Data underlying the publication: Potato Virus Y Detection in Seed Potatoes Using Deep Learning on Hyperspectral Images
Polder, Gerrit ; Marrewijk, Bart van; Blok, Pieter ; Villiers, Hendrik de; Wolf, Jan van der; Kamp, Jan - \ 2020
Wageningen University & Research
disease detection - Erwinia - hyperspectral imaging - imaging spectroscopy - potato plants - Potato Y virus
Bacterial and virus diseases causes major damage in agriculture. Classifying those is an challenging but important task. This dataset contains +-274 000 spectral line images (512 pixels x 56 bands) of 6 different cultivars of healthy, bacterial (Erwinia) and virus (PVY) infected plants. The coordinates of each image is stored and can be compared with the stored position of individual plants labeled by crop experts
Comparison of remote sensing and plant trait-based modelling to predict ecosystem services in subalpine grasslands
Homolova, L. ; Schaepman, M.E. ; Lamarque, P. ; Clevers, J.G.P.W. ; Bello, F. de; Thuiller, W. ; Lavorel, S. - \ 2014
Ecosphere 5 (2014)8. - ISSN 2150-8925
land-use change - leaf chlorophyll content - imaging spectroscopy - water-content - aviris data - spectral reflectance - hyperspectral data - species richness - area index - vegetation
There is a growing demand for spatially explicit assessment of multiple ecosystem services (ES) and remote sensing (RS) can provide valuable data to meet this challenge. In this study, located in the Central French Alps, we used high spatial and spectral resolution RS images to assess multiple ES based on underpinning ecosystem properties (EP) of subalpine grasslands. We estimated five EP (green biomass, litter mass, crude protein content, species diversity and soil carbon content) from RS data using empirical RS methods and maps of ES were calculated as simple linear combinations of EP. Additionally, the RS-based results were compared with results of a plant trait-based statistical modelling approach that predicted EP and ES from land use, abiotic and plant trait data (modelling approach). The comparison between the RS and the modelling approaches showed that RS-based results provided better insight into the fine-grained spatial distribution of EP and thereby ES, whereas the modelling approach reflected the land use signal that underpinned trait-based models of EP. The spatial agreement between the two approaches at a 20-m resolution varied between 16 and 22% for individual EP, but for the total ecosystem service supply it was only 7%. Furthermore, the modelling approach identified the alpine grazed meadows land use class as areas with high values of multiple ES (hot spots) and mown-grazed permanent meadows as areas with low values and only few ES (cold spots). Whereas the RS-based hot spots were a small subset of those predicted by the modelling approach, cold spots were rather scattered, small patches with limited overlap with the modelling results. Despite limitations associated with timing of assessment campaigns and field data requirements, RS offers valuable data for spatially continuous mapping of EP and can thus supply RS-based proxies of ES. Although the RS approach was applied to a limited area and for one type of ecosystem, we believe that the broader availability of high fidelity airborne and satellite RS data will promote RS-based assessment of ES to larger areas and other ecosystems.
Imaging spectroscopy for ecological analysis in forest and grassland ecosystems
Homolova, L. - \ 2014
Wageningen University. Promotor(en): Michael Schaepman, co-promotor(en): Jan Clevers. - Wageningen : Wageningen University - ISBN 9789461738240 - 177
remote sensing - naaldbossen - alpenweiden - picea abies - bladoppervlakte - ecofysiologie - ecosysteemdiensten - vegetatie - chlorofyl - cartografie - beeldvormende spectroscopie - remote sensing - coniferous forests - alpine grasslands - picea abies - leaf area - ecophysiology - ecosystem services - vegetation - chlorophyll - mapping - imaging spectroscopy
Terrestrial vegetation is an important component of the Earth’s biosphere and therefore playing an essential role in climate regulation, carbon sequestration, and it provides large variety of services to humans. For a sustainable management of terrestrial ecosystems it is essential to understand vegetation responses to various pressures, to monitor and to predict the spatial extent and the rate of ecosystem changes. Remote sensing (RS) therefore offers a unique opportunity for spatially continuous, and for some type of RS data, also frequent monitoring of terrestrial ecosystems.
RS of vegetation is a broad research field, where a lot of progress has been made in the last three decades. However, the complexity of interactions between vegetation and solar radiation, constantly modulated by environmental factors, offers room for deeper investigation. Rather than solving one big research problem, this thesis built a few bridges on a way leading towards better understanding of using airborne imaging spectroscopy for ecological analysis in temperate coniferous forests and subalpine grasslands. The research was divided into a theoretical and an applied part. The theoretical part contributed to a critical evaluation of research achievements and challenges in optical RS of plant traits (Chapter 2). The applied part addressed three research topics: i) investigating variability of total to projected leaf area ratio in spruce canopies and its implications on RS of chlorophyll content (Chapter 3), ii) testing chlorophyll retrieval methods based on continuum removal in spruce canopies (Chapter 4), and iii) exploring potentials of imaging spectroscopy to map ecosystem properties and the capacity of subalpine grasslands in providing ecosystem services in comparison with a plant trait-based modelling approach (Chapter 5).
