Articles | Volume 8, issue 3
Biogeosciences, 8, 667–686, 2011
Biogeosciences, 8, 667–686, 2011

Research article 17 Mar 2011

Research article | 17 Mar 2011

Integrating field sampling, geostatistics and remote sensing to map wetland vegetation in the Pantanal, Brazil

J. Arieira1,4, D. Karssenberg2, S. M. de Jong2, E. A. Addink2, E. G. Couto3, C. Nunes da Cunha1,4, and J. O. Skøien2 J. Arieira et al.
  • 1Instituto Nacional de Áreas Úmidas (INAU), Federal University of Mato Grosso, Cuiabá-MT, 78060-900, Brazil
  • 2Department of Physical Geography, Faculty of Geosciences, Utrecht University, P.O. Box 80115, 3508 TC, Utrecht, The Netherlands
  • 3Department of Soils, Faculty of Agronomy, Federal University of Mato Grosso, Cuiabá-MT, 78060-900, Brazil
  • 4Núcleo de estudos ecológicos do Pantanal (NEPA), Instituto de Biociências, Federal University of Mato Grosso, Cuiabá-MT, 78060-900, Brazil

Abstract. Development of efficient methodologies for mapping wetland vegetation is of key importance to wetland conservation. Here we propose the integration of a number of statistical techniques, in particular cluster analysis, universal kriging and error propagation modelling, to integrate observations from remote sensing and field sampling for mapping vegetation communities and estimating uncertainty. The approach results in seven vegetation communities with a known floral composition that can be mapped over large areas using remotely sensed data. The relationship between remotely sensed data and vegetation patterns, captured in four factorial axes, were described using multiple linear regression models. There were then used in a universal kriging procedure to reduce the mapping uncertainty. Cross-validation procedures and Monte Carlo simulations were used to quantify the uncertainty in the resulting map. Cross-validation showed that accuracy in classification varies according with the community type, as a result of sampling density and configuration. A map of uncertainty derived from Monte Carlo simulations revealed significant spatial variation in classification, but this had little impact on the proportion and arrangement of the communities observed. These results suggested that mapping improvement could be achieved by increasing the number of field observations of those communities with a scattered and small patch size distribution; or by including a larger number of digital images as explanatory variables in the model. Comparison of the resulting plant community map with a flood duration map, revealed that flooding duration is an important driver of vegetation zonation. This mapping approach is able to integrate field point data and high-resolution remote-sensing images, providing a new basis to map wetland vegetation and allow its future application in habitat management, conservation assessment and long-term ecological monitoring in wetland landscapes.

Final-revised paper