Reconstructing long time series of burned areas in arid grasslands of southern Russia by satellite remote sensing

Fire is an important natural disturbance process in many ecosystems, but humans can irrevocably change natural fire regimes. Quantifying long-term change in fire regimes is important to understand the driving forces of changes in fire dynamics, and the implications of fire regime changes for ecosystem ecology. However, assessing fire regime changes is challenging, especially in grasslands because of high intra- and inter-annual variation of the vegetation and temporally sparse satellite data in many regions of the world. The breakdown of the Soviet Union in 1991 caused substantial socioeconomic changes and a decrease in grazing pressure in Russia's arid grasslands, but how this affected grassland fires is unknown. Our research goal was to assess annual burned area in the grasslands of southern Russia before and after the breakdown. Our study area covers 19,000 km² in the Republic of Kalmykia in southern Russia in the arid grasslands of the Caspian plains. We estimated annual burned area from 1985 to 2007 by classifying AVHRR data using decision tree algorithm, and validated the results with RESURS, Landsat and MODIS data. Our results showed a substantial increase in burned area, from almost none in the 1980s to more than 20% of the total study area burned in both 2006 and 2007. Burned area started to increase around 1998 and has continued to increase, albeit with high fluctuations among years. We suggest that it took several years after livestock numbers decreased in the beginning of the 1990s for vegetation to recover, to build up enough fuel, and to reach a threshold of connectivity that could sustain large fires. Our burned area detection algorithm was effective, and captured burned areas even with incomplete annual AVHRR data. Validation results showed 68% producer's and 56% user's accuracy. Lack of frequent AVHRR data is a common problem and our burned area detection approach may also be suitable in other parts of the world with comparable ecosystems and similar AVHRR data limitations. In our case, AVHRR data were the only satellite imagery available far enough back in time to reveal marked increases in fire regimes in southern Russia before and after the breakdown of the Soviet Union.     

Monitoring spatial patterns and changes of surface net radiation in urban and suburban areas using satellite remote-sensing data

The net radiation obtained by the earth’s surface drives the exchange and transmission of energy and material in earth systems. Accurately quantifying the surface net radiation (SNR) is a premise for climate change research. Satellite remote-sensing data can provide land-surface information on a regional scale, making it possible to monitor SNR spatial patterns and changes using a cost-effective methodology that is superior to traditional ground observation. Furthermore, it can address the problem that the local-scale observational data does not extend to large areas with sparsely distributed meteorological observation sites. In this article, urban and suburban areas of Beijing were selected as study areas, and the surface physical parameters were retrieved based on Landsat images. Auxiliary data were also applied to model SNR, including Moderate Resolution Imaging Spectroradiometer (MODIS) images and other land-surface observational data obtained from meteorological stations. In addition, the SNR spatial patterns and their changes in urban and suburban areas and the differences in different land-surface types, different years, and different seasons were analysed. The results showed the following: (1) the fractional vegetation cover was one of the principal factors affecting the surface radiation process, and the SNR value was high where the cover value was high. Comparing the SNR in urban areas with that in suburban areas, the value was higher, and there was generally a ‘plateau’ in the spatial distribution characteristics in urban areas. (2) After analysis of the mean SNR value for different land-surface types, the highest mean SNR was for water, followed by vegetation cover, artificial surface and bare land, and the deviation of the mean SNR values of all of the land-surface types in winter was smaller compared with those in summer. (3) With urban sprawl and rapid changes of land-surface cover, there was an increasing trend in the SNR value in urban areas that was more significant in summer than that in winter. According to the SNR values in 2004 and 2014, the areas of all of the land-surface types showed a small increase of approximately 35 W m^–2 in summer and 25 W m^–2 in winter.     