In Chapter 2, we reviewed achievements and challenges in RS estimation of key plant traits and we concentrated our discussion on eight traits with the strongest potential to be mapped using RS (plant growth and life forms, flammability properties, photosynthetic pathways and photosynthesis activity, plant height, leaf lifespan and phenology, specific leaf area, leaf nitrogen and phosphorous). The review indicated that imaging spectroscopy facilitates better retrievals of plant traits related to leaf biochemistry, photosynthesis and phenology rather than traits related to vegetations structure. Estimation of the canopy structure related traits (e.g. plant height) can certainly benefit from increasing synergies between imaging spectroscopy and active RS (radar or laser scanning). One of major challenges in RS of plant traits is to effectively suppress the negative influences of water absorption and canopy structure, which would facilitate more accurate retrievals of biochemical and photosynthesis-related traits. Secondly, a successful integration of RS and plant ecology concepts would require careful matching of spatial scales of in-situ trait data with RS observations.
In Chapter 3, measurement methods and variability of total to projected leaf area within spruce crowns were investigated. Comparison of six laboratory methods revealed that methods using an elliptic approximation of a needle shape underestimated total leaf area compared to methods using a parallelepiped approximation. The variability in total to projected leaf area was primarily driven by the vertical sampling position and less by needle age or forest stand age. We found that total leaf area estimation has an important implication on RS of leaf chlorophyll content. An error associated with biased estimates of total leaf area can reach up to 30% of the expected chlorophyll range commonly found in forest canopies and therefore negatively influences the validation of RS-based chlorophyll maps. In Chapter 4, potentials of the continuum removal transformation for mapping of chlorophyll content in spruce canopies were investigated. We tested two methods based on continuum removal: artificial neural networks and an optical index. The optical index was newly designed here and it was based on the spectral continuum between 650 and 720 nm. Both continuum removal based methods exhibited superior accuracy in chlorophyll retrieval compared to commonly used narrow-band vegetation indices (e.g. NDVI, TCARI/OSAVI). The newly designed index was equally accurate, but certainly provided a more operational approach as compared to the neural network.
In Chapter 5, mapping of ecosystem properties that underline ecosystem services provided by subalpine grasslands using RS methods was tested and further compared with a statistical plant trait-based modelling approach. Imaging spectroscopy in combination with empirical retrieval methods was partly successful to map ecosystem properties. The prediction accuracy at the calibration phase was comparable to the trait-based modelling approach. Spatial comparison between the two approaches revealed rather small agreement. The average fuzzy similarity between the approaches was around 20% for ecosystem properties, but in case of the total ecosystem service supply it decreased below 10%. However, the RS approach detected more variability in ecosystem properties and thereby in services, which was driven by local topography and microclimatic conditions, which could not be detected by the plant trait-based approach. Especially Chapters 2 and 5 indicated that one of the future RS research directions may be in spatial ecology, i.e. spatially explicit mapping of plant traits, ecosystem properties and ecosystem services. High quality RS data are certainly essential building elements for spatial ecology. But in order to address the effects of climate and land use changes on biodiversity and ecosystems, their properties and services, the integration of in-situ and RS data will be ultimately required. Therefore, more coherent experiments, where in-situ and RS data are measured simultaneously at different spatial scales, are needed in the future.
Mapping a priori defined plant associations using remotely sensed vegetation characteristics
Roelofsen, H.D. ; Kooistra, L. ; Bodegom, P.M. van; Verrelst, J. ; Krol, J. ; Witte, J.M.P. - \ 2014
Remote Sensing of Environment 140 (2014). - ISSN 0034-4257 - p. 639 - 651.
ellenberg indicator values - continuous floristic gradients - hyperspectral imagery - imaging spectroscopy - endmember selection - tropical forests - aviris data - classification - regression - moisture
Incorporation of a priori defined plant associations into remote sensing products is a major challenge that has only recently been confronted by the remote sensing community. We present an approach to map the spatial distribution of such associations by using plant indicator values (IVs) for salinity, moisture and nutrients as an intermediate between spectral reflectance and association occurrences. For a 12 km2 study site in the Netherlands, the relations between observed IVs at local vegetation plots and visible and near-infrared (VNIR) and short-wave infrared (SWIR) airborne reflectance data were modelled using Gaussian Process Regression (GPR) (R2 0.73, 0.64 and 0.76 for salinity, moisture and nutrients, respectively). These relations were applied to map IVs for the complete study site. Association occurrence probabilities were modelled as function of IVs using a large database of vegetation plots with known association and IVs. Using the mapped IVs, we calculated occurrence probabilities of 19 associations for each pixel, resulting in both a crisp association map with the most likely occurring association per pixel, as well as occurrence probability maps per association. Association occurrence predictions were assessed by a local vegetation expert, which revealed that the occurrences of associations situated at frequently predicted indicator value combinations were over predicted. This seems primarily due to biases in the GPR predicted IVs, resulting in associations with envelopes located in extreme ends of IVs being scarcely predicted. Although the results of this particular study were not fully satisfactory, the method potentially offers several advantages compared to current vegetation classification techniques, like site-independent calibration of association probabilities, site-independent selection of associations and the provision of IV maps and occurrence probabilities per association. If the prediction of IVs can be improved, this method may thus provide a viable roadmap to bring a priori defined plant associations into the domain of remote sensing.
