MODIS Vegetation Continuous Fields Research Articles

Evaluating global vegetation products for application in heterogeneous forest-savanna landscapes

ABSTRACT While satellite-derived global vegetation structure products are powerful and easy to use, their utility for studying spatial patterns within heterogeneous landscapes such as forest-savanna mosaics has not been extensively evaluated. We explored the application of global vegetation structure products in heterogeneous landscapes by comparing them with Airborne Laser Scanning. Specifically, we assessed the accuracy and bias of two fractional cover products, MODIS Vegetation Continuous Fields (VCF) and Hansen Global Forest Change (GFC), and one global canopy height model, Global Forest Canopy Height Model (CHM), in comparison with the same variables derived from local ALS point clouds. We found that there were limitations to all three products. MODIS VCF was less accurate than its reported accuracy by at least 3%, and GFC was 10% less accurate than MODIS VCF. While Global CHM had a similar magnitude of error to its reported product accuracy, product agreement was much lower (R2 0.19 vs. R2 0.61). We also found that the context of the analysis is important when choosing whether to use one fractional product over the other. Global products should be applied with caution in heterogeneous landscapes. Increased training and validation from these landscapes could improve the performance of these products and their utility for landscape-scale ecological research.

Alternative biome states of African terrestrial vegetation and the potential drivers: A continental-scale study

The alternative stable state (ASS) theory provides a plausible framework to explain the spatial distribution of biomes and their dynamics. Existing studies to test the alternative biome states (ABSs) mainly focused on tree-dominated biomes. It is still uncertain whether ABSs are present in a wide range of terrestrial biomes. This study was to examine the ABSs in the terrestrial vegetated areas of Africa and the maintaining factors. The potential landscapes were reconstructed separately for forest, grassland and shrubland using the MODIS Vegetation Continuous Field data along the gradients of temperature, precipitation and aridity index (AI, the ratio of precipitation to potential evapotranspiration). The differences of soil organic carbon density (SOCD), fire count and grazing intensity were compared to test the feedback hypothesis to maintain the ABSs. The results showed that AI performed well in detecting the ABSs at the continental scale of Africa. Forest (at the wetter end) and shrubland (at the drier end) were well separated along the AI axis. Forest had three stable states (i.e. closed forest, woody savanna, and savanna) and shrubland had two stable states (i.e. closed shrubland and open shrubland). Grassland had two stable states (i.e. dense grassland and sparse grassland) distributing in a large AI range. The stable states that shared a specific AI range were regarded as the ABSs. Climate aridity greatly determined the distribution of the ABSs but the positive feedbacks between vegetation and SOCD, fire count, and livestock density played potential roles in driving the shifts between the ABSs. Our study indicated that the ABSs commonly existed in varied biomes (both tree-dominated and non-tree-dominated) in the African continent, which provided an enlarged picture of the ABSs of the terrestrial biomes. The findings contribute to a deeper understanding of large scale vegetation patterns and their dynamics and facilitate to macro management of the terrestrial ecosystems in facing the possible regime shifts of the biomes.

Characterizing spatial patterns of pine bark beetle outbreaks during the dry and rainy season’s in Honduras with the aid of geographic information systems and remote sensing data

Coniferous and mixed forests cover approximately 42% of Honduras forested areas, however; pine bark beetle (PBB) (Dendroctonus spp.) outbreaks are an environmental hazard that has caused incalculable ecological and economic impacts in Honduras. In this research, in order to plan more focalized measures for controlling the PBB outbreaks, it is essential to identify those areas that have a high susceptibility to a PBB outbreak during the dry and rainy seasons. For this purpose, we require to associate the historical PBB outbreak points (2017–2019), with a series of environmental and anthropogenic variables that according to the literature review have affected the initiation and the spread of PBB outbreaks. To assess the current climatic variables we used MODIS land surface temperature product (LST-MOD11A2), wind speed, precipitation, and temperature acquired from the WorldClim data. To assess vegetation vigor, we use the Normalized drought moisture index estimated from the MODIS surface reflectance product (MOD09A3). To give us an understanding of the density of the forest, we compare two products, the MODIS Leaf Area Index (LAI-MOD15A2H), and the MODIS Vegetation Continuous Fields (VFC-MOD44B). We included elevation, aspect, and slope as variables and acquired this data from the Global multiresolution terrain elevation data (GMTED-2010). Finally, we used geographical information systems data to derive proximity to different types of roads as anthropogenic data and wildfire density. We do a preliminary analysis of the variables and eliminated those which show the least importance. Furthermore, we integrated the most relevant variables identified with the PBB points by using the Random Forest (RF) algorithm to fit the model and then predicted the current PBB outbreak susceptibility for the dry and rainy season. Results indicated that climatic variables weigh heavily in determining high susceptibility areas. Our prediction results show high and very high susceptibility in the North-eastern and Central parts of the country especially. The results acquired, can lead to improved preventive and control measures to reduce the negative ecological effects that are caused by PBB outbreaks.

