Does environmental data increase the accuracy of land use and land cover classification?

Optical image classification converts spectral data into thematic information from the spectral signature of each object in the image. However, spectral separability is influenced by intrinsic characteristics of the targets, as well as the characteristics of the images used. The classification process will present more reliable results when aspects associated with natural environments (climate, soil, relief, water, etc.) and anthropic environments (roads, constructions, urban area) begin to be considered, as they determine and guide land use and land cover (LULC). The objectives of this study are to evaluate the integration of environmental variables with spectral variables and the performance of the Random Forest algorithm in the classification of Landsat-8 OLI images, of a watershed in the Eastern Amazon, Brazil. The classification process used 96 predictive variables, involving spectral, geological, pedological, climatic and topographic data and Euclidean distances. The selection of variables to construct the predictive models was divided into two approaches: (i) data set containing only spectral variables, and (ii) set of environmental variables added to the spectral data. The variables were selected through nonlinear correlation analysis, with the Randomized Dependence Coefficient and the Recursive Feature Elimination (RFE) method, using the Random Forest classifier algorithm. The spectral variables NDVI, bands 2, 4, 5, 6 and 7 of the dry season and band 4 of the rainy season were selected in both approaches (i and ii). The Euclidean distance from the urban area, Arenosol soil class, annual precipitation, precipitation in February and precipitation of the wettest quarter were the variables selected from the auxiliary data set. This study showed that the addition of environmental data to the spectral data reduces the limitation of the latter, regarding the discrimination of the different classes of LULC, in addition to improving the accuracy of the classification. The addition of soil classes to spectral variables provided a reduction in errors for vegetation classification (Evergreen Forest and Cerrado Sensu Stricto), as it was able to inform about nutrient availability and water storage capacity. The study demonstrates that the addition of environmental variables to the spectral variables can be an alternative to improve monitoring in areas of ecotone in Neotropical regions.     

Mapping diversified peri-urban agriculture – potential of object-based versus per-field land cover/land use classification

High spatial resolution satellite data contribute to improving land cover/land use (LCLU) classification in agriculture. A classification procedure based on Quickbird satellite image data was developed to map LCLU of diversified agriculture at sub-communal and communal level (7 km²). Segmentation performance of the panchromatic band in combination with high pass filters (HPF) was tested first. Accuracy of field boundary delineation was evaluated by an object-based segmentation, a per-field and a manual classification, along with a quantitative accuracy assessment. Sub-communal classification revealed an overall accuracy of 84% with a κ coefficient of 0.77 for the per-field vector segmentation compared to an overall accuracy of 56–60% and a κ coefficient of 0.37–0.42 for object-based approaches. Per-field vector segmentation was thus superior and used for LCLU classification at communal level. Overall accuracy scored 83% and the κ coefficient 0.7. In diversified agriculture, per-field vector segmentation and classification achieved higher classification results.     

Quantifying land use/land cover spatio-temporal landscape pattern dynamics from Hyperion using SVMs classifier and FRAGSTATS®

This study aims to quantify the landscape spatio-temporal dynamics including Land Use/Land Cover (LULC) changes occurred in a typical Mediterranean ecosystem of high ecological and cultural significance in central Greece covering a period of 9 years (2001–2009). Herein, we examined the synergistic operation among Hyperion hyperspectral satellite imagery with Support Vector Machines, the FRAGSTATS^® landscape spatial analysis programme and Principal Component Analysis (PCA) for this purpose. The change analysis showed that notable changes reported in the experimental region during the studied period, particularly for certain LULC classes. The analysis of accuracy indices suggested that all the three classification techniques are performing satisfactorily with overall accuracy of 86.62, 91.67 and 89.26% in years 2001, 2004 and 2009, respectively. Results evidenced the requirement for taking measures to conserve this forest-dominated natural ecosystem from human-induced pressures and/or natural hazards occurred in the area. To our knowledge, this is the first study of its kind, demonstrating the Hyperion capability in quantifying LULC changes with landscape metrics using FRAGSTATS^® programme and PCA for understanding the land surface fragmentation characteristics and their changes. The suggested approach is robust and flexible enough to be expanded further to other regions. Findings of this research can be of special importance in the context of the launch of spaceborne hyperspectral sensors that are already planned to be placed in orbit as the NASA’s HyspIRI sensor and EnMAP.     