Non-linear partial least square regression increases the estimation accuracy of grass nitrogen and phosphorus using in situ hyperspectral and environmental data
Ramoelo, A. ; Skidmore, A.K. ; Cho, M.A. ; Mathieu, R. ; Heitkonig, I.M.A. ; Dudeni-Tlhone, N. ; Schlerf, M. ; Prins, H.H.T. - \ 2013
ISPRS Journal of Photogrammetry and Remote Sensing 82 (2013). - ISSN 0924-2716 - p. 27 - 40.
kruger-national-park - multiple linear-regression - band-depth analysis - vegetation indexes - south-africa - chlorophyll estimation - imaging spectroscopy - absorption features - biochemical content - mineral-nutrition
Grass nitrogen (N) and phosphorus (P) concentrations are direct indicators of rangeland quality and provide imperative information for sound management of wildlife and livestock. It is challenging to estimate grass N and P concentrations using remote sensing in the savanna ecosystems. These areas are diverse and heterogeneous in soil and plant moisture, soil nutrients, grazing pressures, and human activities. The objective of the study is to test the performance of non-linear partial least squares regression (PLSR) for predicting grass N and P concentrations through integrating in situ hyperspectral remote sensing and environmental variables (climatic, edaphic and topographic). Data were collected along a land use gradient in the greater Kruger National Park region. The data consisted of: (i) in situ-measured hyperspectral spectra, (ii) environmental variables and measured grass N and P concentrations. The hyperspectral variables included published starch, N and protein spectral absorption features, red edge position, narrow-band indices such as simple ratio (SR) and normalized difference vegetation index (NDVI). The results of the non-linear PLSR were compared to those of conventional linear PLSR. Using non-linear PLSR, integrating in situ hyperspectral and environmental variables yielded the highest grass N and P estimation accuracy (R2 = 0.81, root mean square error (RMSE) = 0.08, and R2 = 0.80, RMSE = 0.03, respectively) as compared to using remote sensing variables only, and conventional PLSR. The study demonstrates the importance of an integrated modeling approach for estimating grass quality which is a crucial effort towards effective management and planning of protected and communal savanna ecosystems.
Measurement methods and variability assessment of the Norway spruce total leaf area: Implications for remote sensing
Homolova, L. ; Lukes, P. ; Malenovsky, Z. ; Lhotakova, Z. ; Kaplan, V. ; Hanus, J. - \ 2013
Trees-Structure and Function 27 (2013)1. - ISSN 0931-1890 - p. 111 - 121.
chlorophyll-a - light-interception - hyperspectral data - picea-abies - imaging spectroscopy - conifer needles - surface-area - gas-exchange - canopy - biochemistry
Estimation of total leaf area (LAT) is important to express biochemical properties in plant ecology and remote sensing studies. A measurement of LAT is easy in broadleaf species, but it remains challenging in coniferous canopies. We proposed a new geometrical model to estimate Norway spruce LAT and compared its accuracy with other five published methods. Further, we assessed variability of the total to projected leaf area conversion factor (CF) within a crown and examined its implications for remotely sensed estimates of leaf chlorophyll content (Cab). We measured morphological and biochemical properties of three most recent needle age classes in three vertical canopy layers of a 30 and 100-year-old spruce stands. Newly introduced geometrical model and the parallelepiped model predicted spruce LAT with an error >5 % of the average needle LAT, whereas two models based on an elliptic approximation of a needle shape underestimated LAT by up to 60 %. The total to projected leaf area conversion factor varied from 2. 5 for shaded to 3. 9 for sun exposed needles and remained invariant with needle age class and forest stand age. Erroneous estimation of an average crown CF by 0. 2 introduced an error of 2-3 µg cm-2 into the crown averaged Cab content. In our study, this error represents 10-15 % of observed crown averaged Cab range (33-53 µg cm-2). Our results demonstrate the importance of accurate LAT estimates for validation of remotely sensed estimates of Cab content in Norway spruce canopies.