Climatic and vegetational drivers of insect beta diversity at the continental scale.

AimWe construct a framework for mapping pattern and drivers of insect diversity at the continental scale and use it to test whether and which environmental gradients drive insect beta diversity.LocationGlobal; North and Central America; Western Europe.Time period21st century.Major taxa studiedInsects.MethodsAn informatics system was developed to integrate terrestrial data on insects with environmental parameters. We mined repositories of data for distribution, climatic data were retrieved (WorldClim), and vegetation parameters inferred from remote sensing analysis (MODIS Vegetation Continuous Fields). Beta diversity between sites was calculated and then modeled with two methods, Mantel test with multiple regression and generalized dissimilarity modeling.ResultsGeographic distance was the main driver of insect beta diversity. Independent of geographic distance, bioclimate variables explained more variance in dissimilarity than vegetation variables, although the particular variables found to be significant were more consistent in the latter, particularly, tree cover. Tree cover gradients drove compositional dissimilarity at denser coverages, in both continental case studies. For climate, gradients in temperature parameters were significant in driving beta diversity more so than gradients in precipitation parameters.Main conclusionsAlthough environmental gradients drive insect beta diversity independently of geography, the relative contribution of different climatic and vegetational parameters is not expected to be consistent in different study systems. With further incorporation of additional temporal information and variables, this approach will enable the development of a predictive framework for conserving insect biodiversity at the global scale.

Power laws and critical fragmentation in global forests

The replacement of forest areas with human-dominated landscapes usually leads to fragmentation, altering the structure and function of the forest. Here we studied the dynamics of forest patch sizes at a global level, examining signals of a critical transition from an unfragmented to a fragmented state, using the MODIS vegetation continuous field. We defined wide regions of connected forest across continents and big islands, and combined five criteria, including the distribution of patch sizes and the fluctuations of the largest patch over the last sixteen years, to evaluate the closeness of each region to a fragmentation threshold. Regions with the highest deforestation rates–South America, Southeast Asia, Africa–all met these criteria and may thus be near a critical fragmentation threshold. This implies that if current forest loss rates are maintained, wide continental areas could suddenly fragment, triggering extensive species loss and degradation of ecosystems services.

Simultaneous comparison and assessment of eight remotely sensed maps of Philippine forests

This article compares and assesses eight remotely sensed maps of Philippine forest cover in the year 2010. We examined eight Forest versus Non-Forest maps reclassified from eight land cover products: the Philippine Land Cover, the Climate Change Initiative (CCI) Land Cover, the Landsat Vegetation Continuous Fields (VCF), the MODIS VCF, the MODIS Land Cover Type product (MCD12Q1), the Global Tree Canopy Cover, the ALOS-PALSAR Forest/Non-Forest Map, and the GlobeLand30. The reference data consisted of 9852 randomly distributed sample points interpreted from Google Earth. We created methods to assess the maps and their combinations. Results show that the percentage of the Philippines covered by forest ranges among the maps from a low of 23% for the Philippine Land Cover to a high of 67% for GlobeLand30. Landsat VCF estimates 36% forest cover, which is closest to the 37% estimate based on the reference data. The eight maps plus the reference data agree unanimously on 30% of the sample points, of which 11% are attributable to forest and 19% to non-forest. The overall disagreement between the reference data and Philippine Land Cover is 21%, which is the least among the eight Forest versus Non-Forest maps. About half of the 9852 points have a nested structure such that the forest in a given dataset is a subset of the forest in the datasets that have more forest than the given dataset. The variation among the maps regarding forest quantity and allocation relates to the combined effects of the various definitions of forest and classification errors. Scientists and policy makers must consider these insights when producing future forest cover maps and when establishing benchmarks for forest cover monitoring.