Production of global land cover data – GLCNMO

Abstract

Global land cover is one of the fundamental contents of Digital Earth. The Global Mapping project coordinated by the International Steering Committee for Global Mapping has produced a 1-km global land cover dataset – Global Land Cover by National Mapping Organizations. It has 20 land cover classes defined using the Land Cover Classification System. Of them, 14 classes were derived using supervised classification. The remaining six were classified independently: urban, tree open, mangrove, wetland, snow/ice, and water. Primary source data of this land cover mapping were eight periods of 16-day composite 7-band 1-km MODIS data of 2003. Training data for supervised classification were collected using Landsat images, MODIS NDVI seasonal change patterns, Google Earth, Virtual Earth, existing regional maps, and expert's comments. The overall accuracy is 76.5% and the overall accuracy with the weight of the mapped area coverage is 81.2%. The data are available from the Global Mapping project website (http://www.iscgm.org/). The MODIS data used, land cover training data, and a list of existing regional maps are also available from the CEReS website. This mapping attempt demonstrates that training/validation data accumulation from different mapping projects must be promoted to support future global land cover mapping.     

Tropical land use land cover mapping in Pará (Brazil) using discriminative Markov random fields and multi-temporal TerraSAR-X data

Remote sensing satellite data offer the unique possibility to map land use land cover transformations by providing spatially explicit information. However, detection of short-term processes and land use patterns of high spatial–temporal variability is a challenging task.We present a novel framework using multi-temporal TerraSAR-X data and machine learning techniques, namely discriminative Markov random fields with spatio-temporal priors, and import vector machines, in order to advance the mapping of land cover characterized by short-term changes. Our study region covers a current deforestation frontier in the Brazilian state Pará with land cover dominated by primary forests, different types of pasture land and secondary vegetation, and land use dominated by short-term processes such as slash-and-burn activities. The data set comprises multi-temporal TerraSAR-X imagery acquired over the course of the 2014 dry season, as well as optical data (RapidEye, Landsat) for reference. Results show that land use land cover is reliably mapped, resulting in spatially adjusted overall accuracies of up to 79% in a five class setting, yet limitations for the differentiation of different pasture types remain.The proposed method is applicable on multi-temporal data sets, and constitutes a feasible approach to map land use land cover in regions that are affected by high-frequent temporal changes.     

Monitoring and analysing the Emirate of Dubai’s land use/land cover changes: an integrated,low-cost remote sensing approach

This study presents a modified low-cost approach, which integrates the spectral angle mapper and image difference algorithms in order to enhance classification maps for the purpose of monitoring and analysing land use/land cover change between 2000 and 2015 for the Emirate of Dubai. The approach was modified by collecting 320 training samples from QuickBird images with a spatial resolution of 0.6 m, as well as carrying out field observations, followed by the application of a 3?×?3 Soble filter, sieving classes, majority/minority analysis, and clump classes of the obtained classification maps. The accuracy assessment showed that the targeted 2000, 2005, 2010 and 2015 classification maps have 88.1252%, 89.0699%, 90.1225% and 96.0965% accuracy, respectively. The results showed that the built-up area increased by 233.721?km² (5.81%) between 2000 and 2005 and continues to increase even up and till the present time. The assessment of changes in the periods 2000–2005 and 2010–2015 confirmed that net vegetation area losses were more pronounced from 2000 to 2005 than from 2010 to 2015, dropping from 47,618 to 40,820?km², respectively. This study is aimed to assist urban planners and decision-makers, as well as research institutes.     

The analysis of SPOT-5 characteristics on land cover Classification

<正>KnowledgeaboutlandcoverandlandusehasbecomeincreasinglyimportantastheNationplanstoovercometheproblemsofuncontrolleddevelopment,deterioratingenvironmentalquality,lossofprimeagriculturallandsetc.Landuseandlandcoverdataareneededintheanalysisofenvironmentalprocessesandproblemstoknowiflivingconditionsandstandardsaretobeimprovedormaintainedatcurrentlevels.     