Comparison of burned area estimates derived from SPOT-VEGETATION and Landsat ETM+ data in Africa: Influence of spatial pattern and vegetation type

An algorithm for burned area mapping in Africa based on classification trees was developed using SPOT-VEGETATION (VGT) imagery. The derived 1 km spatial resolution burned area maps were compared with 30 m spatial resolution maps obtained with 13 Landsat ETM+ scenes, through linear regression analysis. The procedure quantifies the bias in burned area estimation present in the low spatial resolution burned area map. Good correspondence was observed for seven sites, with values of the coefficient of determination (R²) ranging from 0.787 to 0.983. Poorer agreement was observed in four sites (R² values between 0.257 and 0.417), and intermediate values of R² (0.670 and 0.613) were obtained for two sites. The observed variation in the level of agreement between the Landsat and VGT estimates of area burned results from differences in the spatial pattern and size distribution of burns in the different fire regimes encompassed by our analysis. Small and fragmented burned areas result in large underestimation at 1 km spatial resolution. When large and compact burned areas dominate the landscape, VGT estimates of burned area are accurate, although in certain situations there is some overestimation. Accuracy of VGT burned area estimates also depends on vegetation type. Results showed that in forest ecosystems VGT maps underestimate substantially the amount of burned area. The most accurate estimates were obtained for woodlands and grasslands. An overall linear regression fitted with the data from the 13 comparison sites revealed that there is a strong relationship between VGT and Landsat estimates of burned area, with a value of R² of 0.754 and a slope of 0.803. Our findings indicate that burned area mapping based on 1 km spatial resolution VGT data provides adequate regional information.     

Growing period and drought early warning in Africa using satellite data

Vegetation growing periods for 1983-84 were determined for 28 sites in Ethiopia using data from the Advanced Very High Resolution Radiometer (AVHRR) on the NOAA series of meteorological satellites. Results offer promise for drought early warning from space. A strong correlation (r=0·99) was found between rates of change of the normalized difference vegetation index (NDVI) derived from the AVHRR data and threshold values of a soil moisture index at the beginning and ends of growing periods. The moisture index (P + S)/ET_p relates precipitation P, stored soil moisture S and potential evapotranspiration ET_p in a simple moisture balance (LPG) model that requires inputs of standard monthly meteorological data. A moisture threshold of (P + S)/ET_p = 0·5 was used to identify the beginning and end of the growing periods and to calibrate the time series of NDVI responses. Trends also detected in values of the NDVI during vegetation growth cycles suggest useful minima exist at the beginning and end of growing periods. Below respective minima of 0·10 and 0·22, growing periods are unlikely to have been initiated or to continue during a declining growth stage. Correlation analysis indicated a relation between moisture index and NDVI, with NDVI lagging in time, in most cases, by 5 or less weeks during the initial growth stage and 6 or more weeks during declining growth     

A comparative study on spatial and spectral resolutions of satellite data in mapping mangrove forests

Mangrove forests in the western Waitemata Harbour, Auckland, New Zealand were mapped into lush and stunted categories from SPOT HRV and Landsat TM images at 10, 20 and 30m using the maximum likelihood method. It was found that the TM-generated results were the most accurate at 95% for lush mangroves and 87.5% for stunted mangroves. Their corresponding accuracy levels were lowered to 77.5% and 67.5% in the 20m SPOT XS-derived results. Both percentages were improved to 80 after the PAN band was incorporated in the classification at 10m. These results suggest that a high spectral resolution is more important in accurately mapping mangroves in a temperate zone than a fine spatial resolution because it enhances the interpretability of non-mangrove vegetation and thus increases its confusion with mangroves.     

