Estimating Carbon-fixation in
India based on Remote Sensing Data R.S. Hooda1 and
D.G.Dye2 Abstract 1Haryana State Remote Sensing Application Centre, HAU Campus, Hisar 125 004, India. Ph./Fax : 91-1662-32632 2Department of Geography, Boston University, 675 Commonwealth Avenue, Boston, MA, 02215 - 1401, USA. Ph. + 1-67-353-2525, Fax. +1-617-353-5956 The energy emitted by sun and captured by the terrestrial vegetation through the process of carbon-fixation governs the energy fluxes on the earth. Carbon-fixation, thus, is an important parameter in the study of different biological processes. Reflectance based data from the earth observing remote sensing satellites provide vital information regarding terrestrial vegetation. Efforts were made in the present study to estimate carbon-fixation over Indian territory using Production Efficiency Model (PEM). PEM used to evaluate Net Primary Productivity (NPP) requires decomposition of productivity into independent parameters involved in the production built up process. In the first attempt of its kind, most of the important parameters used in the PEM WERE derived from the satellite observations. NASA/NOAA Pathfinder AVHRR Land (PAL) 10 day composted NDVI data with a spatial resolution of 8 km was used from the study. The study area was divided into agricultural and natural vegetation areas based on the NDVI-climatological technique developed by Hoda and Dye (1995). The NDVI for the years 1987, 1988,1989 were used to estimate fraction of PAR absorbed (FAPAR) based on the relationship Fapar = -0.31 +1.9*NDVI provided by the SAIL model. Incident PAR (IPAR) data set for India was extracted from the monthly global IPAR data set already generated using UV reflectivity data from Nimbus Total Ozone Mapping Spectrometer (TOMS). The IPAR data when combined with the fAPAR data, provided absorbed PAR(APAR). APAR was subsequently cumbered to NPP using the mean PAR conversion efficiency values of 2.07 and 0.64 g/M joules for agricultural and non-agricultural vegetations, respectively, calculated based on literature. Annual and interracial variations in carbon-fixation in India have also been discussed. 1. Introduction Productivity is the rate of atmospheric carbon uptake by vegetation through the process of photosynthesis. Built up of productivity is a complex phenomenon which is a culmination of many temporal plant processes. Recent methods to evaluate NPP involves decomposition of productivity into independent parameters such as incoming solar radiation, radiation absorption efficiency and conversion efficiency of absorbed radiation into organic matter (Kumar and Monteith, 1981). The models developed in these studies are an advancement over the statistical models properly accounting for various steps in the productivity built up process. Goward et. Al. (1985) showed that vegetation indices, such as Normalized Difference Vegeta-tion Index (NDVI) are related to net primary production (NPP, g m-2 year1). Monteith (1987) suggested that NPP under non-stressed conditions is linearly related to the amount of Photosynthetically active radiation (PAR, MJ m-2) that is abosrbed by green foliage (APAR, MJ m-2). Further, Kumar and Monteith (1981) showed how the fraction of PAR absorbed (fAPAR) relates to the ratio of red reflectance ® to near infrared (NIR). Asrar et.al. (1984) subsequently related the NDVI to the Fapar; hence NDVI may be used to estimate NPP at global scale. Eck and Dye (1991) describged a simple, physically based, satellite remote sensing method for estimating IPAR that uses ultraviolet (UV) reflectivity data from the Nimubus Total Ozone Mapping Sepectrometer (TOMS). Subsequently, Dye and Gward (1993) also created a global APAR image using spectral reflectance measurements from the NOAA-7 AVHRR and TOMS data. Hunt (1994) suggested that global estimates of NPP based on vegetation indices should include a classification among established forest, young forest and non-forest ecosystems to account for difference in _. To address this problem, Hooda and Dye (1995) developed an automated technique for the identification of agricultural areas using NDVI-climatological modeling. One of the major problem in the NPP estimation is the finding of representative values of PAR Conversion Efficiency (e) for various vegetation types as it changes with the type of vegetation, temperature, water availability and metabolic type of the plant (C3 or C4 type). Prince 1991 and Rumy et. Al. (1994) searched though the literature and listed e values for various vegetation and ecosystem types. Prince and Goward (1995) used the PEM for global productivity studies. In the present study the carbon-fixation in India has been studied using all the PEM parameters derived from remote sensing data. 2. Data Used 2.1 NDVI data NASA/NOAA Pathfinder AVHRR Land (PAL) 10 days composted NDVI data set for year 1987, 1988, and 1989 was procured from the Goddard Distributed Active Archive Center (DAAC), USA. To generate composites data set, 10 consecutive days of data are combined, taking the observation for each 8 km bin from the data with the fewest clouds and atmospheric contaminants as identified by the highest NDVI value. There are three composites per month for each year of data. The composting technique fairly removes the cloud contamination from the data to use in climatic modeling studies (Agbu and James, 1994). The data is available on Goode's Equal Area Projection. 2.2 IPAR data Global IPAR data set generated by Dye (1995) using UV reflectivity data from Nimbus TOMS sensor through the method of Eck and Dye (1991) was used for the present study. This TOMS IPAR data set consists of monthly average estimates, at a spatial resolution of 1o*1o degree from 66oN to 66oS latitude. The data for the Indian region was extracted from the global data set and interpolated to match the 8 km resolution of NDVI data. 3. Methodology 3.1 Separation of agricultural and natural vegetation The NDVI-climatological modeling technique developed by Hooda and Dye (1995) was used for the separation of agriculture of agricultural and natural vegetation areas. 