Satellite detection of increases in global land surface evapotranspiration during 1984–2007

Abstract

As a key component of digital earth, remotely sensed data provides the compelling evidence that the amount of water vapour transferred from the entire global surface to the atmosphere increased from 1984 to 2007. The validation results from the earlier evapotranspiration (ET) estimation algorithm based on net radiation (R _n ), Normalised Difference Vegetation Index (NDVI), air temperature and diurnal air temperature range (DTaR) showed good agreement between estimated monthly ET and ground-measured ET from 20 flux towers. Our analysis indicates that the estimated actual ET has increased on average over the entire global land surface except for Antarctica during 1984–2007. However, this increasing trend disappears after 2000 and the reason may be that the decline in net radiation and NDVI during this period depleted surface soil moisture. Moreover, the good correspondence between the precipitation trend and the change in ET in arid and semi-arid regions indicated that surface moisture linked to precipitation affects ET. The input parameters R _n , T _air, NDVI and DTaR show substantial spatio-temporal variability that is almost consistent with that of actual ET from 1984 to 2007 and contribute most significantly to the variation in actual ET.     

An application of spatial filtering techniques for land cover mapping using TM‐images

Land cover classification using satellite imagery is commonly based on spectral information in the individual pixels. The information in neighbouring pixels is ignored. Spatial filtering techniques using information present in neighbouring pixels may however, contribute significantly to an improvement of the classification. In this study different methods of spatial filtering are applied to a part of a TM‐scene of Kenya to assess their relative reliability. The study area is characterized by extended, relatively homogeneous areas of eucalyptus forests and tea estates and by fragmentated areas of agricultural land use. Spectral information was combined with the results of different spatial filtering methods and then classified. The spatial filtering techniques applied were texture calculation by means of variance, “median minus original” filtering and fractal dimension computations using several sizes of templates. The obtained classification accuracy of several image combinations is compared using the percentage correctly classified and using an overall accuracy measure: the Kappa coefficient. It is concluded that in this case the spatial filtering techniques only slightly improve the classification. From the applied filtering methods texture calculation by means of variance yielded the best results.     

Retrieving spatial variations of land surface temperatures from satellite data–Cairo region,Egypt

Cairo region is characterized by a range of physiographic features, including: flat agricultural lands, bare sandy deserts, highlands, calcareous terrains and urban land use. A time series data-set (300 images) acquired from the Moderate Resolution Imaging Spectroradiometer for the period July 2002–June 2015 were utilized to retrieve the spatial variations in the mean land surface temperature (LST) for the above-mentioned surface features. Results showed that vegetation, topography and surface albedo have negative correlations with LST. Vegetation/LST correlation has the maximum regression coefficient (R² = 0.68) and albedo/LST has the minimum (R² = 0.03). Cultivated lands reveal the lowest mean LST (<32 °C), whereas industrial lands exhibit the highest LST (>40 °C) of Cairo region. There is a considerable urban heat island formed at Helwan south of Cairo, where heavy industries are settled. Industrial activities raised the mean LST of the region by at least 4 °C than the surrounding urban lands.     

Towards automated 10–30 m resolution land cover mapping in insular South-East Asia

In this study, we create and critically analyse an automated decision tree classification approach for regional level land cover mapping in insular South-East Asian conditions, using a combination of 10–30 m resolution optical and radar data. The resulting map contains 11 land cover classes and reveals a great deal of contextual information due to high spatial resolution. A limited accuracy assessment indicates 59–97% class wise accuracies. The unprecedented spatial detail of closed canopy oil palm mapping (with user’s accuracy of 90%) is seen as the most promising feature of the mapping approach. The incapability of separating primary forests from other tree cover, and the large variety of different landscapes (e.g. home gardens and tea plantations) classified as shrubland, are considered the main areas for future improvement. Overall, the study demonstrates the great potential of multi-source 10–30 m resolution high data volume land cover mapping approaches in insular South-East Asian conditions.     

