Stefaan Lhermitte

The effect of atmospheric and topographic correction methods on land cover classification accuracy

a b s t r a c t Mapping of vegetation in mountain areas based on remote sensing is obstructed by atmospheric and topo- graphic distortions. A variety of atmospheric and topographic correction methods has been proposed to minimize atmospheric and topographic effects and should in principle lead to a better land cover classi- fication. Only a limited number of atmospheric and topographic combinations has been tested and the effect on class accuracy and on different illumination conditions is not yet researched extensively. The purpose of this study was to evaluate the effect of coupled correction methods on land cover classifica- tion accuracy. Therefore, all combinations of three atmospheric (no atmospheric correction, dark object subtraction and correction based on transmittance functions) and five topographic corrections (no topo- graphic correction, band ratioing, cosine correction, pixel-based Minnaert and pixel-based C-correction) were applied on two acquisitions (2009 and 2010) of a Landsat image in the Romanian Carpathian moun- tains. The accuracies of the fifteen resulting land cover maps were evaluated statistically based on two validation sets: a random validation set and a validation subset containing pixels present in the dif- ference area between the uncorrected classification and one of the fourteen corrected classifications. New insights into the differences in classification accuracy were obtained. First, results showed that all corrected images resulted in higher overall classification accuracies than the uncorrected images. The highest accuracy for the full validation set was achieved after combination of an atmospheric correction based on transmittance functions and a pixel-based Minnaert topographic correction. Secondly, class accuracies of especially the coniferous and mixed forest classes were enhanced after correction. There was only a minor improvement for the other land cover classes (broadleaved forest, bare soil, grass and water). This was explained by the position of different land cover types in the landscape. Finally, coupled correction methods showed most efficient on weakly illuminated slopes. After correction, accuracies in the low illumination zone (cos ˇ ≤ 0.65) were improved more than in the moderate and high illumination zones. Considering all results, best overall classification results were achieved after combination of the transmittance function correction with pixel-based Minnaert or pixel-based C-topographic correction. Furthermore, results of this bi-temporal study indicated that the topographic component had a higher influence on classification accuracy than the atmospheric component and that it is worthwhile to invest in both atmospheric and topographic corrections in a multi-temporal study.

Evaluating Landsat Thematic Mapper spectral indices for estimating burn severity of the 2007 Peloponnese wildfires in Greece

A vast area (more than 100 000 ha) of forest, shrubs and agricultural land burned on the Peloponnese peninsula in Greece during the 2007 summer. Three pre- and post-fire differenced Landsat Thematic Mapper (TM)-derived spectral indices were correlated with field data of burn severity for these devastating fires. These spectral indices were the Normalised Difference Vegetation Index (NDVI), the Normalised Difference Moisture Index (NDMI) and the Normalised Burn Ratio (NBR). The field data consist of 160 Geo Composite Burn Index (GeoCBI) plots. In addition, indices were evaluated in terms of optimality. The optimality statistic is a measure for the index’s sensitivity to fire-induced vegetation depletion. Results show that the GeoCBI–dNBR (differenced NBR) approach yields a moderately high R2 = 0.65 whereas the correlation between field data and the differenced NDMI (dNDMI) and the differenced NDVI (dNDVI) was clearly lower (respectively R2 = 0.50 and R2 = 0.46). The dNBR also outperformed the dNDMI and dNDVI in terms of optimality. The resulting median dNBR optimality equalled 0.51 whereas the median dNDMI and dNDVI optimality values were respectively 0.50 and 0.40 (differences significant for P < 0.001). However, inaccuracies observed in the spectral indices approach indicate that there is room for improvement. This could imply improved preprocessing, revised index design or alternative methods.

Illumination effects on the differenced Normalized Burn Ratio's optimality for assessing fire severity