Differentiation of plant age in grasses using remote sensing
Knox, N. ; Skidmore, A.K. ; Werff, H.M.A. van der; Groen, T.A. ; Boer, W.F. de; Prins, H.H.T. ; Kohi, E. ; Peel, M. - \ 2013
International Journal of applied Earth Observation and Geoinformation 24 (2013)10. - ISSN 0303-2434 - p. 54 - 62.
difference water index - monitoring vegetation - nitrogen concentration - imaging spectroscopy - hyperspectral data - boreal regions - time-series - green-up - phenology - reflectance
Phenological or plant age classification across a landscape allows for examination of micro-topographical effects on plant growth, improvement in the accuracy of species discrimination, and will improve our understanding of the spatial variation in plant growth. In this paper six vegetation indices used in phenological studies (including the newly proposed PhIX index) were analysed for their ability to statistically differentiate grasses of different ages in the sequence of their development. Spectra of grasses of different ages were collected from a greenhouse study. These were used to determine if NDVI, NDWI, CAI, EVI, EVI2 and the newly proposed PhIX index could sequentially discriminate grasses of different ages, and subsequently classify grasses into their respective age category. The PhIX index was defined as: (An VNIR+ log(An SWIR2))/(An VNIR- log(An SWIR2)), where An VNIRand An SWIR2are the respective nor- malised areas under the continuum removed reflectance curve within the VNIR (500-800 nm) and SWIR2 (2000-2210 nm) regions. The PhIX index was found to produce the highest phenological classification accuracy (Overall Accuracy: 79%, and Kappa Accuracy: 75%) and similar to the NDVI, EVI and EVI2 indices it statistically sequentially separates out the developmental age classes. Discrimination between seedling and dormant age classes and the adult and flowering classes was problematic for most of the tested indices. Combining information from the visible near infrared (VNIR) and shortwave infrared region (SWIR) region into a single phenological index captures the phenological changes associated with plant pigments and the ligno-cellulose absorption feature, providing a robust method to discriminate the age classes of grasses. This work provides a valuable contribution into mapping spatial variation and monitoring plant growth across savanna and grassland ecosystems.
Quantifying structure of Natura 2000 heathland habitats using spectral mixture analysis and segmentation techniques on hyperspectral imagery
Mücher, C.A. ; Kooistra, L. ; Vermeulen, M.H. ; Vanden Borre, J. ; Haest, B. ; Haveman, R. - \ 2013
Ecological Indicators 33 (2013)suppl. C. - ISSN 1470-160X - p. 71 - 81.
imaging spectroscopy - satellite imagery - biodiversity - indicators - california - vegetation - ecosystem - scale - space
Monitoring of habitat types protected under the Annex I of the EU Habitats Directive requires every 6 years information to be reported on their conservation status (area, range, structure and function) in the member states. Hyperspectral imagery can be an important source of information to assist in the evaluation of the habitats’ conservation status, as it can provide continuous maps of habitat quality indicators (e.g., life forms, management activities, grass, shrub and tree encroachment) at the pixel level. Such local level information is highly needed for management purposes, e.g., the location of habitat patches and their sizes and quality within a protected site. This paper focuses on the use of continuous fraction images as derived from spectral mixture analysis of hyperspectral imagery (AHS-160), in combination with segmentation techniques, to facilitate habitat quality assessment in a heathland site in the Netherlands. This combined application of techniques on hyperspectral imagery demonstrates the usefulness of information from continuous fraction maps of grass abundance (Molinia caerulea) in heathlands – at and within the patch level – compared to traditional mapping techniques that assess grass encroachment in a limited number of abundance classes at the patch level. It therefore provides a better basis to monitor large areas for processes such as grass encroachment that largely determine the conservation status of Natura 2000 heathland areas. Timely, accurate and up-to-date spatial information on the encroachment of mosses, grasses, shrubs or trees (dominant species) can help conservation managers to take better decisions and to better evaluate the effect of taken measures. While discrepancies exist between the results of field-based vegetation surveys and the proposed remote sensing approach, we provide a discussion on the uncertainty of determining which of both methods is most accurate in relation to dominant species, which is in our case Molinia caerulea, and set forth several reasons why the remote sensing based approach might form a better basis for the monitoring of abundant species and patch evolution through time.
A New Minimum-Volume Enclosing Algorithm for Endmember Identification and Abundance Estimation in Hyperspectral Data
Hendrix, E.M.T. ; García, I. ; Plaza, J. ; Martín, G. ; Plaza, A. - \ 2012
IEEE Transactions on Geoscience and Remote Sensing 50 (2012)7. - ISSN 0196-2892 - p. 2744 - 2757.
nonnegative matrix factorization - spectral mixture analysis - imaging spectroscopy - n-findr - extraction - imagery
Spectral unmixing is an important technique for hyperspectral data exploitation, in which a mixed spectral signature is decomposed into a collection of spectrally pure constituent spectra, called endmembers, and a set of correspondent fractions, or abundances, that indicate the proportion of each endmember present in the mixture. Over the last years, several algorithms have been developed for automatic or semiautomatic endmember extraction. Some available approaches assume that the input data set contains at least one pure spectral signature for each distinct material and further conduct a search for the most spectrally pure signatures in the high-dimensional space spanned by the hyperspectral data. Among these approaches, those aimed at maximizing the volume of the simplex that can be formed using available spectral signatures have found wide acceptance. However, the presence of spectrally pure constituents is unlikely in remotely sensed hyperspectral scenes due to spatial resolution, mixing phenomena, and other considerations. In order to address this issue, other available algorithms have been developed to generate virtual endmembers (not necessarily present among the input data samples) by finding the simplex with minimum volume that encloses all available observations. In this paper, we discuss maximum-volume versus minimum-volume enclosing solutions and further develop a novel algorithm in the latter category which incorporates the fractional abundance estimation as an internal step of the endmember searching process (i.e., it does not require an external method to produce endmember fractional abundances). The method is based on iteratively enclosing the observations in a lower dimensional space and removing observations that are most likely not to be enclosed by the simplex of the endmembers to be estimated. The performance of the algorithm is investigated and compared to that of other algorithms (with and without the pure pixel assumption) using synthetic a- d real hyperspectral data sets collected by a variety of hyperspectral imaging instruments.