Woody vegetation and land cover changes in the Sahel of Mali (1967–2011)

In the past 50 years, the Sahel has experienced significant tree- and land cover changes accelerated by human expansion and prolonged droughts during the 1970s and 1980s. This study uses remote sensing techniques, supplemented by ground-truth data to compare pre-drought woody vegetation and land cover with the situation in 2011. High resolution panchromatic Corona imagery of 1967 and multi-spectral RapidEye imagery of 2011 form the basis of this regional scaled study, which is focused on the Dogon Plateau and the Seno Plain in the Sahel zone of Mali. Object-based feature extraction and classifications are used to analyze the datasets and map land cover and woody vegetation changes over 44 years. Interviews add information about changes in species compositions. Results show a significant increase of cultivated land, a reduction of dense natural vegetation as well as an increase of trees on farmer's fields. Mean woody cover decreased in the plains (−4%) but is stable on the plateau (+1%) although stark spatial discrepancies exist. Species decline and encroachment of degraded land are observed. However, the direction of change is not always negative and a variety of spatial variations are shown. Although the impact of climate is obvious, we demonstrate that anthropogenic activities have been the main drivers of change.     

Abrupt spatiotemporal land and water changes and their potential drivers in Poyang Lake, 2000–2012

Driven by various natural and anthropogenic factors, Poyang Lake, the largest freshwater lake in China, has experienced significant land use/cover changes in the past few decades. The aim of this study is to investigate the spatial–temporal patterns of abrupt changes and detect their potential drivers in Poyang Lake, using time-series Moderate Resolution Imaging Spectroradiometer (MODIS) 16-day maximum value composite vegetation indices between 2000 and 2012. The breaks for additive seasonal and trend (BFAST) method was applied to the smoothed time-series normalized difference vegetation index (NDVI), to detect the timing and magnitude of abrupt changes in the trend component. Large part of Poyang Lake (98.9% for trend component) has experienced abrupt changes in the past 13 years, and the change patterns, including the distributions in timing and magnitudes of major abrupt trend changes between water bodies and land areas were clearly differentiated. Most water bodies had abrupt increasing NDVI changes between 2010 and 2011, caused by the sequential severe flooding and drought in the two years. In contrast, large parts of the surrounding land areas had abrupt decreasing NDVI changes. Large decreasing changes occurred around 2003 at the city of Nanchang, which were driven by urbanization. These results revealed spatial–temporal land cover changing patterns and potential drivers in the wetland ecosystem of Poyang Lake.     

The effect of imposing ‘fractional abundance constraints’ onto the multilayer perceptron for sub-pixel land cover classification

To be physically interpretable, sub-pixel land cover fractions or abundances should fulfill two constraints, the Abundance Non-negativity Constraint (ANC) and the Abundance Sum-to-one Constraint (ASC). This paper focuses on the effect of imposing these constraints onto the MultiLayer Perceptron (MLP) for a multi-class sub-pixel land cover classification of a time series of low resolution MODIS-images covering the northern part of Belgium. Two constraining modes were compared, (i) an in-training approach that uses ‘softmax’ as the transfer function in the MLP’s output layer and (ii) a post-training approach that linearly rescales the outputs of the unconstrained MLP. Our results demonstrate that the pixel-level prediction accuracy is markedly increased by the explicit enforcement, both in-training and post-training, of the ANC and the ASC. For aggregations of pixels (municipalities), the constrained perceptrons perform at least as well as their unconstrained counterparts. Although the difference in performance between the in-training and post-training approach is small, we recommend the former for integrating the fractional abundance constraints into MLPs meant for sub-pixel land cover estimation, regardless of the targeted level of spatial aggregation.     

Remote sensing monitoring of land restoration interventions in semi-arid environments with a before–after control-impact statistical design

Restoration interventions to combat land degradation are carried out in arid and semi-arid areas to improve vegetation cover and land productivity. Evaluating the success of an intervention over time is challenging due to various constraints (e.g. difficult-to-access areas, lack of long-term records) and the lack of standardised and affordable methodologies. We propose a semi-automatic methodology that uses remote sensing data to provide a rapid, standardised and objective assessment of the biophysical impact, in terms of vegetation cover, of restoration interventions. The Normalised Difference Vegetation Index (NDVI) is used as a proxy for vegetation cover. Recognising that changes in vegetation cover are naturally due to environmental factors such as seasonality and inter-annual climate variability, conclusions about the success of the intervention cannot be drawn by focussing on the intervention area only. We therefore use a comparative method that analyses the temporal variations (before and after the intervention) of the NDVI of the intervention area with respect to multiple control sites that are automatically and randomly selected from a set of candidates that are similar to the intervention area. Similarity is defined in terms of class composition as derived from an ISODATA classification of the imagery before the intervention. The method provides an estimate of the magnitude and significance of the difference in greenness change between the intervention area and control areas. As a case study, the methodology is applied to 15 restoration interventions carried out in Senegal. The impact of the interventions is analysed using 250-m MODIS and 30-m Landsat data. Results show that a significant improvement in vegetation cover was detectable only in one third of the analysed interventions, which is consistent with independent qualitative assessments based on field observations and visual analysis of high resolution imagery. Rural development agencies may potentially use the proposed method for a first screening of restoration interventions.     

