Hiran Zani

Fossil megafans evidenced by remote sensing in the Amazonian wetlands

In the present work, remote sensing data are analyzed for investigating if a set of Late Quaternary triangular, fan-shaped morphologies from a large wetland in northern Brazilian Amazonia corresponds to residual megafan deposits. The results revealed main characteristics suggestive of megafans, which include: fan shape; distribution on a low-lying wetland; wide individual areas of several tens or hundreds of km2 in surface; concave-up and convex-up longitudinal and transverse profiles, respectively; very low gradient; axes pointing toward basement rocks and smooth slopes converging into the basin center; and surfaces with distributary drainage networks. The recognition of megafan systems with distributary drainage in this region is an interesting finding, as these megafans record a past landscape that differs from the modern fluvial tributary drainage of the Amazon basin. Characterizing these palaeomorpho logies is of high relevance for reconstructing changes in the Amazonian landscape during the Late Quaternary.

Multitemporal Landsat data applied for deciphering a megafan in northern Amazonia

In this article, we present a protocol for the recognition of channels and waterbodies and for estimating their hydroperiod to both characterize a megafan system in northern Amazonia and discuss its dynamics in the late Quaternary. Our approach is based on analysis of a multitemporal Landsat Thematic Mapper/Enhanced Thematic Mapper Plus (TM/ETM+) data set consisting of 13 scenes acquired from 2000 to 2009. The images were primarily preprocessed using standard techniques for geometric and radiometric corrections. Following numerous tests with the most common water classifiers, we determined that a simple threshold on an infrared band, which provided a binary image containing both water and non-water classes, would adequately meet the goals of this article. The sum of all 13 classifications produced a new image, the hydroperiod product, with pixel values corresponding to the relative frequency of standing water. The results indicate that there is no permanent drainage network in the Viruá megafan, which displays only temporary channels with medium flood frequency and palaeochannels. These are arranged as complex distributary networks, as is typical of megafan depositional systems. The southern part of the megafan contains distributaries that are seasonally flooded, while the central and northern parts display an increased volume of palaeochannels. These data lead us to conclude that the locus of sediment deposition along this megafan system changed from north to south through time.

Classification of Vegetation over a Residual Megafan Landform in the Amazonian Lowland Based on Optical and SAR Imagery

The origin of large areas dominated by pristine open vegetation that is in sharp contrast with surrounding dense forest within the Amazonian lowland has generally been related to past arid climates, but this is still an issue open for debate. In this paper, we characterize a large open vegetation patch over a residual megafan located in the northern Amazonia. The main goal was to investigate the relationship between this paleolandform and vegetation classes mapped based on the integration of optical and SAR data using the decision tree. Our remote sensing dataset includes PALSAR and TM/Landsat images. Five classes were identified: rainforest; flooded forest; wooded open vegetation; grassy-shrubby open vegetation; and water body. The output map resulting from the integration of PALSAR and TM/Landsat images showed an overall accuracy of 94%. Narrow, elongated and sinuous belts of forest within the open vegetation areas progressively bifurcate into others revealing paleochannels arranged into distributary pattern. Such characteristics, integrated with pre-existing geological information, led us to propose that the distribution of vegetation classes highlight a morphology attributed to a Quaternary megafan developed previous to the modern fluvial tributary system. The characterization of such megafan is important for reconstructing landscape changes associated with the evolution of the Amazon drainage basin.

Impact of sedimentary processes on white-sand vegetation in an Amazonian megafan

Amazonian white-sand vegetation has unique tree communities tolerant to nutrient-poor soils of interest for interpreting processes of adaptation in neotropical forests. Part of this phytophysionomy is confined to Late Quaternary megafan palaeo-landforms, thus we posit that sedimentary disturbance is the main ecological factor controlling tree distribution and structuring in this environment. In this study, we characterize the topographic trend of one megafan palaeo-landform using a digital elevation model and verify its relationship to the forest by modelling the canopy height with remote sensing data. We also compare the composition and structure (i.e. canopy height and diameter at breast height) of tree groups from the outer and inner megafan environments based on the integration of remote sensing and floristic data. The latter consist of field inventories of trees 10 cm dbh using six (500 × 20 m) plots in várzea, terra firme and igapó from the outer megafan and 20 (50 × 20 m) plots in woodlands and forests from the inner megafan. The unweighted pair group method with arithmetic mean (UPGMA) and the non-metric multidimensional scaling (NMDS) were applied for clustering and dissimilarity analyses, respectively. The megafan is a sand-dominated triangular wetland with a topographic gradient of < 15 cm km−1, being more elevated along its axis. The outer megafan has a higher number of tree species (367), taller canopy height (mean of 14.1 m) and higher diameter at breast height (mean of 18.2 cm) than the white-sand forest. The latter records 89 tree species, mean canopy height of 8.4 cm and mean diameter at breast height of 15.3 cm. Trees increase in frequency closer to channels and toward the megafan’s axis. The flooded and nutrient-poor sandy megafan substrate favoured the establishment of white-sand vegetation according to the overall megafan topography and morphological heterogeneities inherent to megafan sub-environments.

Automatic detection of burned areas in wetlands by remote sensing multitemporal images

In this paper, a methodology for automatic detection of burned areas is suggested. The classification criterion is performed using Bayesian statistical parameter (mean and covariance matrix) extracted automatically using the Expectation Maximization algorithm and taking into account the spectral similarity between burned and flooded areas. In this work the final process involves the application of morphological operators of erosion and dilation of images in order to insert information from the spatial context, refining the final map. Experiments were conducted to a TM-Landsat scene with areas affected by fires and seasonal flooding. The results show that the accuracy is increased with the consideration of flooding mask and the subsequent application of spatial context, reaching values up to 97% of accuracy when compared with a reference map.