Simultaneous detection of burned areas of multiple fires in the tropics using multisensor remote-sensing data

Fires associated with recurrent El Niño events have caused severe damage to tropical peat swamp forests. Accurate quantitative information about the frequency and distribution of the burned areas is imperative to fire management but is lacking in the tropics. This article examines a novel method based on principal component analysis (PCA) of the normalized difference water index (NDWI) from multisensor data for simultaneously detecting areas burned due to multiple El Niño–related fires. The principal components of multitemporal NDWI (NDWI-PCs) were able to capture the areas burned in the 1998 and 2003 El Niño fires in NDWI-PC3 and 2, respectively. The proposed method facilitates the reduction of dimensionality in detecting the burned areas. From 22 image bands, the proposed method was able to accurately detect the burned areas of multiple fires with only three NDWI-PCs. The proposed method also shows superior performance to unsupervised classifications of the principal components of combined image bands, multitemporal NDWI, NDWI differencing and post-classification comparison methods. The results show that the 1998 El Niño fire was devastating especially to intact peat swamp forest. For degraded peat swamp forest, there was an increase in the burned area from 1998 to 2003. The proposed method offers the retrieval of accurate and reliable quantitative information on the frequency and spatial distribution of burned areas of multiple fires in the tropics. This method is also applicable to the detection of changes in general as well as the detection of vegetation changes.     

Detecting shadows in high-resolution remote-sensing images of urban areas using spectral and spatial features

In accordance with the characteristics of urban high-resolution (HR) remote-sensing images, we propose a shadow detection algorithm that combines spectral and spatial features. Rather than pixel-based shadow features, the proposed features are based on shadow regions obtained by the object-based segmentation method. First, based on the shadow ratio map, the candidate shadow pixels are acquired by the Otsu method. The candidate shadow regions can be identified using connected component analysis. In the candidate shadow regions, shadow spectral and spatial features are calculated. With these two features, the true shadow regions can be distinguished from candidate shadow regions. Experiments and comparisons indicate that our proposed algorithm is feasible and effective for shadow detection in both aerial and satellite images.     

Assessing the landscape context and conversion risk of protected areas using satellite data products

Since the establishment of the first national park (Yellowstone National Park in 1872) and the first wildlife refuge (Pelican Island in 1903), dramatic changes have occurred in both ecological and cultural landscapes across the U.S. The ability of these protected areas to maintain current levels of biodiversity depend, at least in part, on the integrity of the surrounding landscape. Our objective was to quantify and compare the extent and pattern of natural land cover, risk of conversion, and relationships with demographic and economic variables in counties near National Park Service units and U.S. Fish and Wildlife Service refuges with those counties distant from either type of protected area in the coterminous United States. Our results indicate that landscapes in counties within 10 km of both parks and refuges and those within 10 km of just parks were more natural, more intact, and more protected than those in counties within 10 km of just refuges and counties greater than 10 km from either protected area system. However, they also had greater human population density and change in population, indicating potential conversion risk since the percent of landscape protected averaged < 5% in both groups and human population dynamics are primary drivers of change in many landscapes. Conversion outweighed protection by at least two times (Conservation Risk Index > 2) in 76% of counties near both parks and refuges, 81% of counties near just parks, 91% of counties near just refuges, and 93% of distant counties. Thirteen percent of counties in the coterminous U.S. had moderate to high amounts of natural land cover (> 60%), low protection (< 20%), and the greatest change in population (> 20%). Although these areas are not the most critically endangered, they represent the greatest conservation opportunity, need, and urgency. Our approach is based on national level metrics that are simple, general, informative, and can be understood by broad audiences and by policy makers and managers to assess the health of lands surrounding parks and refuges. Regular monitoring of these metrics with satellite data products in counties surrounding protected areas provides a consistent, national level assessment of management opportunities and potentially adverse changes on adjacent lands.     