В. Аналіз точності глобальних карт зімкнутості деревостанів для території рівнинної частини України

В. Аналіз точності глобальних карт зімкнутості деревостанів для території рівнинної частини України

Evaluating MODIS-vegetation continuous field products to assess tree cover change and forest fragmentation in India – A multi-scale satellite remote sensing approach

Monitoring the changes in forest-cover and understanding the dynamics of the forest is becoming increasingly important for the sustainable management of forest ecosystems. This paper uses temporal MODIS Vegetation Continuous Field (MODIS-VCF) to monitor the tree cover change in the Indian region over a period of 6years (2000–2005). Pixel-based linear regression model is developed to identify rate of deforestation and fragmentation at landscape level. The regression parameters viz., slope, offset and variance are used to identify threshold between forest and non-forest classes. The classification algorithm resulted into change area, no change area, positive change and negative changes. MODIS-VCF raw product of 2005 was validated using the field data and showed a coefficient of determination (R2=0.85) between percent tree cover and individual plot wise canopy cover information. The results were overlaid with UNEP protected area boundary. On a long-term basis, the forest cover change was monitored using medium spatial resolution (Landsat and IRS) satellite data to identify the rate of deforestation and fragmentation at landscape level. The developed approach is efficient and effective for regional monitoring of forest cover change. It could be automated for regular usage and monitoring.

Multiple remote sensing data sources to assess spatio-temporal patterns of fire incidence over Campos Amazônicos Savanna Vegetation Enclave (Brazilian Amazon)

Fire activity plays an important role in the past, present and future of Earth system behavior. Monitoring and assessing spatial and temporal fire dynamics have a fundamental relevance in the understanding of ecological processes and the human impacts on different landscapes and multiple spatial scales. This work analyzes the spatio-temporal distribution of burned areas in one of the biggest savanna vegetation enclaves in the southern Brazilian Amazon, from 2000 to 2016, deriving information from multiple remote sensing data sources (Landsat and MODIS surface reflectance, TRMM pluviometry and Vegetation Continuous Field tree cover layers). A fire scars database with 30 m spatial resolution was generated using a Landsat time series. MODIS daily surface reflectance was used for accurate dating of the fire scars. TRMM pluviometry data were analyzed to dynamically establish time limits of the yearly dry season and burning periods. Burned area extent, frequency and recurrence were quantified comparing the results annually/seasonally. Additionally, Vegetation Continuous Field tree cover layers were used to analyze fire incidence over different types of tree cover domains. In the last seventeen years, 1.03millionha were burned within the study area, distributed across 1432 fire occurrences, highlighting 2005, 2010 and 2014 as the most affected years. Middle dry season fires represent 86.21% of the total burned areas and 32.05% of fire occurrences, affecting larger amount of higher density tree surfaces than other burning periods. The results provide new insights into the analysis of burned areas of the neotropical savannas, spatially and statistically reinforcing important aspects linked to the seasonality patterns of fire incidence in this landscape.

Forest cover mapping in post-Soviet Central Asia using multi-resolution remote sensing imagery

Despite rapid advances and large-scale initiatives in forest mapping, reliable cross-border information about the status of forest resources in Central Asian countries is lacking. We produced consistent Central Asia forest cover (CAFC) maps based on a cost-efficient approach using multi-resolution satellite imagery from Landsat and MODIS during 2009–2011. The spectral-temporal metrics derived from 2009–2011 Landsat imagery (overall accuracy of 0.83) was used to predict sub-pixel forest cover on the MODIS scale for 2010. Accuracy assessment confirmed the validity of MODIS-based forest cover map with a normalized root-mean-square error of 0.63. A general paucity of forest resources in post-Soviet Central Asia was indicated, with 1.24% of the region covered by forest. In comparison to the CAFC map, a regional map derived from MODIS Vegetation Continuous Fields tended to underestimate forest cover, while the Global Forest Change product matched well. The Global Forest Resources Assessments, based on individual country reports, overestimated forest cover by 1.5 to 147 times, particularly in the more arid countries of Turkmenistan and Uzbekistan. Multi-resolution imagery contributes to regionalized assessment of forest cover in the world’s drylands while developed CAFC maps (available at https://data.zef.de/) aim to facilitate decisions on biodiversity conservation and reforestation programs in Central Asia.

