Multiple Endmember Spectral Mixture Analysis (MESMA) for Dryland Applications
|
|
- Harry Leonard
- 7 years ago
- Views:
Transcription
1 Multiple Endmember Spectral Mixture Analysis (MESMA) for Dryland Applications M. Bachmann Hyper-I-Net 1st School on Hyperspectral Imaging October 29-31, 2007 Cáceres, Spain
2 Tutorial Structure Recapitulation: Basics of Spectral Unmixing (see talk by José Bioucas) MESMA Basic principle Different approaches Specific parameters Application Example Mapping Ground Cover Fractions in Drylands Methodological approach Results Conclusions Recommended Reading Folie 2
3 Spectral Unmixing Basics Linear Spectral Unmixing Subpixel classifier / subpixel detection i.e., objects smaller than one pixel can be detected Fuzzy classifier i.e., one pixel can belong to multiple classes Quantitative classifier i.e., results are area coverage of materials. Unit: % of pixel area [m 2 ] Physically based model - no statistical approach! i.e., (approx.) linear relationship between reflectance value of a pixel and the area of materials within this pixel Folie 3
4 Spectral Unmixing Basics Causes of spectral mixtures: Object borders Discrete objects Intimate material mixture Shading Adjacency effects PSF effects Multiple scattering not all can be addressed by linear spectral unmixing! Folie 4
5 Spectral Unmixing Basics ρ = ρ * ρ + ρ measured material_ 1 f material_1+ material_2* fmaterial_2+... material_ n* f material_ n Where a mn : reflectance of EM n in band m b m : measured reflectance in band m x m : abundance for EM n Where A: m*n EM-matrix x: abundance vector for n EM b: measured spectrum in m bands if A quadratic (i.e., m=n): Using hyperspectral data: normally overdetermined (m >> n), thus solving by least-squares approximation, e.g. Pseudo-Inverse (Moore-Penrose-inversion) where Folie 5
6 Spectral Unmixing Basics Since not all parameters are known exactly and can be included: introducing error term e: and minimizing error function F in least-squares sense: The RMS error of the mixing model is: i.e., the difference between measured signal (b) and modelled signal (b*) over all bands. Residual spectrum: bandwise difference r i = b i - b i * Folie 6
7 Spectral Unmixing Basics Matrix inversion may result in numerical problems! Linear dependencies between spectrally similar classes Higher number of EM higher probability of linear dependencies Result: ill-conditioned problem Thus check: Condition number of EM matrix A: where denotes euklidian L2-norn Κ >> 1: ill-conditioned problem Eigenvalues 0 of EM correlation matrix DET(A T A) 0 Folie 7
8 MESMA Introduction Multiple Endmember Spectral Mixture Analysis Linear spectral unmixing approach where particular EMs used to model a pixel number of EMs to be used vary on a per-pixel basis Folie 8
9 MESMA Introduction Change in lithology & soil types RMS, Fixed EM RMS, MESMA Folie 9
10 MESMA Basic Principle Linear spectral mixture model: ρ = ρ * f + ρ * f ρ measured material_ 1 material_1 material_2 material_2 material_ n* f material_ n Unmixing: solve overdetermined linear equation system for using constrained least-squares approaches (non-negative, sum-to-one) f material_1... n If we don t know which materials are in the pixel, why not include all EM spectra in one model? Linear dependencies between EM spectra (ill-conditioned EM-matrix) (Intrinsic) data dimensionality restricts number of EMs Folie 10
11 MESMA Basic Principle Example using 3 EM-Classes: - PV: photosynthetic active vegetation - NPV: non-photosynthetic active veg. - Soil: soils & rocks Thus using MESMA: only a small number of EMs are used at the same time, but EM can vary. 2 approaches: Unmix using all combinations of 2,3 n EMs & select best-fitting model or: Group EM into (thematic) classes and vary EM within these classes Folie 11
