Eduarda Martiniano de Oliveira Silveira

CHANGE DETECTION IN BRAZILIAN SAVANNAS USING SEMIVARIOGRAMS DERIVED FROM NDVI IMAGES

The Normalized Difference Vegetation Index (NDVI) is often used to extract information from vegetated areas since it is directly related to vegetation parameters such as percent of ground cover, photosynthetic activity of the plant and leaf area index. The aim of this paper was to analyze the potencial of semivariograms generated from NDVI values to detect changes in vegetated areas, analyzing their behavior (shape) and derived metrics (range, sill and nugget). Semivariograms were generated from NDVI values derived from Landsat TM images of May 2010, June 2010 and July 2011. The study area is located in the northern state of Minas Gerais, Brazil, and is covered by Brazilian savannas vegetation, totalizing 1,596 ha. Semivariograms were generated after the exploratory data analysis. Models were fitted, validated and their metrics analyzed. The results showed a very clear trend where the shape of semivariograms, sill and range were different when deforestation occurred and were similar when the area had not been changed. The model that generated best fit was the Gaussian, however, the three models tested showed behavior that makes it possible to detect changes in vegetation. It suggests that further researches should explore the degree to which the semivariogram can be used to quantify this spatial variability as well as to analyze the influence of sazonality for changing detection in vegetated areas.

Assessment of geostatistical features for object-based image classification of contrasted landscape vegetation cover

Accurate mapping and monitoring of savanna and semiarid woodland biomes are needed to support the selection of areas of conservation, to provide sustainable land use, and to improve the understanding of vegetation. The potential of geostatistical features, derived from medium spatial resolution satellite imagery, to characterize contrasted landscape vegetation cover and improve object-based image classification is studied. The study site in Brazil includes cerrado sensu stricto, deciduous forest, and palm swamp vegetation cover. Sentinel 2 and Landsat 8 images were acquired and divided into objects, for each of which a semivariogram was calculated using near-infrared (NIR) and normalized difference vegetation index (NDVI) to extract the set of geostatistical features. The features selected by principal component analysis were used as input data to train a random forest algorithm. Tests were conducted, combining spectral and geostatistical features. Change detection evaluation was performed using a confusion matrix and its accuracies. The semivariogram curves were efficient to characterize spatial heterogeneity, with similar results using NIR and NDVI from Sentinel 2 and Landsat 8. Accuracy was significantly greater when combining geostatistical features with spectral data, suggesting that this method can improve image classification results.Accurate mapping and monitoring of savanna and semiarid woodland biomes are needed to support the selection of areas of conservation, to provide sustainable land use, and to improve the understanding of vegetation. The potential of geostatistical features, derived from medium spatial resolution satellite imagery, to characterize contrasted landscape vegetation cover and improve object-based image classification is studied. The study site in Brazil includes cerrado sensu stricto, deciduous forest, and palm swamp vegetation cover. Sentinel 2 and Landsat 8 images were acquired and divided into objects, for each of which a semivariogram was calculated using near-infrared (NIR) and normalized difference vegetation index (NDVI) to extract the set of geostatistical features. The features selected by principal component analysis were used as input data to train a random forest algorithm. Tests were conducted, combining spectral and geostatistical features. Change detection evaluation was performed using a confusion matrix and its accuracies. The semivariogram curves were efficient to characterize spatial heterogeneity, with similar results using NIR and NDVI from Sentinel 2 and Landsat 8. Accuracy was significantly greater when combining geostatistical features with spectral data, suggesting that this method can improve image classification results.

Object-based land-cover change detection applied to Brazilian seasonal savannahs using geostatistical features

ABSTRACT A new method for remote-sensing land-use/land-cover (LULC) change detection is proposed to eliminate the effects of forest phenology on classification results. This method is insensitive to spectral changes caused by vegetation seasonality and uses an object-based approach to extract geostatistical features from bitemporal Landsat TM (Thematic Mapper) images. We first create image objects by multiresolution segmentation to extract geostatistical features (semivariogram parameters and indices) and spectral information (average values) from NDVI (normalized difference vegetation index), acquired in the wet and dry seasons, as input data to train a Support Vector Machine algorithm. We also used the image difference traditional change-detection method to validate the effectiveness of the proposed method. We used two classes: (1) LULC change class and (2) seasonal change class. Using the most geostatistical features, the change detection results are considerably improved compared with the spectral features and image differencing technique. The highest accuracy was achieved by the sill (σ2 overall variability) semivariogram parameter (95%) and the AFM (area first lag–first maximum) semivariogram index (88.33%), which were not affected by vegetation seasonality. The results indicate that the geostatistical context makes possible the use of bitemporal NDVI images to address the challenge of accurately detecting LULC changes in Brazilian seasonal savannahs, disregarding changes caused by phenological differences, without using a dense time series of remote-sensing images. The challenge of extracting accurate semivariogram curves from objects of long and narrow shapes requires further study, along with the relationship between the scale of segmentation and image spatial resolution, including the type of change and the initial land-cover class.

Determinação do volume de madeira em povoamento de eucalipto por escâner a laser aerotransportado

The objective of this work was to evaluate the possibility of estimating the diameter at breast height (DBH) with tree height and number data derived from airborne laser scanning (LiDAR, light detection and ranging) dataset, and to determine the timber volume of an Eucalyptus sp. stand from these variables. The total number of detected trees was obtained using a local maxima filtering. Plant height estimated by LiDAR showed a nonsignificant tendency to underestimation. The estimate for DBH was coherent with the results found in the forest inventory; however, it also showed a tendency towards underestimation due to the observed behavior for height. The variable number of stems showed values close to the ones observed in the inventory plots. LiDAR underestimated the total timber volume in the stand in 11.4%, compared to the total volume delivered to the industry. The underestimation tendency of tree height (5% mean value) impacted the individual tree volume estimate and, consequently, the stand volume estimate. However, it is possible to obtain regression equations that estimate DBH with good precision, from the LiDAR plant height derived data. The parabolic model is the one that provides the best estimates for timber volumetric yield of eucalyptus stands.