Analysis of the pattern of potential woody cover in Texas savanna

While woody plant encroachment has been observed worldwide in savannas and adversely affected the ecosystem structure and function, a thorough understanding of the nature of this phenomenon is urgently required for savanna management and restoration. Among others, potential woody cover (the maximum realizable woody cover that a given site can support), especially its variation over environment has huge implication on the encroachment management in particular, and on tree-grass interactions in general. This project was designed to explore the pattern of potential woody cover in Texas savanna, an ecosystem with a large rainfall gradient in west–east direction. Substantial random pixels were sampled across the study area from MODIS Vegetation Continuous Fields (VCF) tree cover layer (250m). Since potential woody cover is suggested to be limited by water availability, a nonlinear 99th quantile regression was performed between the observed woody cover and mean annual precipitation (MAP) to model the pattern of potential woody cover. Research result suggests a segmented relationship between potential woody cover and MAP at MODIS scale. Potential biases as well as the practical and theoretical implications were discussed. Through this study, the hypothesis about the primary role of water availability in determining savanna woody cover was further confirmed in a relatively understudied US-located savanna.

Assessment of the three factors affecting Myanmar's forest cover change using Landsat and MODIS vegetation continuous fields data

ABSTRACTLong-term observation of the earth is essential for studying the factors affecting global environmental changes. Digital earth technology can facilitate the monitoring of global environmental change with its ability to process vast amounts of information. In this study, we map the forest cover change of Myanmar from 2000 to 2005 using a training data automation procedure and support vector machines algorithm. Our results show that Myanmar's forests have declined 0.68% annually over this six-year period. We validated our derived change results and found the overall accuracy to be greater than 88%. We also assessed forest loss from protected areas, areas close to roads, and areas subject to fire, which were most likely to lose forested area. The results revealed the main reasons for forest losses in some hotspots to be increased agricultural conversion, fire, and the construction of highways. This information is useful for identifying the driving forces behind forest changes and to support environmental policy development in Myanmar.

Validation of MODIS Vegetation Continuous Fields for monitoring deforestation and forest degradation: two cases in Mexico

This study assesses whether MODIS Vegetation Continuous Fields percent tree cover (PTC) data can detect deforestation and forest degradation. To assess the usefulness of PTC for detecting deforestation, we used a data set consisting of eight forest and seven non-forest categories. To evaluate forest degradation, we used data from two temperate forest types in three conservation states: primary (dense), secondary (moderately degraded) and open (heavily degraded) forest. Our results show that PTC can differentiate temperate forest from non-forest categories (p = 0.05) and thus suggests PTC can adequately detect deforestation in temperate forests. In contrast, single-date PTC data does not appear to be adequate to detect forest degradation in temperate forests. As for tropical forest, PTC can partially discriminate between forest and non-forest categories.

Development of a global hybrid forest mask through the synergy of remote sensing, crowdsourcing and FAO statistics

A number of global and regional maps of forest extent are available, but when compared spatially, there are large areas of disagreement. Moreover, there is currently no global forest map that is consistent with forest statistics from FAO (Food and Agriculture Organization of the United Nations). By combining these diverse data sources into a single forest cover product, it is possible to produce a global forest map that is more accurate than the individual input layers and to produce a map that is consistent with FAO statistics. In this paper we applied geographically weighted regression (GWR) to integrate eight different forest products into three global hybrid forest cover maps at a 1km resolution for the reference year 2000. Input products included global land cover and forest maps at varying resolutions from 30m to 1km, mosaics of regional land use/land cover products where available, and the MODIS Vegetation Continuous Fields product. The GWR was trained using crowdsourced data collected via the Geo-Wiki platform and the hybrid maps were then validated using an independent dataset collected via the same system. Three different hybrid maps were produced: two consistent with FAO statistics, one at the country and one at the regional level, and a “best guess” forest cover map that is independent of FAO. Independent validation showed that the “best guess” hybrid product had the best overall accuracy of 93% when compared with the individual input datasets. The global hybrid forest cover maps are available at http://biomass.geo-wiki.org.