Remote sensing of forage nutrients: Combining ecological and spectral absorption feature data
Knox, N. ; Skidmore, A.K. ; Prins, H.H.T. ; Heitkonig, I.M.A. ; Slotow, R. ; Waal, C. van der; Boer, W.F. de - \ 2012
ISPRS Journal of Photogrammetry and Remote Sensing 72 (2012). - ISSN 0924-2716 - p. 27 - 35.
south-african savanna - multiple linear-regression - kruger-national-park - mineral-nutrition - leaf biochemistry - hyperspectral reflectance - nitrogen concentration - imaging spectroscopy - grass - quality
Forage quality in grassland-savanna ecosystems support high biomass of both wild ungulates and domestic livestock. Forage quality is however variable in both space and time. In this study findings from ecological and laboratory studies, focused on assessing forage quality, are combined to evaluate the feasibility of a remote sensing approach for predicting the spatial and temporal variations in forage quality. Spatially available ecological findings (ancillary data), and physically linked spectral data (absorption data) are evaluated in this study and combined to create models which predict forage quality (nitrogen, phosphorus and fibre concentrations) of grasses collected in the Kruger National Park, South Africa, and analysed in both dry and wet seasons. Models were developed using best subsets regression modelling. Ancillary data alone, could predict forage components, with a higher goodness of fit and predictive capability, than absorption data (Ancillary: R2 adj ¼ 0:42—0:74 compared with absorption: R2 adj ¼ 0:11—0:51, and lower RMSE values for each nutrient produced by the ancillary models). Plant species and soil classes were found to be ecological variables most frequently included in prediction models of ancillary data. Models in which both ancillary and absorption variables were included, had the highest predictive capabilities ( R2 adj ¼ 0:49—0:74 and lowest RMSE values) compared to models where data sources were derived from only one of the two groups. This research provides an important step in the process of creating biochemical models for mapping forage nutrients in savanna systems that can be generalised seasonally over large areas.
Geosensors to support crop production: current applications and user requirements
Thessler, S. ; Kooistra, L. ; Teye, F. ; Huitu, H. ; Bregt, A.K. - \ 2011
Sensors 11 (2011)7. - ISSN 1424-8220 - p. 6656 - 6684.
wireless sensor networks - soil organic-carbon - diffuse-reflectance spectroscopy - grain protein-concentration - unmanned aerial vehicles - weed-control - winter-wheat - irrigation management - precision agriculture - imaging spectroscopy
Sensor technology, which benefits from high temporal measuring resolution, real-time data transfer and high spatial resolution of sensor data that shows in-field variations, has the potential to provide added value for crop production. The present paper explores how sensors and sensor networks have been utilised in the crop production process and what their added-value and the main bottlenecks are from the perspective of users. The focus is on sensor based applications and on requirements that users pose for them. Literature and two use cases were reviewed and applications were classified according to the crop production process: sensing of growth conditions, fertilising, irrigation, plant protection, harvesting and fleet control. The potential of sensor technology was widely acknowledged along the crop production chain. Users of the sensors require easy-to-use and reliable applications that are actionable in crop production at reasonable costs. The challenges are to develop sensor technology, data interoperability and management tools as well as data and measurement services in a way that requirements can be met, and potential benefits and added value can be realized in the farms in terms of higher yields, improved quality of yields, decreased input costs and production risks, and less work time and load
Impact of multiangular information on empirical models to estimate canopy nitrogen concentration in mixed forest
Huber, S. ; Koetz, B. ; Psomas, A. ; Kneubuehler, M. ; Schopfer, J.T. ; Itten, K.I. ; Zimmermann, N.E. - \ 2010
Journal of Applied Remote Sensing 4 (2010)1. - ISSN 1931-3195
hyperspectral brdf data - imaging spectroscopy - bidirectional reflectance - spectral measurements - absorption features - ecosystem processes - chlorophyll content - carbon - leaf - photosynthesis
Directional effects in remotely sensed reflectance data can influence the retrieval of plant biophysical and biochemical estimates. Previous studies have demonstrated that directional measurements contain added information that may increase the accuracy of estimated plant structural parameters. Because accurate biochemistry mapping is linked to vegetation structure, also models to estimate canopy nitrogen concentration (C-N) may be improved indirectly from using multiangular data. Hyperspectral imagery with five different viewing zenith angles was acquired by the spaceborne CHRIS sensor over a forest study site in Switzerland. Fifteen canopy reflectance spectra corresponding to subplots of field-sampled trees were extracted from the preprocessed CHRIS images and subsequently two-term models were developed by regressing C-N on four datasets comprising either original or continuum-removed reflectances. Consideration is given to the directional sensitivity of the C-N estimation by generating regression models based on various combinations (n=15) of observation angles. The results of this study show that estimating canopy C-N with only nadir data is not optimal irrespective of spectral data processing. Moreover adding multiangular information improves significantly the regression model fits and thus the retrieval of forest canopy biochemistry. These findings support the potential of multiangular Earth observations also for application-oriented ecological monitoring.