Monitoring post-burn recovery of chaparral vegetation in southern California using multi-temporal satellite data

Monitoring the post-burn recovery condition of chaparral vegetation in southern California is important for managers to determine the appropriate time to conduct controlled burns. Due to the difficulty of monitoring post-fire recovery over large areas and the absence of detailed fire records in many areas, we examined the possibility of using satellite observations to establish the postfire recovery stage of chamise chaparral stands in this region. SPOT XS data collected on three dates between 1986 and 1992 were analysed to determine if temporal changes in a spectral vegetation index tracked the expected post-fire recovery trajectory of the above-ground biomass of chamise chaparral stands of varying post-fire ages. Results of the study indicated that neither the normalized difference vegetation index nor the soil adjusted vegetation index followed the expected post-fire recovery patterns in these stands. These findings are explained by interannual variations in precipitation having a larger than expected effect on the growth of this drought-resistant evergreen community, with changes in green leaf area dominating the temporal variations in the spectral vegetation indices.     

A typical pattern of vegetation conditions in southern Africa during El Nino years detected from AVHRR data using three-channel numerical index

The 1997-98 El Nino is one of the strongest events of the last 50 years. Its absolute magnitude, areal extent, and rapid development have raised a serious concern among decision makers because of its possible impact on global ecosystems. In this study, El Nino consequences for land ecosystem were examined using the new AVHRR-based three-channel index (VT), widely used for monitoring drought around the world. The VT index was used to identify a typical pattern of vegetation conditions in southern Africa during the recent El Nino years. Features and trends of the 1997-98 El Nino including its intensity, extent, and impact on vegetation are also discussed.     

Estimating spatial variations in soil organic carbon using satellite hyperspectral data and map algebra

This study evaluated the effectiveness of using Hyperion hyperspectral data in improving existing remote-sensing methodologies for estimating soil organic carbon (SOC) content on farmland. The study area is Big Creek Watershed in Southern Illinois, USA. Several data-mining techniques were tested to calibrate and validate models that could be used for predicting SOC content using Hyperion bands as predictors. A combined model of stepwise regression followed by a five hidden nodes artificial neural network was selected as the best model, with a calibration coefficient of determination (R ²) of 78.9% and a root mean square error (RMSE) of 3.3 tonnes per hectare (t ha^?1). The validation RMSE, however, was found to be 11.3 t ha^?1. Map algebra was implemented to extrapolate this model and produce a SOC map for the watershed. Hyperspectral data improved marginally the predictability of SOC compared to multispectral data under natural field conditions. They could not capture small annual variations in SOC, but could measure decadal variations with moderate error. Satellite-based hyperspectral data combined with map algebra can measure total SOC pools in various ecosystem or soil types to within a few per cent error.     

Dynamics of the relationship between NDVI and SWIR32 vegetation indices in southern Africa: implications for retrieval of fractional cover from MODIS data

Fractional cover of photosynthetic vegetation (F_PV), non-photosynthetic vegetation (F_NPV), and bare soil (F_BS) has been retrieved for Australian tropical savannah based on linear unmixing of the two-dimensional response envelope of the normalized difference vegetation index (NDVI) and short wave infrared ratio (SWIR)32 vegetation indices (VI) derived from Moderate Resolution Imaging Spectroradiometer (MODIS) reflectance data. The approach assumes that cover fractions are made up of a simple mixture of green leaves, senescent leaves, and bare soil. In this study, we examine retrieval of fractional cover using this approach for a study area in southern Africa with a more complex vegetation structure. Region-specific end-members were defined using Hyperion images from different locations and times of the season. These end-members were applied to a 10-year time series of MODIS-derived NDVI and SWIR32 (from 2002 to 2011) to unmix F_PV, F_NPV, and F_BS. Results of validation with classified high-resolution imagery indicated major bias in estimation of F_NPV and F_BS, with regression coefficients for predicted versus observed data substantially less than 1.0 and relatively large intercept values. Examination with Hyperion images of the inverse relationship between the MODIS-equivalent SWIR32 index and the Hyperion-derived cellulose absorption index (CAI) to which it nominally approximates revealed: (1) non-compliant positive regression coefficients for certain vegetation types; and (2) shifts in slope and intercept of compliant regression curves related to day of year and geographical location. The results suggest that the NDVI–SWIR32 response cannot be used to approximate the NDVI–CAI response in complex savannah systems like southern Africa that cannot be described as simple mixtures of green leaves, dry herbaceous material high in cellulose, and bare soil. Methods that use a complete set of multispectral channels at higher spatial resolution may be needed for accurate retrieval of fractional cover in Africa.     

