ANALYSIS OF BIOME-BGC MODEL FOR DRY TROPICAL FORESTS OF VINDHYAN HIGHLANDS , INDIA

A process-based model BIOME-BGC was run for sensitivity analysis to see the effect of ecophysiological parameters on net primary production (NPP) of dry tropical forest of India. The sensitivity test reveals that the forest NPP was highly sensitive to the following ecophysiological parameters: Canopy light extinction coefficient (k), Canopy average specific leaf area (SLA), New stem C : New leaf C (SC:LC), Maximum stomatal conductance (gs,max), C:N of fine roots (C:Nfr), All-sided to projected leaf area ratio and Canopy water interception coefficient (Wint). Therefore, these parameters need more precision and attention during estimation and observation in the field studies.


INTRODUCTION
There are various types of models that may be used in ecosystem analysis.Ruimy et al. (1994), categorized the NPP model into three groups: (1) statistical models (Lieth, 1975), (2) parametric models (Potter et al., 1993) and (3) process models (Foley, 1994).Traditional types of models are regression models, which are based on empirically derived statistical relationships.Such models may be used for predicting stand development under stable conditions and in regions where the built-in relationships were derived.Naturally, such models are not so useful for incorporating changing growth conditions and for spatial extrapolation.Moreover, such models remain descriptive and do not offer much explanatory power for ecosystem analysis.For this, so-called process-based models must be deployed.These models simulate ecosystem development as a result of ecophysiological processes described mechanistically and usually incorporate the effect of environmental change on ecosystem functioning and are able to quantify effects of, e.g., change in climate, elevated CO 2 , nitrogen deposition and land use scenarios (Cienciala and Tatarinov, 2006).
Sensitivity analysis is the study of how the variation (uncertainty) in the output of a model can be apportioned, qualitatively or quantitatively, to different sources of variation in the input of the model (Xu et al., 2004).It can be used to examine which variables/parameters have the largest effect on the model output.
Sensitivity analysis determines what level of accuracy is necessary for a parameter to make the model sufficiently useful and valid.If the sensitivity tests reveal that the model is insensitive to a particular factor, then it may be possible to use an estimate rather than a value with greater.
For ecological analysis Biogeochemical (BGC) model is most preferred.BIOME-BGC (Running and Coughlan, 1988;Thornton, 1998) White et al. (2000) and Tatarinov and Cienciala (2006) for temperate biomes.Reassessment of model sensitivity for dry tropical forest of India is needed because effect of parameters for different combinations of site and eco-physiological parameters may differ.Therefore, the aim of the present study is (i) to collect ecophysiological and site parameters for dry tropical forest of India from available literature and observations and (ii) to reveal the effect of ecophysiological parameters on NPP and also identify critically sensitive input parameters.

BIOME-BGC model description
The BIOME-BGC version 4.1.2was provided by Peter Thornton at the National Center for Atmospheric Research (NCAR), and the Numerical Terradynamic Simulation Group (NTSG) at the University of Montana (available online at http://www.ntsg.umt.edu.).The model is used for this study along with MT-CLIM (Mountain Microclimate Simulator).BIOME-BGC (Thornton 1998(Thornton , 2000) ) is a biogeochemical model that simulates the storage and fluxes of water, C and N in terrestrial ecosystems using a daily time step and at different ecological scales (Churkina et al., 2003).The model requires daily weather data including radiation, maximum and minimum temperatures, precipitation and daytime VPD.It also requires information describing the soil properties (Table 2) and eco-physiological traits of vegetation (Table 3).Allometric relationships are used to initialize plant and soil C and N pools based on the leaf pools of these elements.BIOME-BGC estimates NPP on the basis of GPP and Ra (i.e.NPP).

Sensitivity analysis
Sensitivity analysis was performed as documented in Cienciala and Tatarinov (2006), "output variables (y) to input parameters (x) (or the effect of parameter x on the variable y), ¨y/¨x was calculated as a ratio of output variable change to parameter change (both in %)".A negative ratio would mean a decrease in variable with an increased parameter value and vice versa (Cienciala and Tatarinov 2006).As for the absolute quantity (|¨y/¨x|), the parameters were ranked in terms of their effect on the modeled variable as (i) parameters with a strong effect (|¨y/¨x| larger than 0.2), (ii) parameters with a medium effect (|¨y/¨x| between 0.1 and 0.2) and (iii) parameters with low effect (|¨y/¨x| less than 0.1).