Mapping wetland water depths over the central Congo Basin using PALSAR ScanSAR, Envisat altimetry, and MODIS VCF data

We have, for the first time, developed a simple linear regression model based on PALSAR ScanSAR backscattering coefficients (σ0), water levels from Envisat altimetry, and MODIS Vegetation Continuous Field (VCF) product to generate water depth maps over flooded forest in the central Congo Basin. The water depth maps we generated are relative to the lowest water level from Envisat altimetry, which is assumed to be a base level with essentially zero depth. The predicted and observed water depths along the Envisat pass showed excellent agreements with RMS differences of 12.8cm to 17.8cm. The water depth maps were also independently validated with ∂h/∂t obtained from PALSAR interferometry, and storage anomalies estimated by multiplying inundation extents from PALSAR ScanSAR with spatially averaged water level anomalies from Envisat altimetry. The water storage volumes calculated from our water depth maps were calculated to be 11.3±2.0km3, 10.3±2.3km3, and 9.3±1.8km3 for 12/05/2006, 12/08/2007, and 12/10/2008, respectively. It is expected that our method can be applied to other river basins that have flooded forest with satellite radar altimeter over-passes. The water depth maps can be directly used to calibrate and validate the spatial and temporal variation of inundation extent and water depth in the wetland derived from a 2-D hydrodynamic model. Furthermore, our water depth maps can be used as a “true” dataset to perform a pre-launch “virtual mission” study of the Surface Water Ocean Topography (SWOT) mission to be launched in 2020.

Regional-scale mapping of tree cover, height and main phenological tree types using airborne laser scanning data

Airborne Laser Scanning (ALS) data for generating Digital Terrain Models (DTM) at national level are often collected during leaf-off seasons. Leaf-off data are useful for classifying evergreen and deciduous trees since echoes at lower intensity are returned from deciduous trees when compared to evergreen trees. In addition to this, the proportion of echoes from the ground is higher for deciduous trees than for evergreen trees. In this study, we classified land cover, including evergreen and deciduous trees, using a Random Forest classifier based on LiDAR-metrics generated from leaf-off ALS data, and estimated tree cover and tree heights for the whole of Denmark. The results were compared with the CORINE Land Cover (CLC2006) data, percentage of tree cover from MODIS Vegetation Continuous Fields (VCF) and a global tree height map based on ICESat data.Considering tree class alone, deciduous and evergreen trees could be classified with an overall accuracy of 94% using validation data generated using aerial imagery from a 60-km strip across Central Jutland. The lower values of ALS-based percentage tree cover were overestimated and the higher values underestimated by MODIS VCF data, with a root–mean–square (RMS) deviation of 18.26%. The tree heights estimated using ALS data were generally lower than the global estimates of tree height with an RMS deviation of 5.1m. The ALS intensity values were useful for classifying evergreen and deciduous trees. These findings show that ALS datasets collected for generating national DTMs can be used for tree cover and tree height mapping, as well as for regional classification of trees if data over the whole area are collected within a few months in the leaf-off season.

Integrating global land cover products for improved forest cover characterization: an application in North America

Six widely used coarse-resolution global land cover data-sets – Global Land Cover Characterization (GLCC), Global Land Cover 2000 (GLC2000), GlobCover land cover product (GlobCover), MODIS land cover product (MODIS LC), the University of Maryland land cover product (UMD LC), and the MODIS Vegetation Continuous Fields tree cover layer (MODIS VCF) disagree substantially in their estimates of forest cover. Employing a regression tree model trained on higher-resolution, Landsat-based data, these multisource multiresolution maps were integrated for an improved characterization of forest cover over North America. Evaluated using a withheld test sample, the integrated percent forest cover (IPFC) data-set has a root mean square error of 11.75% – substantially better than the 17.37% of GLCC, 17.61% of GLC2000, 17.96% of GlobCover, 15.23% of MODIS LC, 19.25% of MODIS VCF, and 15.15% of UMD LC, respectively. Although demonstrated for forest, this approach based on integration of multiple products has potential for improved characterization of other land cover types as well.

