Vegetation indices have been used extensively to estimate the vegetation density from satellite and airborne images for many years. In this paper, we focus on one of the most popular of such indices, the normalized difference vegetation index (NDVI), and we introduce a statistical framework to analyze it. As the degree of vegetation increases, the corresponding NDVI values begin to saturate and cannot represent highly vegetated regions reliably. By adopting the statistical viewpoint, we show how to obtain a linearized and more reliable measure. While the NDVI uses only red and near-infrared bands, we use the statistical framework to introduce new indices using the blue and green bands as well. We compare these indices with that obtained by linearizing the NDVI with extensive experimental results on real IKONOS multispectral images.

Linearized vegetation indices based on a formal statistical framework

Changing environmental and socio-economic conditions make land degradation, a major concern in Central and East Asia. Globally satellite imagery, particularly Normalized Difference Vegetation Index...

Tracking desertification on the Mongolian steppe through NDVI and field-survey data

PORÓWNANIE KILKU WKANIKÓW ROLINNOCI WYLICZONYCH NA PODSTAWIE JEDNOSEZONOWEJ SERII OBRAZÓW SPOT XS I ZBADANIE ICH U ̄YTECZNOCI DO OKRELANIA TYPU POKRYCIA TERENU ORAZ IDENTYFIKACJI WYBRANYCH UPRAW ROLNICZYCH Marek Mróz, Anna Sobieraj Katedra Fotogrametrii i Teledetekcji Uniwersytet Warmiñsko-Mazurski w Olsztynie S 3 o w a k l u c z o w e: wskaniki rolinnoci, identyfikacja upraw, SPOT XS. 40 Marek Mróz, Anna Sobieraj S t r e s z c z e n i e Znaczna liczba zaproponowanych dotychczas w literaturze wskaników rolinnoci sk3ania do poszukiwania najlepszego, najbardziej odpowiedniego wskanika do danego zastosowania teledetekcji. Porównano i zanalizowano wybrane wskaniki obliczone na podstawie obrazów satelitarnych SPOT XS z jednego sezonu wegetacyjnego (2003) dla obszaru ̄u3aw Wilanych. U¿ytecznoæ wskaników zbadano z punktu widzenia ich przydatnoci w identyfikacji upraw rolniczych i form pokrycia terenu. Wyniki wskazuj1, ¿e nie ma znacz1cych ró¿nic miêdzy prostymi i zaawansowanymi wskanikami rolinnoci, ani dla ró¿nych typów pokrycia terenu, ani odmiennych upraw rolniczych. Introduction Natural vegetation cover and agricultural crops are frequently the subjects of remote sensing studies. The information obtained in this way is a source of knowledge used for monitoring and evaluating the Earths vegetative cover. It is also used for environmental resources management. One of the ways enabling to get such information is determination of vegetation indices (VIs). Over years many vegetation indices have been proposed for determining the vigor and health of vegetation. Theirs formulas resulted from attempts to create better indices, which means that they take into account many factors, like soil reflectance, atmosphere, vegetation density etc. The aim of all these improvements and modifications is to get more reliable information about vegetation based on remotely sensed values (reflectances). The usual form of a vegetation index is a ratio of reflectance measured in two bands, or their algebraic combination. Spectral ranges (bands) to be used in VIs calculation are selected depending on the spectral properties of plants. First researchers interested in the spectral properties of plants were WILLSTATTER and STOLL (1913). They studied the phenomena of light entering plant cells. Later studies, concerning interactions between light of a different wavelength with vegetation, allowed to determine bands used in VIs. Nearly all commonly used VIs are based on near-infrared (NIR) and red (R) bands. In general, reflectance in Red band depends on chlorophyll content, and reflectance in Near Infrared  on internal structure of the plant cell. Reflectances in these spectra are uncorrelated with each other and they show high spectral contrast for vegetation. It should be mentioned that despite many studies it is sometimes difficult to determine precisely what is expressed by VIs. Although clearly correlated with chlorophyll absorption by foliage, it has been related to many plant properties, including the percentage of green cover and biomass, productivity, absorbed photosynthetically active radiation  APAR, leaf area index  LAI and biophysical properties such as photosynthetic capacity. The correlations between the biophysical parameters of plants and VIs, especially between LAI and the most popular index NDVI, are frequentlt the subjects of investigations. According to many authors when LAI > 3, NDVI 41 Comparison of Several Vegetation Indices Calculated on the Basis... becomes saturated. It means that information about vegetation is much poorer in a such situation. Hence some other VIs were proposed and elaborated. As demonstrated in many papers, the reflectance in R and NIR bands, and in consequence Vis, are also related to:  vegetation type (trees, bushes, grass, etc.), precisely type of leaves, their shape, inclination angle,  crop architecture,  plant growth stage,  leaf pigment content: chlorophyll and other pigments like carotene, xantophyll etc. (their composition and distribution in leaves),  water content of plants, and many other. Also the soil, atmosphere and geometry of Sun  target  sensor affect the radiation recorded by the sensor, and then influence the values of VIs. These factors introduce noise to the information which can be acquired from vegetation indices. Soil color can change the value of VI. Some studies have shown that dark soil resulted in higher values of VIs for parcels partially covered by vegetation, compared with bright soil when the NDVI index was calculated, and smaller values in the case of the PVI index. Changes in VIs accounted for approx. 