Journal of Geographical Sciences ›› 2012, Vol. 22 ›› Issue (4): 669-680.doi: 10.1007/s11442-012-0955-9

• Ecology and Environment • Previous Articles     Next Articles

Construction of aboveground biomass models with remote sensing technology in the intertropical zone in Mexico

AGUIRRE-SALADO Carlos Arturo1,2, TREVI?O-GARZA Eduardo Javier1, AGUIRRE-CALDERÓN Oscar Alberto1, JIMÉNEZ-PÉREZ Javier1, GONZÁLEZ-TAGLE Marco Aurelio1, VALDEZ-LAZALDE José|René3, MIRANDA-ARAGÓN Liliana1, AGUIRRE-SALADO Alejandro Iván3   

  1. 1. Autonomous University of Nuevo Leon, Linares NL 67700, Mexico;
    2. Autonomous University of San Luis Potosi, SLP 78290, Mexico;
    3. The College of Postgraduates, Texcoco MEX 56230, Mexico
  • Received:2011-05-12 Revised:2012-03-10 Online:2012-08-15 Published:2012-07-10

Abstract:

Spatially-explicit estimation of aboveground biomass (AGB) plays an important role to generate action policies focused in climate change mitigation, since carbon (C) retained in the biomass is vital for regulating Earth’s temperature. This work estimates AGB using both chlorophyll (red, near infrared) and moisture (middle infrared) based normalized vegetation indices constructed with MCD43A4 MODerate-resolution Imaging Spectroradiometer (MODIS) and MOD44B vegetation continuous fields (VCF) data. The study area is located in San Luis Potosí, Mexico, a region that comprises a part of the upper limit of the intertropical zone. AGB estimations were made using both individual tree data from the National Forest Inventory of Mexico and allometric equations reported in scientific literature. Linear and nonlinear (exponential) models were fitted to find their predictive potential when using satellite spectral data as explanatory variables. Highly-significant correlations (p=0.01) were found between all the explaining variables tested. NDVI62, linked to chlorophyll content and moisture stress, showed the highest correlation. The best model (nonlinear) showed an index of fit (Pseudo-r2) equal to 0.77 and a root mean square error equal to 26.00 Mg/ha using NDVI62 and VCF as explanatory variables. Validation correlation coefficients were similar for both models: linear (r=0.87**) and nonlinear (r=0.86**).

Key words: MODIS, MCD43A4, MOD44B, forest inventory, regression