Comparison of a New Percent Tree Cover Dataset with Existing One and Categorical Land Cover Datasets in Eurasia

The global tree cover percentage is an important parameter used to understand the global environment. However, the available percent tree cover products on global or continental-scale are few, and efforts to quantitatively validate these maps have been limited. We produced a new percent tree cover dataset at 500 m resolution in 2008 for Eurasia using reference data interpreted from Google Earth. It is a part of percent tree cover (PTC) data in Global Mapping project. In this study, the dataset was compared with existing global percent tree cover dataset, MODIS Vegetation Continuous Fields, MOD44B. We assessed the agreement of these datasets with two existing global categorical land cover datasets and statistic data in Eurasia. The result showed that estimates of tree cover in our new map and MOD44B were relatively similar at randomly sampled sites. Our map and MOD44B agreed with either or both of land cover maps at 93% of sites and 91% of sites, respectively, for pixel blocks. However, we found that MOD44B disagreed with our map and categorical land cover datasets at about half of the sampled sites where the difference of tree cover percentage between our map and MOD44B was large, especially in the areas with significant differences (more than 50%). Disagreed areas were concentrated in forests of Russia and Indonesia, and in herbaceous dominated vegetation of UK and Ireland. We also found that both our map and MOD44B were somewhat different from the data reported by FRA 2010.

Validation of modelled forest biomass in Germany using BETHY/DLR

Abstract. We present a new approach to the validation of modelled forest Net Primary Productivity (NPP), using empirical data on the mean annual increment, or MAI, in above-ground forest stock. The soil-vegetation-atmosphere-transfer model BETHY/DLR is used, with a particular focus on a detailed parameterization of photosynthesis, to estimate the NPP of forest areas in Germany, driven by remote sensing data from VEGETATION, meteorological data from the European Centre for Medium-Range Weather Forecasts (ECMWF), and additional tree coverage information from the MODIS Vegetation Continuous Field (VCF). The output of BETHY/DLR, Gross Primary Productivity (GPP), is converted to NPP by subtracting the cumulative plant maintenance and growth respiration, and then validated against MAI data that was calculated from German forestry inventories. Validation is conducted for 2000 and 2001 by converting modelled NPP to stem volume at a regional level. Our analysis shows that the presented method fills an important gap in methods for validating modelled NPP against empirically derived data. In addition, we examine theoretical energy potentials calculated from the modelled and validated NPP, assuming sustainable forest management and using species-specific tree heating values. Such estimated forest biomass energy potentials play an important role in the sustainable energy debate.

Large area forest stem volume mapping in the boreal zone using synergy of ERS-1/2 tandem coherence and MODIS vegetation continuous fields

ERS-1/2 tandem coherence was reported to have high potential for the mapping of boreal forest stem volume (e.g. Santoro et al., 2002, 2007a; Wagner et al., 2003; Askne & Santoro, 2005). Large-scale application of the data for forest stem volume mapping, however, is hindered by the variability of coherence with meteorological and environmental acquisition conditions. The traditional way of stem volume retrieval is based on the training of models, relating coherence to stem volume, with the aid of forest inventory data which is generally available for a few small test sites but not for large areas. In this paper a new approach is presented that allows model training using the MODIS Vegetation Continuous Fields canopy cover product (Hansen et al., 2003) without further need for ground data. A comparison of the new approach with the traditional regression-based and ground-data dependent model training is presented in this paper for a multi-seasonal ERS-1/2 tandem dataset covering several well known Central Siberian forest sites. As a test scenario for large-area application, the approach was applied to a multi-seasonal ERS-1/2 tandem dataset of 223 ERS-1 and ERS-2 image pairs covering Northeast China (~ 1.5 million km 2) to map four stem volume classes (0–20, 20–50, 50–80, and > 80 m 3/ha).