50%. Moreover, the atmosphere and all suspended particles significantly influence the radiation reaching the sensor, and  indirectly  vegetation indices. In order to separate the influence of atmosphere from the vegetation signal, bands other than NIR and R are introduced to VIs formulas. This is usually a blue band. It is assumed that this band contains almost all effect of atmosphere on radiation (influence of scattering, absorbing etc.). In this study the blue band was unavailable among all SPOT specific spectral bands. Despite many factors disturbing the vegetation signal, VIs are sensitive to plant physiology, internal structure of the crop/vegetation community or their spatial pattern, and can be used for land cover and agricultural crop mapping. In the natural and intuitive approach, VIs are perceived as neochannels in colour compositions, substituting original bands as colour plans RGB (Red-Green-Blue). The use of VIs is recommended by e.g. the JRC/ /IPSC Institute in remote sensing methodology of agricultural crop identification within the framework of Common Agricultural Policy. The advantage of VIs is that they can exhibit, in a single index image, the properties of the objects originally registered in two bands. Thus we can reduce the number of bands and include in the composition additional channels obtained with one or several sensors, at the same or different times. Objective Many different vegetation indices (VIs) have been developed in last decades. Tasking into account their number, it seems that the following ques42 Marek Mróz, Anna Sobieraj tions should be answered: which VIs are the most suitable for mapping purposes, and at which stage of crop development? Are we allowed to use VIs arbitrarily, or should we determine which of them improves the accuracy of land cover and crop identification to the highest degree? These questions are stemming from fact that the sensitivity of VIs to many factors (e.g. plant growth stage, chlorophyll content, etc.) is different. The aim of the present study was to find VIs best suited for crop identification with a given multitemporal but uniseasonal SPOT XS data set. We had at our disposal five programmed multispectral SPOT images taken for the same season (2003) and agricultural test area. The satellite registrations were separated in time by some weeks or even a month, so the crop development phase, density, percentage of foliage cover and architecture have changed over this period. We have distinguished a priori 4 types of vegetation: cereals, rapeseed, sugar beet and maize that can be considered examples of different crop architecture. Thus it seemed interesting to compare several most popualr VIs for this data set, and choose the most suitable one in terms of its further use for land cover/crop identification. We examined the differences between VIs for the same and various crops on all registration dates, to find the most distinctive Vis. Then several VIs were used as input channels for making colour compositions. The measure of their suitability was the colour contrast of output images. Types of vegetation indices Vegetation indices can be classified into the following groups: 1) Slope-based VIs  the position of each point in 2-dimensional NIR-Red space is geometrically equivalent to the slope (tangent) of the line connecting the origin of reference and this particular point on a scattergram. 2) Distance-based VIs  they require to establish the soil line and measure the perpendicular distance of each point from this line. On scattergrams presenting soils and various plants, points of the soil form a line at approximately 45, passing through the origin. Some plants overlap this line (mostly arid-region plants and litter), whereas the majority of them are spread out in a fan towards the NIR axis. The determination of the slope and intercept of the soil line requires to draw a line through this cloud of points representing soil in NIR-Red space. Another difference between distance-based and slope-based VIs is the orientation of lines of equal vegetation intensity/vividness (isovegetation lines), regardless of moisture conditions. In slope-based VIs isolines converge at the origin, whereas in distance-based VIs all isolines remain parallel to the soil line. SAVI family VIs are also classified into this group, although isolines for these VIs are neither parallel nor convergent at the origin. 43 Comparison of Several Vegetation Indices Calculated on the Basis... 3) Orthogonal transformation VIs  based on decorrelation of a set of spectral bands through orthogonalization. These VIs produce new uncorrelated bands. Each band represents a different dimension of the information contained in the original band set. 4) Red Edge Inflection Point (REIP)  VIs based on waveform analysis techniques (REIP_Gaus (MILLER et al. 1990), REIP_Poly (BROGE, LEBLANC, 2001), REIP_Lagr (DAWSON, CURRAN 1998). They make use of the Gaussian, polynomial and Lagrangian models, respectively. The red edge is a characteristic feature of the spectral response of vegetation. It is a rapid change in reflectance observed in the spectrum of green plants at the transition between the visible and near-infrared wavelengths. The red edge is a fairly wide feature of approximately 30 nm. The measure