Comparing ten classification methods for burned area mapping in a Mediterranean environment using Landsat TM satellite data

Various methods have been developed during the past three decades to improve the classification accuracy in burned area mapping using satellite data captured by different sensors. In this article, we compare ten such classification approaches using Landsat Thematic Mapper (TM) imagery on three Mediterranean test sites by evaluating the classification accuracy using (i) a traditional pixel-based approach, (ii) the concept of the Pareto boundary of efficient solution and (iii) linear regression analysis. Additionally, we make a discrimination of errors depending on their distribution and causal factor. The classification approaches compared resulted in not statistically significant differences in the accuracy of the burned area maps. Differences between the methods were also observed when considering the accuracy along the edges of the burned patches; however, again these were not statistically significant. The findings of our study in a Mediterranean environment clearly demonstrate that, for the selection of the most suitable classification approach, other factors could be given more weight, such as computational resources, imagery characteristics, availability of ancillary data, available software and the analyst's experience. Maybe the most important finding of our work is that the variance imposed by the methods is less than the variance imposed by factors differentiated locally in the three study sites since the between-group variance of the overall accuracy is higher than that of the within groups.     

Mapping vegetation-soil-climate complexes in southern Africa using temporal Fourier analysis of NOAA-AVHRR NDVI data

The distribution and phenology of vegetation is largely associated with climate, terrain characteristics and human activity. Satellite data provide the opportunity to monitor continuously the dynamics of vegetation, its changes and its impact on the environment. Monthly normalized difference vegetation index (NDVI) values which had been extracted from NOAA-AVHRR GAC data for Southern Africa (south of the Equator) from 1981 till 1991 were analysed using the Fast Fourier Transform (FFT). The FFT gives a new representation of the time series of images, i.e. a set of images of amplitude and phase (pixelwise) of periodic functions with different frequencies, which allows the analysis of the vegetation phenology using only the amplitude and phase of the most important periodic components. This approach is a powerful way to monitor various dynamic parameters of the vegetation in Southern Africa. The relationship between amplitude values and aridity and vegetation type was studied by correlating the amplitude and phase images with maps of the Budyko Aridity Index (ratio of net radiation to rainfall) and of the vegetation (White map). Images of amplitude and phase values were then used as attributes to map land units homogeneous as regards leaf display phenology. The map of these land units in Southern Africa, which has a 7.6 km resolution, was based on long term (9 years) information of 19 different vegetation-soil-climate complexes. Ten out of the 19 vegetation-soil-climate complexes are described in detail.     

Multi-class grasping classifiers using EEG data and a common spatial pattern filter

This paper describes the classification of various human actions from brain activity. In particular, we focus on grasping movements and estimate grasping patterns from electroencephalogram (EEG) data. EEG data is converted to grasping features by using a common spatial pattern filter (CSP filter), and the features are subsequently classified into grasping categories by using the k-nearest neighbor method. We tested the pipeline of feature extraction and classification on the EEG dataset. The EEG data were acquired while participants grasped an object according to the Cutkosky’s grasping taxonomy, in which grasping movements are categorized into nine power-type grasping patterns and seven precision-type grasping patterns. The best classification rate for 9-class power-type grasping patterns was 48% and for 7-class precision-type grasping patterns was 40%.     

Use of satellite spectral data in crop yield estimation surveys

Abstract

Landsat Thematic Mapper data are utilized to improve upon the ground yield survey estimates for wheat in India. The normalized difference and ratio vegetation indices computed from the spectral responses are used to obtain homogeneous vegetation vigour stratifications of the cropland of the study area. Certain post-stratified estimators that make use of these stratifications are investigated and are shown to provide improved crop yield estimates.     