Study area
7KH VWXG\ DUHD LV ORFDWHG RQ WKH 9LQGK\DQ 3ODWHDX LQ WKH 0DULKDQ UDQJH RI (DVW 0LU]DSXU )RUHVW 'LYLVLRQ RI 8WWDU 3UDGHVK ,QGLD DW -1 DQG -E.The total forest area in this block is 10360 ha.The climate is tropical and characterized by monsoon conditions.There are three seasons: rainy (mid June -Sept.),winter (November -February) and summer (March -mid June).The mean monthly temperatue varies from 17.5 C (January) to 37.5 C (May) and total annual rainfall averages 821 mm, of which 86 % occurs in the rainy season.The potential natural vegetation is northern tropical dry deciduous forest (Champion and Seth, 1968).The details about dominated species of study area were described by Singh and Singh (1991).

RESULT
Sensitivity analysis for eco-physiological parameters revealed that SLA, g s,max , W int and LAI all:pro exert a strong negative effect on NPP, while k, SC:LC and C:N fr showed positive effect (Table 3).C:N lit , N R , CRC:SC and C:N leaf showed positive medium effect on NPP (Table 3).Input parameters like SLA shd:su , T t , VPD i , m t , LWP f , FRC:LC, T lf , VPD f etc. had low sensitivity to model output value of NPP (Table 4).m l m fr , LWC:TWC, g cut has produced almost negligible effect on the sensitivity of the model.

DISCUSSION
For dry tropical forest observed k had a high positive impact on NPP.The value of k depends upon intercepted photosynthetically active radiation (PAR) and LAI (Lagergren et al. 2005).LAI has major influence on BIOME-BGC since it controls canopy radiation absorption, water interception, photosynthesis, and litter inputs to detrital pools (White et al. 2000).Increased specific leaf area (SLA) resulted in higher LAI (LAI = SLA × leaf carbon) without altering photosynthetic capacity, increasing water stress and thereby reducing NPP (White et al. 2000).SC:LC allocation ratio has a high positive effect on NPP of dry tropical forest.This International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXVIII-8/W20, 2011 ISPRS Bhopal 2011 Workshop, 8 November 2011, Bhopal, India is due to the redistribution of biomass into the woody compartment with a low turnover rate (Tatarinov and Cienciala 2006).Maximum stomatal conductance (g s,max ) showed a negatively high sensitivity to NPP; this was in contrast to the report of European temperate managed forestry (Tatarinov and Cienciala 2006).Increases in g s,max reduced NPP by increasing water stress.Increased C:N fr , reduces root nitrogen requirements and diverts nitrogen to increased photosynthetic capacity which promotes higher NPP for most biomes (White et al. 2000).
Fraction of leaf N in Rubisco of dry tropical forest showed positive sensitivity to NPP similar to major natural temperate biomes (White et al. 2000) and for beech and small or medium spruce species (Tatarinov and Cienciala 2006).This effect follows from the fact that maximum rate of carboxylation in the model is proportional to N R (Tatarinov and Cienciala 2006).The effect of the C:Nl leaf was similar to the report of Tatarinov and Cienciala (2006) for beach species which was in contrast to White et al. (2000), who found, that the increase of C:N leaf decreased NPP in all woody biomes.Such an ambiguous effect of C:N leaf might be due to the trade-off between the increase of photosynthesis and foliage respiration with an increasing foliage nitrogen content (Tatarinov and Cienciala 2006).VPD i and VPD f showed low negative effect on NPP.This negative relation with NPP was due to increased VPD which causes closure of stomata resulting in inhibition of photosynthesis (Jarvis 1976;Stewart 1988) (1/yr) 0.00 Table 4: Sensitivity Index (SI) for Ecophysiological parameters

CONCLUSION
The study identified key eco-physiological parameters of a process model BIOME-BGC based on a detailed sensitivity analysis.Among the eco-physiological parameters k, SLA, SC:LC, g s,max , C:N fr , LAIall:pro and W int showed the strongest effect on simulated NPP.Four ecophysiological parameter viz., C:N lit , N R , C:Nleaf and CRC:SC had medium influence on simulated NPP value.Therefore these parameters need more precision and attention during estimation and observation in the field study.

Table 1 :
is such model which employs a simplified biochemical model of photosynthesis, environmentally regulated stomatal conductance, and explicit calculations of respiration for various plant pools to calculate NPP.It is a process-based model requiring a considerable number of ecophysiological and site parameters.Therefore, application of this model requires parameterization, including sensitivity analysis of model output to the input data.List of abbreviations International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXVIII-8/W20, 2011 ISPRS Bhopal 2011 Workshop, 8 November 2011, Bhopal, India

Table 2 :
Model input parameters for site characteristics

Table 3 :
White et al. (2000))iala 2006)w negative effect on NPP, which primarily occurs in summer season in dry deciduous forest, i.e., it is affected by fire mortality for a shorter time(Tatarinov and Cienciala 2006).Ecophysiological input parameters for model (*DVM = Default value of model for deciduous broadleaf forest (DBF) biome asWhite et al. (2000))