Print PageBiomass estimation and mapping of crops in cropland and non-crop vegetation in abandoned croplands make us understand the contributions of crops and vegetation to primary productivity. The digital number (DN) of a QuickBird (QB) satellite image was converted to a reflectance factor using the dark object subtraction method and the spectral reflectance of asphalt. We used a field spectra-based Normalized Difference Vegetation Index (NDVI) and sampled the biomass of crop and non-crop vegetations (CNCVs) including cultivated and abandoned croplands to develop a simple biomass estimation model. Incorporating this model into QB image gives us the biomass distribution map.
Results/Conclusions
The measured biomass was significantly correlated with field-based NDVI (R2=0.79, p<0.001). QB imagery of 12 April and 23 May 2007 was used for agricultural land (AL) classification, and image obtained on 8 July 2007, was used to validate the field-based biomass estimation model and to map biomass. Four classes including paddy, corn, abandoned cropland and poly house were classified by the ISODATA clustering technique. Overall accuracy of the classification was 94% with Kappa statistics of 0.92. The final classified image was used to produce a biomass distribution map of CNCVs. The biomass estimation model was evaluated by incorporating it into the July QB image to obtain the observed biomass. The relationship between field-measured and QB-observed biomass appeared to be linear (R2=0.79). Finally, a field-based biomass estimation model was applied to the July QB NDVI image in the area of paddy, corn and abandoned cropland. The biomass map gives us information regarding the status of biomass in CNCVs ranged between <140 to >900 g◦m-2 (DW). In the map, biomass appeared to be higher in the abandoned croplands than that in the croplands. Thus, models using field-spectra based NDVI and sampled biomass could be applied to QB images for remote estimation and mapping of biomass in mountain agricultural land.