A practical approach to retrieving the finer areas of algal bloom in inland lakes from coarse spatial resolution satellite data

Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) images have been used to monitor algal blooms in the open ocean, coastal waters, and inland lakes. However, it is difficult to obtain an accurate definition of algal bloom areas in inland lakes due to the spatial resolution of the generated images. This study developed a practical approach that uses a linear spectral mixing model with a moving window (LSMM), to obtain a finer algal bloom area. The approach analyses the differences in areas of algal bloom retrieved from MODIS images with 250 and 500 m spatial resolutions from 2012 to 2015 and synchronous VIIRS images with 750 m spatial resolution. Forty-two data sets with 126 satellite images were selected. The results showed that the average relative area difference (RAD) of algal bloom in the MODIS 500 m image was approximately 21.31% compared with the MODIS 250 m image and approximately 33.77% compared with the VIIRS image. A 5 × 5 window size was selected for the MODIS 500 m and VIIRS images. The results demonstrated that the approach can be successfully applied to MODIS 500 m and VIIRS images because the RAD significantly improved. The average RAD decreased to 9.39% in the MODIS 500 m image and to 12.84% in the VIIRS image. The relationship between the landscape of the algal bloom patch and the RAD showed that the performance of the LSMM method improved as the patch density (PD) increased from 0 to 2. When the perimeter-area ratio (PARA) is greater than 2 and the mean patch size (MPS) is less than approximately 5 km², the LSMM method significantly improved the RAD. An independent validation demonstrated that the LSMM method developed for MODIS and VIIRS images can be successfully applied to other coarser-resolution spatial imageries such as Geostationary Ocean Color Imager (GOCI) images. The LSMM method is more effective than the other methods for determining the fragmented landscape of algal blooms.     

Combined spectral and spatial processing of ERTS imagery data

In addition to spectral features, texture is an important spatial feature used in identifying objects or regions of interest in an image. Although texture is relatively easy for human observers to recognize and describe in empirical terms, it has been extremely refractory to precise definition and analysis by digital computers. This paper describes a procedure for extracting some easily computable features for the texture of blocks of digital image data and illustrates the applications of combined textural (spatial) and spectral features for identifying the land use categories of blocks of ERTS MSS (Earth Resources Technology Satellite Multi Spectral Scanner) data.The land use classification algorithm based on textural and spectral features was developed and tested using 614 image blocks of 64 × 64 resolution cells derived from an ERTS image over the Monterrey Bay area of the California coast line. The algorithm was applied on a training set of 314 blocks and tested on a set of 310 blocks. The overall accuracy of the classifier was found to be 83.5% on seven land use categories.     

Temporal and spatial variability in biomass burned areas across the USA derived from the GOES fire product

Burned area is a critical input to the algorithms of biomass burning emissions and understanding variability in fire activity due to climate change but it is difficult to estimate. This study presents a robust algorithm to reconstruct the patterns in burned areas across Contiguous United States (CONUS) in diurnal, seasonal, and interannual scales from 2000-2006. Specifically, burned areas in individual fire pixels are empirically calculated using diurnal variations in instantaneous fire sizes from the Geostationary Operational Environmental Satellites (GOES) WF_ABBA (Wildfire Automated Biomass Burning Algorithm) fire product. GOES burned areas exhibit diurnal variability with a temporal scale of half hours. The cumulative burned area during 9:00-16:00 local solar time accounts for 65%-81% of the total daily burned area. The diurnal variability is strongest in croplands compared to shrublands, grasslands, savannas, and forests. Analysis on a seasonal scale indicates that over 56% of burning occurs during summer (June-August). On average, the total annual burned area during the last seven years is 2.12 × 10⁴ ± 0.41 × 10⁴ km². The algorithm developed in this study can be applied to obtain burned area from the detections of GOES active fires at near real time, which can greatly improve the estimates of biomass burning emissions needed for predicting air quality.