3.2 Productivity Estimation Productivity estimation was done as described in the following steps: 3.2.1 Estimation of Fraction of IPAR absorbed by vegetation (Fapar) The spectral vegetation index measurements produced by calculating the NDVI have been shown, empirically and theoretically, to be related to fAPAR in vegetation canopies (Ruimy. et.al., 1994). Although there are several possible limitations to such an inference, it does appear that an approximation of this fAPAR can be derived from the NDVI (Myneni and Williams, 1994). Ruimy et.al. (1994), after an extensive search through the literature, tabulated various relationships between Fapar and NDVI developed by different workers. For the present study relationship based on SAIL model simulation was used which is represented as: The 10 day composited NDVI data was first averaged to give average monthly NDVI for al the three years. Calibrations for negative values on land in the NDVI data were made in way to set the bare soil Fapar TO zero. This calibration required a uniform enhancement of 0.1 NDVI units in the data. From average monthly NDVI, fAPAR for each month was calculated using the above equation. 3.2.2 Estimation of Absorbed Photosynthetically Active Radiations (APAR) The APAR calculations required IPAR and fAPAR Data sets for India. Monthly fAPAR data set of India for the three years was generated as described in the previous step. The monthly Indian IPAR data extracted from TOMS global data set of Dye (1995) was combined with the respective fAPAR data to give monthly APAR in MJ m-2. 3.2.3 Estimation of Absorbed Photosynthetically Active Radiations (APAR) The APAR calculations required IPAR and fAPAR data sets for India. Monthly fAPAR data set of India for the three years was generated as described in the previous step. The monthly Indian IPAR data extracted from TOMS global data set of Dye (1995) was combined with the respective fAPAR data to give monthly APAR in MJ m-2. 3.2.4 Estimation of NPP, biomass and Carbon-fixation The conversion of APAR into productivity requires conversion efficiencies of APAR into dry matter (e) of various crops. Since we have a map differentiating the area into agricultural and natural vegetation, mean conversion efficiency values for the two types of vegetation are required for use in the mode. Ruimy et. Al. (194) conducted an extensive literature survey and tabulated the e values for different types of vegetation reported by different workers. But most of the workers reported e values in terms of above ground dry matter only. To overcome this problem they also searched through the literature to estimate a mean ratio of below ground NPP to above ground NPP and based on this factor they calculated the e values for different agricultural and forest vegetation. From this data we calculated the weighted average e values for cultivated and forested areas in India which were found to be 2.07 and 0.644 g dry matter (above ground and below ground ) MJ-1, respectively. These values were used for converting the APAR into NPP and subsequently to biomass in the present study. The average value of carbon in the vegetative dry matter has been reported to be about 0.45 per cent which was used to estimate carbon-fixation from the estimated biomass. 4.Results and Discussions 4.1 Interannual variations in Carbon-fixation The monthly biomass and carbon fixed in the Indian territory during the years 1987, 1988 and 1989, as calculated using the above steps, are shown in table 1. For all the three years, there seems to be a general trend of change in monthly biomass generation and C-fixation. The biomass and cosequently C-fixation, starts building up in the month of January and reaches its peak in the month of February/March and then drops suddenly in April. The biomass generation and C-fixation remains low in the summer months of May and June. This again starts building up in July/August and reaches its peak in September/October and the falls suddenly in November. Thus, we observed two peaks of carbon-fixation; first in the months of Feb/March and second in the month on October/November. Whereas, two low C-fixation peaks are also observed in the months of April and November. These high and low peaks of carbon-fixation clearly corresponds with the two crop growing seasons in India. The winter crops are sown in the month of November/December and they reach their peak vegetative state in the month of Feb./March and are harvested in the month of April. The agricultural lands generally remain fellow during the summer months of May and June, corresponding to low rates of C-fixation. The summer crops are sown in the months of June/July after the onset of monsoon and they rearch their peak vegetative stage in the month of Sept./Oct. before harvesting in November. Sowing of winter seson crops starts in the end of November or December and therefore, the biomass remains low during these months. Thus, this interannual variation in the rates of C-fixation indicated that agricultural C-fixation plays a major part in the total terrestrial biomass production in India as more than 45 per cent of the total geographical area in India is under cultivation.
4.2 Annual variations in Carbon-fixation Total biomass generation of 878.77, 945.96, and 12778.35 million tons was estimated in India for the years 1987, 1988 and 1989, respectively with a corresponding annual C-fixation of 395.45, 425.68 and 181.16 million tons for same years (Table 1). The performance of monsoon is the single most important factor effecting the growth of vegetation and consequenly monsoon is the single most important factor effecting the growth of vegetation and consequenly agricultural productivity in India. A study in the behavior of Indian monsoon showed that it was normal for the year 1989. But it showed a negative anomaly during 1987 and a positive anomaly during 1988 causing drought and floods in the two years, respectively. Therefore, the growth of vegetation and agricultural productivity was reduced in both the years. This interannual anomaly in carbon-fixation could also be described through the present methodology using PEM. The highest annual carbon-fixation of 575.26 million tons was estimated for the year 1989 as compared to 425.68 and 395.45 million tons for the years 1988 and 1987, respectively. 4. Conclusions Based upon the present study it could be concluded that the PEM cvan be used to make fairly correct estimates of biomass and carbon-fixation on a regional and global basis. Most of the parameters used in the PEM can be derived from the remotely sensed data. The model was also able to describe the annual and interannual variations in biomass production and carbon-fixation in the region. Therefore, the technique developed in the present study based upon the use of remote sensing data seems to have a great potential for making quick and accurate estimates of biomass and carbon-fixation over a large region. 8. Reference
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||