Object-oriented semantic labelling of spectral–spatial LiDAR point cloud for urban land cover classification and buildings detection

The urban land cover mapping and automated extraction of building boundaries is a crucial step in generating three-dimensional city models. This study proposes an object-based point cloud labelling technique to semantically label light detection and ranging (LiDAR) data captured over an urban scene. Spectral data from multispectral images are also used to complement the geometrical information from LiDAR data. Initial object primitives are created using a modified colour-based region growing technique. Multiple classifier system is then applied on the features extracted from the segments for classification and also for reducing the subjectivity involved in the selection of classifier and improving the precision of the results. The proposed methodology produces two outputs: (i) urban land cover classes and (ii) buildings masks which are further reconstructed and vectorized into three-dimensional buildings footprints. Experiments carried out on three airborne LiDAR datasets show that the proposed technique successfully discriminates urban land covers and detect urban buildings.     

A spatial–temporal Hopfield neural network approach for super-resolution land cover mapping with multi-temporal different resolution remotely sensed images

The mixed pixel problem affects the extraction of land cover information from remotely sensed images. Super-resolution mapping (SRM) can produce land cover maps with a finer spatial resolution than the remotely sensed images, and reduce the mixed pixel problem to some extent. Traditional SRMs solely adopt a single coarse-resolution image as input. Uncertainty always exists in resultant fine-resolution land cover maps, due to the lack of information about detailed land cover spatial patterns. The development of remote sensing technology has enabled the storage of a great amount of fine spatial resolution remotely sensed images. These data can provide fine-resolution land cover spatial information and are promising in reducing the SRM uncertainty. This paper presents a spatial–temporal Hopfield neural network (STHNN) based SRM, by employing both a current coarse-resolution image and a previous fine-resolution land cover map as input. STHNN considers the spatial information, as well as the temporal information of sub-pixel pairs by distinguishing the unchanged, decreased and increased land cover fractions in each coarse-resolution pixel, and uses different rules in labeling these sub-pixels. The proposed STHNN method was tested using synthetic images with different class fraction errors and real Landsat images, by comparing with pixel-based classification method and several popular SRM methods including pixel-swapping algorithm, Hopfield neural network based method and sub-pixel land cover change mapping method. Results show that STHNN outperforms pixel-based classification method, pixel-swapping algorithm and Hopfield neural network based model in most cases. The weight parameters of different STHNN spatial constraints, temporal constraints and fraction constraint have important functions in the STHNN performance. The heterogeneity degree of the previous map and the fraction images errors affect the STHNN accuracy, and can be served as guidances of selecting the optimal STHNN weight parameters.     

Analysis and simulation of land use spatial pattern in Harbin prefecture based on trajectories and cellular automata—Markov modelling

There have been rapid population and accelerating urban growth with associated changes in land use and soil degradation in northeast China, an important grain-producing region. The development of integrated use of remote sensing, geographic information systems, and combined cellular automata– Markov models has provided new means of assessing changes in land use and land cover, and has enabled projection of trajectories into the future. We applied such techniques to the prefecture-level city of Harbin, the tenth largest city in China. We found that there had been significant losses of the land uses termed “cropland”, “grassland”, “wetland”, and “floodplain” in favour of “built-up land” and lesser transformations from “floodplain” to “forestland” and “water body” over the 18-year period. However, the transition was not a simple process but a complex network of changes, interchanges, and multiple transitions. In the absence of effective land use policies, projection of past trajectories into a balance state in the future would result in the decline of cropland from 65.6% to 46.9% and the increase of built-up area from 7.7% to 23.0% relative to the total area of the prefecture in 1989. It also led to the virtual elimination of land use types such as unused wetland and floodplain.     