Comparison of several vegetation indices calculated on the basis of a seasonal SPOT XS time series, and their suitability for land cover and agricultural crop identification

It is well known that combining spatial and spectral information can improve land use classification from satellite imagery. Human activity on the ground, such as construction, induces changes in both the photometric structure of the image and in its spectral content owing to, primarily, changes in vegetation density and surface materials. This paper introduces a novel approach to combine spatial (more precisely, structural) information extracted from (1-m resolution) panchromatic Ikonos imagery with the multispectral response (4-m resolution) available from the same sensor. Of the prior work combining spatial and spectral information, none has extracted structural features as we do, and none has combined these information sources as early in the process. The classifier we describe here, discriminating urban and rural regions, is a front-end component of a fairly complete satellite image analysis system that identifies suburban residential areas and extracts their street networks and single-family houses. We extract structural information in the form of photometric straight lines and their spatial arrangement over relatively small neighborhoods. To capture the multispectral information, we turn to the well-known normalized difference vegetation index (NDVI) and an improved linearized version of our own development (details of the structural analysis and the theoretical development of the linearized NDVI appear elsewhere). This paper addresses the novel combination of these types of features (hybrids) by using the structural features, straight line support regions based on gradient orientation, as cue regions for multispectral analysis. We test the hybrid features in a range of parametric and nonparametric classifiers. We also implement and test a probabilistic relaxation algorithm followed by the maximum a priori decision rule. We report extensive results that indicate significant improvements in classification accuracy using the hybrid features.

Classifying land development in high-resolution Satellite imagery using hybrid structural-multispectral features

Pathogen infection has greatly reduced crop production. As the symptoms of diseases usually appear when the plants are infected severely, rapid identification approaches are required to monitor plant diseases at early the infection stage and optimize control strategies. Hyperspectral imaging, as a fast and nondestructive sensing technology, has achieved remarkable results in plant disease identification. Various models have been developed for disease identification in different plants such as arable crops, vegetables, fruit trees, etc. In these models, important algorithms, such as the vegetation index and machine learning classification and methods have played significant roles in the detection and early warning of disease. In this paper, the principle of hyperspectral imaging technology and common spectral characteristics of plant disease symptoms are discussed. We reviewed the impact mechanism of pathogen infection on the photo response and spectrum features of the plants, the data processing tools and algorithms of the hyperspectral information of pathogen-infected plants, and the application prospect of hyperspectral imaging technology for the identification of plant diseases.

/pdf/hyperspectral-sensing-of-plant-diseases-principle-and-2ue0jxh1.pdf

M. Kato

Papers

Relationship among vegetation variables and vegetation features of arid lands derived from satellite data