CONTINUIDADE ESPACIAL DE CARACTERÍSTICAS DENDROMÉTRICAS EM POVOAMENTOS CLONAIS DE EUCALYPTUS SP. AVALIADA AO LONGO DO TEMPO

O objetivo deste estudo foi avaliar a continuidade espacial ao longo do tempo (para 5 idades consecutivas) das caracteristicas dendrometricas altura dominante media, volume e incremento medio anual em povoamentos clonais de Eucalyptus sp no estado de Minas Geais. A area foi plantada em 2003, perfazendo um total de 1.072,6 hectares. Os dados foram oriundos de um conjunto de 116 parcelas permanentes, onde foi realizado inventario florestal sucessivo entre os anos de 2006 e 2010. Aos semivariogramas experimentais, foram ajustados os modelos esferico, exponencial e gaussiano pelo Metodo dos Minimos Quadrados Ponderados, para cada idade, selecionando-se o mais adequado com base no erro medio reduzido e desvio padrao do erro medio reduzido da validacao cruzada. O comportamento dos semivariogramas foram comparados entre as respectivas medicoes com base no plotagem dos mesmos de forma escalonada, permitindo avaliar se a estrutura espacial foi alterada com a idade do plantio. O modelo exponencial apresentou-se como o de melhor ajuste e todas as caracteristicas apresentaram-se estruturadas espacialmente, com os modelos espaciais sendo semelhantes entre as idades do povoamento e para as caracteristicas estudadas. Foi possivel verificar que o grau de continuidade espacial se manteve ao longo dos anos para as caracteristicas avaliadas e que o semivariograma escalonado demonstrou que a estrutura espacial das caracteristicas e semelhante entre as diferentes idades do povoamento. Os resultados evidenciam que a utilizacao de metodos geoestatisticos para avaliar o crescimento de povoamentos de eucalipto ao longo do tempo, consiste de uma importante ferramenta de planejamento, permitindo um melhor acompanhamento e uma predicao mais precisa do volume de madeira da floresta, levando em consideracao a estrutura de dependencia espacial.

Object-based change detection using semivariogram indices derived from NDVI images: The environmental disaster in Mariana, Brazil

Object-based change detection is a powerful analysis tool for remote sensing data, but few studies consider the potential of temporal semivariogram indices for mapping land-cover changes using object-based approaches. In this study, we explored and evaluated the performance of semivariogram indices calculated from remote sensing imagery, using the Normalized Differential Vegetation Index (NDVI) to detect changes in spatial features related to land cover caused by a disastrous 2015 dam failure in Brazil’s Mariana district. We calculated the NDVI from Landsat 8 images acquired before and after the disaster, then created objects by multiresolution segmentation analysis based on post-disaster images. Experimental semivariograms were computed within the image objects and semivariogram indices were calculated and selected by principal component analysis. We used the selected indices as input data to a support vector machine algorithm for classifying change and no-change classes. The selected semivariogram indices showed their effectiveness as input data for object-based change detection analysis, producing highly accurate maps of areas affected by post-dam-failure flooding in the region. This approach can be used in many other contexts for rapid and accurate assessment of such land-cover changes.

Using Spatial Features to Reduce the Impact of Seasonality for Detecting Tropical Forest Changes from Landsat Time Series

In forested areas that experience strong seasonality and are undergoing rapid land cover conversion (e.g., Brazilian savannas), the accuracy of remote sensing change detection is affected by seasonal changes that are erroneously classified as having changed. To improve the quality and consistency of regionally important forest change maps, we aim to separate process related change (for example, spectral variability due to phenology) from changes related to deforestations or fires. Seasonal models are typically used to account for seasonality, but fitting a model is difficult when there are insufficient data points in the time series. In this research, we utilize remotely sensed data and related spectral trends and the spatial context at the object level to evaluate the performance of geostatistical features to reduce the impact of seasonality from the NDVI (Normalized Difference Vegetation Index) of Landsat time series. The study area is the Sao Romao municipality, totaling 2440 km2, and is part of the Brazilian savannas biome. We first create image objects via multiresolution segmentation, basing the objects on the characteristics found in the first image (2003) of the 13-year time series. We intersected the objects with the NDVI images in order to extract semivariogram indices, the RVF (Ratio Variance—First lag) and AFM (Area First lag—First Maximum), and spectral information (average and standard deviation of NDVI values) to generate the time series from these features and to derive Spatio-Temporal Metrics (change and trend) to train a Random Forest (RF) algorithm. The NDVI spatial variability, captured by the AFM semivariogram index time series produced the best result, reaching 96.53% of the overall accuracy (OA) to separate no-change from forest change, while the greatest inter-class confusion occurred using the average of the NDVI values time series (OA = 63.72%). The spatial context approach we presented is a novel approach for the detection of forest change events that are subject to seasonality (and possible miss-classification of change) and mitigating the effects of forest phenology without the need for specific de-seasoning models.

CHARACTERIZING LANDSCAPE SPATIAL HETEROGENEITY USING SEMIVARIOGRAM PARAMETERS DERIVED FROM NDVI IMAGES

The authors are grateful to the Coordenacao de Aperfeicoamento de Pessoal de Nivel Superior (CAPES), Department of Forest Science of the Federal University of Lavras (UFLA) and the ONF Brazil group for supporting this work.