Effects of woody elements on simulated canopy reflectance: implications for forest chlorophyll content retrieval
Verrelst, J. ; Schaepman, M.E. ; Malenovsky, Z. ; Clevers, J.G.P.W. - \ 2010
Remote Sensing of Environment 114 (2010)3. - ISSN 0034-4257 - p. 647 - 656.
leaf optical-properties - light-use efficiency - resolution satellite imagery - radiative-transfer models - high-spatial-resolution - spectral reflectance - imaging spectroscopy - vegetation indexes - hyperspectral reflectance - sensitivity-analysis
An important bio-indicator of actual plant health status, the foliar content of chlorophyll a and b (Cab), can be estimated using imaging spectroscopy. For forest canopies, however, the relationship between the spectral response and leaf chemistry is confounded by factors such as background (e.g. understory), canopy structure, and the presence of non-photosynthetic vegetation (NPV, e.g. woody elements)—particularly the appreciable amounts of standing and fallen dead wood found in older forests. We present a sensitivity analysis for the estimation of chlorophyll content in woody coniferous canopies using radiative transfer modeling, and use the modeled top-of-canopy reflectance data to analyze the contribution of woody elements, leaf area index (LAI), and crown cover (CC) to the retrieval of foliar Cab content. The radiative transfer model used comprises two linked submodels: one at leaf level (PROSPECT) and one at canopy level (FLIGHT). This generated bidirectional reflectance data according to the band settings of the Compact High Resolution Imaging Spectrometer (CHRIS) from which chlorophyll indices were calculated. Most of the chlorophyll indices outperformed single wavelengths in predicting Cab content at canopy level, with best results obtained by the Maccioni index ([R780 - R710] / [R780 - R680]). We demonstrate the performance of this index with respect to structural information on three distinct coniferous forest types (young, early mature and old-growth stands). The modeling results suggest that the spectral variation due to variation in canopy chlorophyll content is best captured for stands with medium dense canopies. However, the strength of the up-scaled Cab signal weakens with increasing crown NPV scattering elements, especially when crown cover exceeds 30%. LAI exerts the least perturbations. We conclude that the spectral influence of woody elements is an important variable that should be considered in radiative transfer approaches when retrieving foliar pigment estimates in heterogeneous stands, particularly if the stands are partly defoliated or long-lived
Forage quality of savannas - Simultaneously mapping foliar protein and polyphenols for trees and grass using hyperspectral imagery
Skidmore, A.K. ; Ferwerda, J.G. ; Mutanga, O. ; Wieren, S.E. van; Peel, M.J.S. ; Grant, R.C. ; Prins, H.H.T. ; Balcik, F. ; Venus, V. - \ 2010
Remote Sensing of Environment 114 (2010)1. - ISSN 0034-4257 - p. 64 - 72.
kruger-national-park - african savanna - south-africa - imaging spectroscopy - mammalian herbivores - mineral-nutrition - diet selection - neural-network - nitrogen - vegetation
Savanna covers about two-thirds of Africa, with forage quantity and quality being important factors determining the distribution and density of wildlife and domestic stock. Testing hypotheses about the distribution of herbivores is hampered by the absence of reliable methods for measuring the variability of vegetation quality (e.g. biochemical composition) across the landscape. It is demonstrated that hyperspectral remote sensing fills this gap by revealing simultaneously the spatial variation of foliar nitrogen (crude protein) as well as the total amount of polyphenols, in grasses and trees. For the first time, the pattern of resources important for feeding preferences in herbivores (polyphenols and nitrogen) is mapped across an extensive landscape and the modeled foliar concentrations are shown to fit with ecological knowledge of the area. We explain how estimates of nitrogen (crude protein) and polyphenols may be scaled up from point-based observations to reveal their spatial pattern, and how the variation in forage quality can influence the management of savannas, including farms, communal grazing areas, and conservation areas. It provides a glimpse of the choices herbivores must face in selecting food resources of different qualities.
Quantitative remote sensing for monitoring forest canopy structural variables in the Three Gorges region of China
Zeng, Y. - \ 2008
Wageningen University. Promotor(en): Michael Schaepman, co-promotor(en): Jan Clevers; B. Wu. - S.l. : s.n. - ISBN 9789085049111 - 119
kroondak - bossen - remote sensing - schaalverandering - china - bosstructuur - beeldvormende spectroscopie - canopy - forests - remote sensing - scaling - china - forest structure - imaging spectroscopy
Bridging various scales ranging from local to regional and global, remote sensing has facilitated extraordinary advances in modeling and mapping ecosystems and their functioning. Since forests are one of the most important natural resources on the terrestrial Earth surface, accurate and up-to-date information on forest structure and its changes are essential for many aspects of forest management. In particular the quantitative monitoring of forest structure using remote sensing techniques strongly supports conservation strategies that take into account biodiversity and the impact of the global carbon cycle.
