity over the agricultural lands in the country. The spatial pattern in NPP trend illustrated that large increase of NPP occurred over the Indo-Gangetic plains, desert tracts of northwestern India, most parts of the central peninsular plateau and eastern and southeastern part of India. The NPP growth rate is positive and large (10%) during the early 15-year period and reduced to 2.7% during the later 15-year period. The reduction in NPP growth rate during the later period is primarily due to decline of NPP over the Indo-Gangetic plains. Diagnostic analysis suggests that climate had a relatively small but significant control (15%) on the NPP trend over the country during the past 25-year study period. Climate played a much stronger role on the enhancement of NPP during the early 15-year period, and on reduction (decline) of NPP during later 15 years of the study period.

Comparison between the anomalies of NPP over India with atmospheric CO2 suggests that Indian terrestrial biosphere could play significant role in the control of atmospheric CO2 variability similar to the global terrestrial system reported earlier by Neemani et al. (2003). It is further revealed that NPP over India is highly sensitive to the global ENSO events. During all the four major El Nino events (1982–1983, 1987–1988, 1997–1998, 2002–2003), NPP over India declined in keeping with the global NPP reported earlier (Nemani et al., 2003).

Acknowledgements

This research work is carried out as part of the National Carbon Project, ISRO-Geosphere and Biosphere Programme. We thank Global Inventory Modeling and Mapping Studies (GIMMS), NOAA for providing bi-monthly NDVI data and Climate Research Unit of East Anglia, UK, for providing the various climate data used in this study. We thank the anonymous reviewer for his valuable comments and suggestions which has helped us to improve our manuscript.

Ancillary