No abstract

Abstract:Fire is one of the main factors directly impacting Amazonian forest biomass and dynamics. Because of Amazonia's large geographical extent, remote sensing techniques are required for comprehensively assessing forest fire impacts at the landscape level. In this context, Light Detection and Ranging (LiDAR) stands out as a technology capable of retrieving direct measurements of vegetation vertical arrangement, which can be directly associated with aboveground biomass. This work aims, for the first time, to quantify post-fire changes in forest canopy height and biomass using airborne LiDAR in western Amazonia. For this, the present study evaluated four areas located in the state of Acre, called Rio Branco, Humaitá, Bonal and Talismã. Rio Branco and Humaitá burned in 2005 and Bonal and Talismã burned in 2010. In these areas, we inventoried a total of 25 plots (0.25 ha each) in 2014. Humaitá and Talismã are located in an open forest with bamboo and Bonal and Rio Branco are located in a dense forest. Our results showed that even ten years after the fire event, there was no complete recovery of the height and biomass of the burned areas (p < 0.05). The percentage difference in height between control and burned sites was 2.23% for Rio Branco, 9.26% for Humaitá, 10.03% for Talismã and 20.25% for Bonal. All burned sites had significantly lower biomass values than control sites. In Rio Branco (ten years after fire), Humaitá (nine years after fire), Bonal (four years after fire) and Talismã (five years after fire) biomass was 6.71%, 13.66%, 17.89% and 22.69% lower than control sites, respectively. The total amount of biomass lost for the studied sites was 16,706.3 Mg, with an average loss of 4176.6 Mg for sites burned in 2005 and 2890 Mg for sites burned in 2010, with an average loss of 3615 Mg. Fire impact associated with tree mortality was clearly detected using LiDAR data up to ten years after the fire event. This study indicates that fire disturbance in the Amazon region can cause persistent above-ground biomass loss and subsequent reduction of forest carbon stocks. Continuous monitoring of burned forests is required for depicting the long-term recovery trajectory of fire-affected Amazonian forests.

Very few studies have been devoted to understanding the digital terrain model (DTM) creation for Amazon forests. DTM has a special and important role when airborne laser scanning is used to estimate vegetation biomass. We examined the influence of pulse density, spatial resolution, filter algorithms, vegetation density and slope on the DTM quality. Three Amazonian forested areas were surveyed with airborne laser scanning, and each original point cloud was reduced targeting to 20, 15, 10, 8, 6, 4, 2, 1, 0.75, 0.5 and 0.25 pulses per square meter based on a random resampling process. The DTM from resampled clouds was compared with the reference DTM produced from the original LiDAR data by calculating the deviation pixel by pixel and summarizing it through the root mean square error (RMSE). The DTM from resampled clouds were also evaluated considering the level of agreement with the reference DTM. Our study showed a clear trade-off between the return density and the horizontal resolution. Higher forest canopy density demanded higher return density or lower DTM resolution.

No abstract

Spectral-Temporal Analysis by Response Surface (STARS), which exploits both multispectral and multitemporal information using fitted response surfaces, was used to describe deforestation patterns in the Brazilian Amazon. The STARS was conducted upon a MODIS dataset formed by 21 selected cloud free images (eight days composition) acquired from August 2003 to August 2004. The Multi-Coefficient Image (MCI) resulted from the STARS was used as input attributes for the three tested classifiers: Instance Based Knearest neighbor (IBK), Decision Tree (DT) and Neural Network (NN). The IBK classifier presented the highest accuracy (κ=0.93) in detecting deforestation also indicating the deforestation period (early or late in the year). The results showed that the STARS is promising to describe spectral change patterns over time, allowing detection of the deforestation process which occurs in the Brazilian Amazon.

No abstract