China is a vast country with abundant forest resources. This thesis focuses in particular on the Three Gorges region of China, where currently major changes are taking place in the forest ecosystem. Certainly, the Three Gorges region is widely known due to the construction of the Three Gorges Dam. But the Chinese government also puts great importance on eco-environmental aspects of the Three Gorges Dam project and has therefore implemented a long-term investigation intending to monitor the changing environment. Within the Three Gorges region, the Longmenhe forest nature reserve has been selected as one of the main study sites for this thesis. This forest nature reserve is dominated by subtropical broadleaved and coniferous forests and the pilot study in the reserve enables monitoring forest structural variables as well as detecting their changes in the whole Three Gorges region.
Quantitative retrieval methods for assessing forest canopy structural variables using remote sensing are commonly grouped into statistical and physical approaches. Inverting physical-based canopy reflectance models for estimating forest variables generally can be applied at different sites and with different sensors. Dealing with scales and scaling currently is one of the central issues in quantitative remote sensing. A better understanding of the different spectral, spatial and temporal scales and a further study on scaling the information from local to regional scales are necessary. Therefore, the main objective of this thesis is to develop a methodology for quantitatively monitoring forest canopy structural variables and their change by integrating multiple scale remote sensing techniques.
In Chapter 2, the potential of hyperspectral EO-1 Hyperion data combined with the inverted physical-based Li-Strahler geometric-optical model for retrieving mean crown closure (CC) and mean crown diameter (CD) as forest canopy structural variables in the Longmenhe forest nature reserve is studied. One of the most important inputs for the model inversion is the fractional contribution of sunlit background (Kg), which is obtained by using linear spectral unmixing methods based on image-derived endmembers of the viewed scene components (sunlit and shaded canopy, sunlit and shaded background). Validation results (37 field samples) show confidence (R2CC=0.61, RMSECC=0.046, R2CD=0.39 and RMSECD=0.984) in the approach selected.
Chapter 3 studies the feasibility of up-scaling from very high spatial resolution data (QuickBird) to high spatial resolution hyperspectral data (Hyperion) for extracting the endmembers of sunlit canopy, sunlit background and shadow. It can be concluded that the regional scaling-based endmembers calculated in the overlapping region of QuickBird and Hyperion using the linear unmixing model are the best ones to be used in combination with the Li-Strahler model inversion for mapping CC and CD in the Longmenhe forest nature reserve. Additionally, the estimation of CC is better than that of CD by inverting the Li-Strahler model on a per-pixel basis.
The inverted Li-Strahler model combined with the regional scaling method, used at a local scale in the Longmenhe study area with QuickBird and Hyperion images, can also be applied at the scale of the Three Gorges region by using the combination of Landsat TM and MODIS images as shown in Chapter 4. For the two years 2002 and 2004, this methodology yields similar accuracies in CC estimation based on 25 field validation samples (R22002=0.614, RMSE2002=0.060; and R22004=0.631, RMSE2004=0.052). The produced map with changes in CC from 2002 to 2004 shows a decrease in CC in the eastern counties of the Three Gorges region located close to the Three Gorges Dam and an increase in CC in other counties implying a positive response to certain policies taken safeguarding forest resources.
The inversion of two canopy reflectance models (the Kuusk-Nilson forest reflectance and transmittance (FRT) model and the Li-Strahler geometric-optical model) for estimating forest CC using Hyperion data in the Longmenhe study area is compared in Chapter 5. The “infeasible” areas (i.e. pixels for which the estimated fraction sunlit background falls not in the range between [0, 1]) from the Li-Strahler model inversion are filled by using a spatial interpolation algorithm based on regression kriging. Validation results (40 field samples) show that the estimated CC by the FRT model inversion has a limited range of variation and is less accurate (R2=0.53, RMSE=0.072) than the estimation by inverting the Li-Strahler model combined with the scaling method and interpolation (R2=0.67, RMSE=0.043). Consequently, in Chapter 6, spatially continuous CC maps for the Three Gorges region in both 2002 and 2004 are produced by integrating the results of Chapter 4 and this spatial interpolation technique. The final improved change map of CC is more suitable to predict and analyze the overall situation of the forest structural change in the whole Three Gorges region.
The main contribution of this work is the integration of the inverted Li-Strahler model, a regional scaling-based endmember extraction method and a spatial interpolation technique to achieve quantitative monitoring of forest canopy structural changes. The approach includes the careful assessment of various scaling aspects namely ranging from multi-spectral to hyperspectral, from high spatial resolution to low spatial resolution, from mono-temporal to multi-temporal and from local to regional study areas. Systematic and structural monitoring of forest ecosystem changes will be feasible at unprecedented quality based on the suggested approach.
