Remote Sens.2015, 7(9), 11061-11082; doi:10.3390/rs70911061 (registering DOI) - published 27 August 2015 Show/Hide Abstract
Abstract: The upper ranges of the northern Andes are characterized by unique Neotropical, high altitude ecosystems known as paramos. These tundra-like grasslands are widely recognized by the scientific community for their biodiversity and their important ecosystem services for the local human population. Despite their remoteness, limited accessibility for humans and waterlogged soils, paramos are highly flammable ecosystems. They are constantly under the influence of seasonal biomass burning mostly caused by humans. Nevertheless, little is known about the spatial extent of these fires, their regime and the resulting ecological impacts. This paper presents a thorough mapping and analysis of the fires in one of the world’s largest paramo, namely the “Complejo de Páramos” of Cruz Verde-Sumapaz in the Eastern mountain range of the Andes (Colombia). Landsat TM/ETM+ and MODIS imagery from 2001 to 2013 was used to map and analyze the spatial distribution of fires and their intra- and inter-annual variability. Moreover, a logistic regression model analysis was undertaken to test the hypothesis that the dynamics of the paramo fires can be related to human pressures. The resulting map shows that the burned paramo areas account for 57,179.8 hectares, of which 50% (28,604.3 hectares) are located within the Sumapaz National Park. The findings show that the fire season mainly occurs from January to March. The accuracy assessment carried out using a confusion matrix based on 20 reference burned areas shows values of 90.1% (producer accuracy) for the mapped burned areas with a Kappa Index of Agreement (KIA) of 0.746. The results of the logistic regression model suggest a significant predictive relevance of the variables road distance (0.55 ROC (receiver operating characteristic)) and slope gradient (0.53 ROC), indicating that the higher the probability of fire occurrence, the smaller the distance to the road and the higher the probability of more gentle slopes. The paper sheds light on fires in the Colombian paramos and provides a solid basis for further investigation of the impacts on the natural ecosystem functions and biodiversity.
Remote Sens.2015, 7(9), 11036-11060; doi:10.3390/rs70911036 (registering DOI) - published 27 August 2015 Show/Hide Abstract
Abstract: Generating accurate and unbiased wall-to-wall canopy height maps from airborne lidar data for large regions is useful to forest scientists and natural resource managers. However, mapping large areas often involves using lidar data from different projects, with varying acquisition parameters. In this work, we address the important question of whether one can accurately model canopy heights over large areas of the Southeastern US using a very heterogeneous dataset of small-footprint, discrete-return airborne lidar data (with 76 separate lidar projects). A unique aspect of this effort is the use of nationally uniform and extensive field data (~1800 forested plots) from the Forest Inventory and Analysis (FIA) program of the US Forest Service. Preliminary results are quite promising: Over all lidar projects, we observe a good correlation between the 85th percentile of lidar heights and field-measured height (r = 0.85). We construct a linear regression model to predict subplot-level dominant tree heights from distributional lidar metrics (R2 = 0.74, RMSE = 3.0 m, n = 1755). We also identify and quantify the importance of several factors (like heterogeneity of vegetation, point density, the predominance of hardwoods or softwoods, the average height of the forest stand, slope of the plot, and average scan angle of lidar acquisition) that influence the efficacy of predicting canopy heights from lidar data. For example, a subset of plots (coefficient of variation of vegetation heights <0.2) significantly reduces the RMSE of our model from 3.0–2.4 m (~20% reduction). We conclude that when all these elements are factored into consideration, combining data from disparate lidar projects does not preclude robust estimation of canopy heights.
Remote Sens.2015, 7(9), 11016-11035; doi:10.3390/rs70911016 - published 26 August 2015 Show/Hide Abstract
Abstract: One of the most important linkages that couple terrestrial carbon and water cycles is ecosystem water use efficiency (WUE), which is relevant to the reasonable utilization of water resources and farming practices. Eddy covariance techniques provide an opportunity to monitor the variability in WUE and can be integrated with Moderate Resolution Imaging Spectroradiometer (MODIS) observations. Scaling up in situ observations from flux tower sites to large areas remains challenging and few studies have been reported on direct estimation of WUE from remotely-sensed data. This study examined the main environmental factors driving the variability in WUE of corn/soybean croplands, and revealed the prominent role of solar radiation and temperature. Time-series of MODIS-derived enhanced vegetation indices (EVI), which are proxies for the plant responses to environmental controls, were also strongly correlated with ecosystem WUE, thereby implying great potential for remote quantification. Further, both performance of the indirect MODIS-derived WUE from gross primary productivity (GPP) and evapotranspiration (ET), and the direct estimates by exclusive use of MODIS EVI data were evaluated using tower-based measurements. The results showed that ecosystem WUE were overpredicted at the beginning and ending of crop-growth periods and severely underestimated during the peak periods by the indirect estimates from MODIS products, which was mainly attributed to the error source from MODIS GPP. However, a simple empirical model that is solely based on MODIS EVI data performed rather well to capture the seasonal variations in WUE, especially for the growing periods of croplands. Independent validation at different sites indicates the method has potential for broad application.