The influence of illumination effects on the optimality of the dNBR (differenced Normalized Burn Ratio) was evaluated for the case of the 2007 Peloponnese (Greece) wildfires using a pre/post-fire Landsat TM (Thematic Mapper) image couple. Well-illuminated pixels (south and south-east facing slopes) exhibited more optimal displacements in the bi-spectral feature space than more shaded pixels (north and north-west exposed slopes). Moreover, pixels experiencing a small image-to-image difference in illumination obtained a higher optimality than pixels with a relatively large difference in illumination. To correct for illumination effects, the c-correction method and a modified c-correction technique were applied. The resulting median dNBR optimality of uncorrected, c-corrected and modified c-correction data was respectively 0.58, 0.60 and 0.71 (differences significant for p < 0.001). The original c-correction method improved the optimality of badly illuminated pixels while deteriorating the optimality of well-illuminated pixels. In contrast, the modified c-correction technique improved the optimality of all the pixels while retaining the prime characteristic of topographic correction techniques, i.e. detrending the illumination–reflectance relationship. For a minority of the data, for shaded pixels and/or pixels with a high image-to-image difference in illumination, the original c-correction outperformed the modified c-correction technique. In this study conducted in rugged terrain and with a bi-temporal image acquisition scheme that deviated up to two months from the ideal anniversary date scheme the modified c-correction technique resulted in a more reliable change detection.

Spatio-temporal variability in remotely sensed land surface temperature, and its relationship with physiographic variables in the Russian Altay Mountains

Abstract Spatio-temporal variability in energy fluxes at the earths surface implies spatial and temporal changes in observed land surface temperatures (LST). These fluxes are largely determined by variation in meteorological conditions, surface cover and soil characteristics. Consequently, a change in these parameters will be reflected in a different temporal LST behavior which can be observed by remotely sensed time series. Therefore, the objective of this paper is to perform a quantitative analysis on the parameters that determine this variability in LST to estimate the impact of changes in these parameters on the surface thermal regime. This study was conducted in the Russian Altay Mountains, an area characterized by strong gradients in meteorological conditions and surface cover. Spatio-temporal variability in LST was assessed by applying the fast Fourier transform (FFT) on 8 year of MODIS Aqua LST time series, herein considering both day and nighttime series as well as the diurnal difference. This FFT method was chosen as it allows to discriminate significant periodics, and as such enables distinction between short-term weather components, and strong, climate related, periodic patterns. A quantitative analysis was based on multiple linear regression models between the calculated, significant Fourier components (i.e. the annual and average component) and five physiographic variables representing the regional variability in meteorological conditions and surface cover. Physiographic predictors were elevation, potential solar insolation, topographic convergence, vegetation cover and snow cover duration. Results illustrated the strong inverse relationship between averaged daytime and diurnal difference LST and snow duration, with a R adj 2 of 0.85 and 0.60, respectively. On the other hand, nocturnal LST showed a strong connection with elevation and the amount of vegetation cover. Amplitudes of the annual harmonic experienced both for daytime and for nighttime LST similar trends with the set of physiographic variables – with stronger relationships at night. As such, topographic convergence was found to be the principal single predictor which demonstrated the importance of severe temperature inversions in the region. Furthermore, limited contribution of the physiographic predictors to the observed variation in the annual signal of the diurnal difference was retrieved, although a significant phase divergence was noticed between the majority of the study region and the perennial snowfields. Hence, this study gives valuable insights into the complexity of the spatio-temporal variability in LST, which can be used in future studies to estimate the ecosystems’ response on changing climatic conditions.

Magnitude- and Shape-Related Feature Integration in Hyperspectral Mixture Analysis to Monitor Weeds in Citrus Orchards

Traditionally, spectral mixture analysis (SMA) fails to fully account for highly similar ground components or endmembers. The high similarity between weed and crop spectra hampers the implementation of SMA for steering weed control management practices. To address this problem, this paper presents an alternative SMA technique, referred to as Integrated Spectral Unmixing (InSU). InSU combines both magnitude (i.e., reflectance) and shape (i.e., derivative reflectance) related features in an automated waveband selection protocol. Analysis was performed on different simulated mixed pixel spectra sets compiled from in situ-measured weed canopy, Citrus canopy, and soil spectra. Compared to traditional linear SMA, InSU significantly improved weed cover fraction estimations. An average decrease in fraction abundance error (Deltaf) of 0.09 was demonstrated for a signal-to-noise ratio (SNR) of 500 : 1, while for a SNR of 50 : 1, the decrease was 0.06.