CARBIS final report : detecting soil carbon and its spatial variability by imaging spectroscopy
Stevens, A. ; Wesemael, B. van; Bartholomeus, H. ; Rossilon, D. ; Tychon, B. ; Ben-Dor, E. - \ 2007
Louvain-la-Neuve : Belspo - 53
organische koolstof - grondanalyse - koolstofvastlegging in de bodem - beeldvormende spectroscopie - organic carbon - soil analysis - soil carbon sequestration - imaging spectroscopy
Combining close-range and remote sensing for local assessment of biophysical characteristics of arable land
Heijden, G.W.A.M. van der; Clevers, J.G.P.W. ; Schut, A.G.T. - \ 2007
International Journal of Remote Sensing 28 (2007)24. - ISSN 0143-1161 - p. 5485 - 5502.
hyperspectral vegetation indexes - imaging spectroscopy - grass swards - precision agriculture - narrow-band - broad-band - models - yield - prediction - quality
For crop management, information on the actual status of the crop is important for taking decisions on nitrogen supply, water supply or harvesting. One would also like to take into account the local spatial variation of the crop. Remote sensing has proved to be a useful technique for estimating and mapping the spatial variation of various biophysical variables. Calibration of the image data is crucial in the performance and applicability of this technique. The aim of this paper is to show the possibility to calibrate remotely sensed imagery using fast and non-destructive close-range (below 1.3m height) sensing instruments, thus providing a means for the assessment of plant characteristics over large areas at low costs. This concept was tested on a homogeneously managed grassland field, subdivided into 20 plots of 15 x 3m, at the end of July 2004. Reflected radiation was recorded with an active close-range sensing device, consisting of a visible light and near-infrared (NIR) imaging spectrograph, and a 3CCD camera, equipped with special band filters (central wavelengths are at 600, 710 and 800 nm). An airborne campaign with a four-band UltraCam digital CCD camera was used for extrapolation to larger scales. Plots were harvested, and fresh and dry biomass and leaf nitrogen content were determined. Partial least squares (PLS) models combining spectral and spatial information from the close-sensing device yielded acceptable results in predicting grassland yields and nitrogen content. Subsequently, these predictions were used to calibrate a model with the image data of the remote sensing device. These were then compared, using leave-one-out cross-validation, with the measured field variables, and the model proved to have an acceptable predictive power.
Imaging spectroscopy : applications in agriculture
Zedde, H.J. van de; Brakel, R.P. van - \ 2007
spectroscopie - kwaliteitscontroles - monitoring - sensorische evaluatie - voedselinspectie - tomaten - patates frites - beeldvormende spectroscopie - spectroscopy - quality controls - monitoring - sensory evaluation - food inspection - tomatoes - chips (French fries) - imaging spectroscopy
Imaging Spectroscopy is the study of light as a function of spatial distribution and wavelength that has been transmitted, emitted, reflected or scattered from an object. This allows us to derive information about the spatial relation of the chemistry of the object. Imaging spectroscopy is suited for the following tasks: • Quality control: detection of latent defects in agri-products, e.g. vegetables and fruit. • Quantification of compounds: carotenes, proteins, sugars, moisture etc. In this poster the following two applications are discussed: 1) Measuring of compounds in tomatoes and 2) Detection and classification of latent defects in French Fries
Measuring compounds in fruits using spectral image analysis
Zedde, H.J. van de; Heijden, G.W.A.M. van der; Polder, G. - \ 2007
rijp worden - rijpheid - multispectrale beelden - chemische samenstelling - vruchten - voedseltechnologie - chemische verbindingen - beeldvormende spectroscopie - ripening - maturity - multispectral imagery - chemical composition - fruits - food technology - chemical compounds - imaging spectroscopy
In a greyvalue (black&white) image, a pixel contains a single value, representing light intensity. In an RGB color image, a pixel contains three values, corresponding with the light intensity at the red, green and blue band of the electromagnetic spectrum. In a spectral image, each pixel consists of an array of intensity values corresponding with small bands (
FLORES : identifying flowers by image content
Zedde, H.J. van de; Heijden, G.W.A.M. van der; Keizer, L.C.P. - \ 2007
multispectrale beelden - spectraalanalyse - principale componentenanalyse - kwaliteit - kenmerken - rozen - beeldvormende spectroscopie - patroonherkenning - uitwendige kenmerken - multispectral imagery - spectral analysis - principal component analysis - quality - traits - roses - imaging spectroscopy - pattern recognition - external traits
For auctions and plant variety testing, flowers need to be identified and compared. This is typically done by an expert. We try to develop a system to automatically compare an image of a flower with stored images of known varieties and retrieve the most similar ones. Spectral imaging allows calibration, making flower color invariant of recording equipment. Using simple similarity measures, spectral images of flowers proved to be superior to RGB images. The aim is to develop feature descriptors and similarity measures that will further increase the precision and recall of FLORES