Remote Sens.2015, 7(8), 10996-11015; doi:10.3390/rs70810996 - published 24 August 2015 Show/Hide Abstract
Abstract: Automatic extraction of ground points, called filtering, is an essential step in producing Digital Terrain Models from airborne LiDAR data. Scene complexity and computational performance are two major problems that should be addressed in filtering, especially when processing large point cloud data with diverse scenes. This paper proposes a fast and intelligent algorithm called Semi-Global Filtering (SGF). The SGF models the filtering as a labeling problem in which the labels correspond to possible height levels. A novel energy function balanced by adaptive ground saliency is employed to adapt to steep slopes, discontinuous terrains, and complex objects. Semi-global optimization is used to determine labels that minimize the energy. These labels form an optimal classification surface based on which the points are classified as either ground or non-ground. The experimental results show that the SGF algorithm is very efficient and able to produce high classification accuracy. Given that the major procedure of semi-global optimization using dynamic programming is conducted independently along eight directions, SGF can also be paralleled and sped up via Graphic Processing Unit computing, which runs at a speed of approximately 3 million points per second.
Remote Sens.2015, 7(8), 10973-10995; doi:10.3390/rs70810973 - published 24 August 2015 Show/Hide Abstract
Abstract: Vegetation phenology is a key biological indicator for monitoring terrestrial ecosystems and global change, and regions with the most obvious phenological changes in vegetation are primarily located at high latitudes and altitudes. Over the past three decades, investigations of obvious phenological changes in vegetation at middle and high latitudes in the Northern Hemisphere have provided significant contributions to understanding global climate change. In this study, phenological parameters were extracted from the Global Inventory Modeling and Mapping Studies (GIMMS) Normalized Difference Vegetation Index (NDVI3g) to analyze the spatial and temporal characteristics of vegetation phenological changes above 40°N in the Northern Hemisphere from 1982–2013. The results showed that the start of season (SOS) was significantly advanced (−2.2 ± 0.6 days·decade−1, p < 0.05) and that the end of season (EOS) was slightly delayed (0.78 ± 0.6 days·decade−1, p = 0.21) over the entire study area in the initial 21 years (1982–2002). When the time scale was extended to 2013, the change rate of the SOS and EOS was significantly reduced; in addition, the SOS was delayed (3.2 ± 1.7 days·decade−1, p < 0.05), and the EOS was advanced (4.5 ± 0.9 days·decade−1, p < 0.05) over the entire study area in the last 11 years (2003–2013). The trends of advanced SOS and delayed EOS over the past three decades were slower than those over the initial two decades on a hemispheric scale. The change trends showed obvious variability with different vegetation types and were greater for woody plants than for herbaceous plants. For broad-leaved forest, the SOS was significantly advanced (2 ± 0.5 days·decade−1, p < 0.05) and the EOS was significantly delayed (2.7 ± 0.6 days·decade−1, p < 0.05) from 1982–2013. The trend of delayed EOS was greater than that of advanced SOS for different vegetation types. With respect to the spatial distribution of phenological trends in the Northern Hemisphere, the trends of advanced SOS and delayed EOS were strongest in Europe followed by North America, and the trends were least significant in Asia. Coniferous forest, shrub forest, grassland, and the entire study area have the same change trends for the two time periods (1982–2002 and 2003–2013), and the increased rate of the phenology parameters has decelerated over the most recent decade. The length of season (LOS) of broad-leaved forest and mixed forest over the past 32 years shows a strong increased trend, and simultaneously, the SOS and EOS show an advanced trend and a delayed trend, respectively
Remote Sens.2015, 7(8), 10951-10972; doi:10.3390/rs70810951 - published 24 August 2015 Show/Hide Abstract
Abstract: Land surface albedo data with high spatio-temporal resolution are increasingly important for scientific studies addressing spatially and/or temporally small-scale phenomena, such as urban heat islands and urban land surface energy balance. Our previous study derived albedo data with 2–4-day and 30-m temporal and spatial resolution that have better spatio-temporal resolution than existing albedo data, but do not completely satisfy the requirements for monitoring high-frequency land surface changes at the small scale. Downscaling technology provides a chance to further improve the albedo data spatio-temporal resolution and accuracy. This paper introduces a method that combines downscaling technology for land surface reflectance with an empirical method of deriving land surface albedo. Firstly, downscaling daily MODIS land surface reflectance data (MOD09GA) from 500 m to 30 m on the basis of HJ-1A/B BRDF data with 2–4-day and 30-m temporal and spatial resolution is performed: this is the key step in the improved method. Subsequently, the daily 30-m land surface albedo data are derived by an empirical method combining prior knowledge of the MODIS BRDF product and the downscaled daily 30-m reflectance. Validation of albedo data obtained using the proposed method shows that the new method has both improved spatio-temporal resolution and good accuracy (a total absolute accuracy of 0.022 and a total root mean squared error at six sites of 0.028).