How much variation in land surface phenology can climate oscillation modes explain at the scale of mountain pastures in Kyrgyzstan?

Climate oscillation modes can shape weather across the globe due to atmospheric teleconnections. We built on the findings of a recent study to assess whether the impacts of teleconnections are detectable and significant in the early season dynamics of highland pastures across five rayons in Kyrgyzstan. Specifically, since land surface phenology (LSP) has already shown to be influenced by snow cover seasonality and terrain, we investigated here how much more explanatory and predictive power information about climatic oscillation modes might add to explain variation in LSP. We focused on seasonal values of five climate oscillation indices that influence vegetation dynamics in Central Asia. We characterized the phenology in highland pastures with metrics derived from LSP modeling using Landsat NDVI time series together with MODIS land surface temperature (LST) data: Peak Height (PH), the maximum modeled NDVI and Thermal Time to Peak (TTP), the quantity of accumulated growing degree-days based on LST required to reach PH. Next, we calculated two metrics of snow cover seasonality from MODIS snow cover composites: last date of snow (LDoS), and the number of snow covered dates (SCD). For terrain features, we derived elevation, slope, and TRASP index as linearization of aspect. First, we used Spearman’s rank correlation to assess the geographical differentiation of land surface phenology metrics responses to environmental variables. PH showed weak correlations with TTP (positive in western but negative in eastern rayons), and moderate relationships with LDoS and SCD only in one northeastern rayon. Slope was weakly related to PH, while TRASP showed a consistent moderate negative correlation with PH. A significant but weak negative correlation was found between PH and SCAND JJA, and a significant weak positive correlation with MEI MAM. TTP showed consistently strong negative relationships with LDoS, SCD, and elevation. Very weak positive correlations with TTP were found for EAWR DJF, AMO DJF, and MEI DJF in western rayons only. Second, we used Partial Least Squares regression to investigate the role of oscillation modes altogether. PLS modelling of TTP showed that thermal time accumulation could be explained mostly by elevation and snow cover metrics, leading to reduced models explaining 55 to 70% of observed variation in TTP. Variable selection indicated that NAO JJA, AMO JJA and SCAND MAM had significant relationships with TTP, but their input of predictive power was neglible. PLS models were able to explain up to 29% of variability in PH. SCAND JJA and MEI MAM were shown to be significant predictors, but adding them into models did not influence modeling performance. We concluded the impacts of climate oscillation anomalies were not detectable or significant in mountain pastures using LSP metrics at fine spatial resolution. Rather, at a 30 m resolution, the indirect effects of seasonal climatic oscillations are overridden by terrain influences (mostly elevation) and snow cover timing. Whether climate oscillation mode indices can provide some new and useful information about growing season conditions remains a provocative question, particularly in light of the multiple environmental challenges facing the agropastoralism livelihood in montane Central Asia.     

The nature and possible causes of land cover change (1984 –1996) along a rainfall gradient in southeastern Botswana

Land cover change analysis was undertaken in semi-arid southeastern Botswana. The aim was to determine how remotely sensed data could be applied over time and under different rainfall regimes to help assess the relative significance of biophysical and human factors in causing land cover change in a rapidly evolving developing world context. To this purpose, land cover changes were studied along an east (hardveld)-west (sandveld) gradient of decreasing rainfall and decreasing population density. Three years of Thematic Mapper imagery from 1984, 1994 and 1996, covering the period from the 1980s drought to the 1990s ‘normal’ rainfall regime were analysed using supervised classification techniques. Land cover change analysis revealed that over a large part of the study area the dry and more biophysically vulnerable western sandveld showed greater vegetation recovery than the eastern hardveld with its more productive soils and higher rainfall. Underlying causes behind this apparent reversal of trends are inferred to be mainly socioeconomic in nature and particularly related to higher population density due to the rise of salaried urban occupation opportunities in the hardveld. This work concludes that, while biophysical causes of change are important, the human dimension is regarded as being more significant especially where human factors negate otherwise positive biophysical effects in an agrarian developing country.