Climate and soil are among the most important factors determining variation in tree communities, but their effects have not been thoroughly elucidated to date for many vegetation features. In this study, we evaluate how climate and soil gradients affect gradients of vegetation composition, species diversity and dominance, structure and functional traits (seed mass and wood density) using over 327 000 trees in 158 sites distributed along environmental gradients in the transitions among the Atlantic forest, Cerrado and Caatinga in Minas Gerais State (MG), Brazil (nearly 600 000 km²). Gradients in species, genus and family abundance in addition to basal area, stem density, species diversity (Fisher’s alpha), dominance percentage, seed mass and wood density were correlated using multiple regressions with environmental variables, as summarized in four principal component analysis axes (two climatic—precipitation seasonality and temperature range—and two edaphic—soil fertility and soil moisture). Additionally, ordinary kriging maps were used to better illustrate the gradients. Multiple regression models indicate that all variables but dominance percentage were affected by one or more of the environmental gradients, but the average R² was low (26.25%). Kriging maps reinforced the patterns observed in the regression models. Precipitation seasonality and soil moisture gradients were the most important gradients affecting vegetation features. This finding suggests that water availability is an important determinant of vegetation features in these vegetation transitions.

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 São Romão 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.

Change detection methods are often incapable of accurately detecting changes within time series that are heavily influenced by seasonal variations. Techniques for de-seasoning time series or methods that apply the spatial context have been used to improve the results of change detection. However, few studies have explored Landsat’s shortwave infrared channel (SWIR 2) to discriminate between seasonal changes and land use/land cover changes (LULCC). Here, we explored the effectiveness of Operational Land Imager (OLI) spectral bands and vegetation indices for detecting deforestation in highly seasonal areas of Brazilian savannas. We adopted object-based image analysis (OBIA), applying a multidate segmentation to an OLI time series to generate input data for discrimination of deforestation from seasonal changes using the Random Forest (RF) algorithm. We found adequate separability between deforested objects and seasonal changes using SWIR 2. Using spectral indices computed from SWIR 2, the RF algorithm generated a change map with an overall accuracy of 88.3%. For deforestation, the producer’s accuracy was 88.0% and the user’s accuracy was 84.6%. The SWIR 2 channel as well as the mid-infrared burn index presented the highest importance among spectral variables computed by the RF average impurity decrease measure. Our results give support to further change detection studies regarding to suitable spectral channels and provided a useful foundation for savanna change detection using an object-based method applied to Landsat time series.

We present the Change, Aftereffect, and Trend (CAT) transform for visualizing and analyzing landscape dynamics from dense, multi-annual satellite vegetation index (VI) time series. The transform compresses a temporally detailed, multi-annual VI dataset into three new variables capturing change events and trends occurring within that period. First, peak annual greenness is extracted from each year. Then a series of simple calculations generate the three CAT variables: 1) Change: the maximum interannual absolute difference in peak greenness between consecutive years; 2) Aftereffect: the mean peak greenness after Change occurred; and 3) Trend: the slope of a linear regression applied to the entire annual peak greenness time series. We demonstrate the CAT transform by applying it to a MODIS 16-day 250-m normalized difference VI (NDVI) dataset covering the province of Alberta, Canada, for 2001 through 2011. We find that the CAT variables capture much of the non-seasonal change in the original NDVI time series. When displayed as an RGB color composite (the CAT image), the transform provides a striking visualization of both drastic and gradual decadal-scale landscape dynamics. Its application to quantitative analyses is demonstrated by an urban sprawl case study conducted around the city of Calgary, Alberta, where a simple decision-tree-based classification of the CAT transform variables was superior to a bitemporal, image-differencing approach. The simple yet powerful CAT transform is easily applicable to other study areas and datasets, and could foster a wider usage and understanding of the many archived high-temporal-resolution satellite datasets currently available.

Detecting disturbances in native vegetation is a crucial component of many environmental management strategies, and remote sensing-based methods are the most efficient way to collect multi-temporal disturbance data over large areas. Given that there is a large range of datasets for monitoring, analyzing, and detecting disturbances, many methods have been well-studied and successfully implemented. However, factors such as the vegetation type, input data, and change detection method can significantly alter the outcomes of a disturbance-detection study. We evaluated the spatial agreement of disturbance maps provided by the Breaks For Additive Season and Trend (BFAST) algorithm, evaluating seven spectral indices in three distinct vegetation domains in Brazil: Atlantic forest, savanna, and semi-arid woodland, by assessing levels of agreement between the outputs. We computed individual map accuracies based on a reference dataset, then ranked their performance, while also observing their relationships with specific vegetation domains. Our results indicated a low rate of spatial agreement among index-based disturbance maps, which itself was minimally influenced by vegetation domain. Wetness indices produced greater detection accuracies in comparison to greenness-related indices free of saturation. The normalized difference moisture index performed best in the Atlantic forest domains, yet performed poorest in semi-arid woodland, reflecting its specific sensitivity to vegetation and its water content. The normalized difference vegetation index led to high disturbance detection accuracies in the savanna and semi-arid woodland domains. This study offered novel insight into vegetation disturbance maps, their relationship to different ecosystem types, and corresponding accuracies. Distinct input data can produce non-spatially correlated disturbance maps and reflect site-specific sensitivity. Future research should explore algorithm limitations presented in this study, as well as the expansion to other techniques and vegetation domains across the globe.