Evaluation of pre/post-fire differenced spectral indices for assessing burn severity in a Mediterranean environment with Landsat Thematic Mapper

In this study several pre/post-fire differenced spectral indices for assessing burn severity in a Mediterranean environment are evaluated. GeoCBI (Geo Composite Burn Index) field data of burn severity were correlated with remotely sensed measures, based on the NBR (Normalized Burn Ratio), the NDMI (Normalized Difference Moisture Index) and the NDVI (Normalized Difference Vegetation Index). In addition, the strength of the correlation was evaluated for specific fuel types and the influence of the regression model type is pointed out. The NBR was the best remotely sensed index for assessing burn severity, followed by the NDMI and the NDVI. For this case study of the 2007 Peloponnese fires, results show that the GeoCBI–dNBR (differenced NBR) approach yields a moderate–high R 2 = 0.65. Absolute indices outperformed their relative equivalents, which accounted for pre-fire vegetation state. The GeoCBI–dNBR relationship was stronger for forested ecotypes than for shrub lands. The relationship between the field data and the dNBR and dNDMI (differenced NDMI) was nonlinear, while the GeoCBI–dNDVI (differenced NDVI) relationship appeared linear.

Hyperspectral Reflectance and Fluorescence Imaging to Detect Scab Induced Stress in Apple Leaves

Abstract: Apple scab causes significant losses in the production of this fruit. A timely and more site-specific monitoring and spraying of the disease could reduce the number of applications of fungicides in the fruit industry. The aim of this leaf-scale study therefore lies in the early detection of apple scab infections in a non-invasive and non-destructive way. In order to attain this objective, fluorescence- and hyperspectral imaging techniques were used. An experiment was conducted under controlled environmental conditions, linking hyperspectral reflectance and fluorescence imaging measurements to scab infection symptoms in a susceptible apple cultivar ( Malus x domestica Borkh. cv. Braeburn). Plant stress was induced by inoculation of the apple plants with scab spores. The quantum efficiency of Photosystem II (PSII) photochemistry was derived from fluorescence images of leaves under light adapted conditions. Leaves inoculated with scab spores were expected

A pixel based regeneration index using time series similarity and spatial context

Although the regeneration index based on control plots provides a valuable tool to quantify fire impact and subsequent vegetation regrowth, the practical implementation at large scale levels remains limited due to the need for detailed reference maps. The objective of this research therefore was the development of an image-based selection approach for control pixels based on time series similarity (TSS). The TSS approach allows the computation of a perpixel regeneration index on regional to global scale without the need for reference maps. Evaluation of the control plot selection approaches based on un-burnt focal pixels confirmed the validity of the TSS approach and showed optimal results for the TSSRMSD approach with and due to beneficial averaging effects and minimal window size effects. As such, the effects of spatial heterogeneity and noise are minimized and a preliminary quality indicator can be derived

The effect of atmospheric and topographic correction on pixel-based image composites: Improved forest cover detection in mountain environments

Abstract Quantification of forest cover is essential as a tool to stimulate forest management and conservation. Image compositing techniques that sample the most suited pixel from multi-temporal image acquisitions, provide an important tool for forest cover detection as they provide alternatives for missing data due to cloud cover and data discontinuities. At present, however, it is not clear to which extent forest cover detection based on compositing can be improved if the source imagery is firstly corrected for topographic distortions on a pixel-basis. In this study, the results of a pixel compositing algorithm with and without preprocessing topographic correction are compared for a study area covering 9 Landsat footprints in the Romanian Carpathians based on two different classifiers: Maximum Likelihood (ML) and Support Vector Machine (SVM). Results show that classifier selection has a stronger impact on the classification accuracy than topographic correction. Finally, application of the optimal method (SVM classifier with topographic correction) on the Romanian Carpathian Ecoregion between 1985, 1995 and 2010 shows a steady greening due to more afforestation than deforestation.

Webcams for Bird Detection and Monitoring: A Demonstration Study

Better insights into bird migration can be a tool for assessing the spread of avian borne infections or ecological/climatologic issues reflected in deviating migration patterns. This paper evaluates whether low budget permanent cameras such as webcams can offer a valuable contribution to the reporting of migratory birds. An experimental design was set up to study the detection capability using objects of different size, color and velocity. The results of the experiment revealed the minimum size, maximum velocity and contrast of the objects required for detection by a standard webcam. Furthermore, a modular processing scheme was proposed to track and follow migratory birds in webcam recordings. Techniques such as motion detection by background subtraction, stereo vision and lens distortion were combined to form the foundation of the bird tracking algorithm. Additional research to integrate webcam networks, however, is needed and future research should enforce the potential of the processing scheme by exploring and testing alternatives of each individual module or processing step.