12 MESMA Basic Principle PV: photosynthetic active vegetation NPV: non-photosynthetic active veg. Soil: soils & rocks MESMA assuming 3 classes PV, NPV & Soil per pixel with l, m, and n EM 1st model: EM_PV-1, EM_NPV-1, EM_Soil-1 2nd model: EM_PV-1, EM_NPV-1, EM_Soil-2 nth model: EM_PV-1, EM_NPV-1, EM_Soil-n n+1 th model: EM_PV-1, EM_NPV-2, EM_Soil-1 2*n th model: EM_PV-1, EM_NPV-2, EM_Soil-n 2n+1 th model: EM_PV-1, EM_NPV-3, EM_Soil-1 m*n th model: EM_PV-1, EM_NPV-m, EM_Soil-n m*n+1 th model: EM_PV-2, EM_NPV-m, EM_Soil-n l*m*n th model: EM_PV-l, EM_NPV-m, EM_Soil-n Folie 12
13 MESMA Parameters How many EM models i.e., how many EM classes? and how many EM per class? Large numer of EM: more materials & spectral variability considered. But: calculation time increases Large numer of EM classes: more parts of image can be modeled / higher thematical content But: more problems with linear dependencies between EM Unmixing with large number of EM classes will (almost always) have smaller RMSE But: reasonable or only mathematical optimization? Test: if plenty of pixels have small EM-fractions (<10% abundance) => likely to be only mathematical optimization Folie 13
14 MESMA Parameters Thus: use as few EM classes as possible Calculate with n and n+1 EM-classes, analyze RMSE & then select model per pixel Or: fixed nummer of EM-classes & use algorithm like BVLS which can exclude EM classes Depending on application / study area typically EM classes for urban applications: Pervious Impervious ( Shade), PV NPV Roofs Other for dryland applications: Green Veg Dry Veg Soils ( Shade) Thematically reasonable number of classes In total about EM models with 3-15 EM per class Folie 14
15 MESMA Parameters Model Selection Criteria: EM model with minimum RMSE same with 0.95 f same with no residual deviation in the same direction in more than n successive bands (n ~ 7) Automated residual analysis Check residual spectra for diagnostic absorption features Identify & parameterize these features Take advantage of spectroscopic data! Folie 15
16 Selecting EM for MESMA Derivation of Image EM: As for all unmixing approaches (see papers by PLAZA et al.) First step: selection of spectrally extreme pixels using AMEE, SMACC, 2nd step: identification of these EM spectra Selection of EM for MESMA: 3rd step: selection of a suitable EM subset: Datatests like Class / EM Average RMSE (CAR, EAR) (DENNISON & ROBERTS) EM which model most (subset) pixels Manual interpretation & selection of EM candidates, field knowledge Recently no optimal automated & data-driven solution exists 4th step depending on MESMA approach: grouping of suitable EM into EM classes Folie 16
17 MESMA Published Approaches I MESMA Approaches for EM model selection Brute Force (i.e., calculation of all EM combinations) Pro: best EM model guaranteed Con: time consuming Monte Carlo approach (ASNER et al., randomly selecting EM from bundles of similar spectra) Pro: fast Con: monte carlo approach possible selection of unsuitable EM model Iterative EM selection based on residual analysis (BACHMANN et al.) Pro: fast (7x), adjusted to improved model selection criteria Con: decreased accuracy (by 2-7% abundance absolute) Folie 17
18 MESMA Published Approaches II MESMA Approaches for EM model selection Cont Neighborhood to crisp classes (SEGL et al.) Pro: fast, good probability for best EM model Con: existence of pure pixels, depending on crisp pre-classification Colinearity factor (GARCIA-HARO et al., pre-selection of promising EM models) Pro: fast Con: - Note: final EM selection is done brute force Segment-based EM model selection as a modification of these approaches Folie 18
19 Application Example Ground Cover Fractions Folie 19
20 Application Example Ground Cover Fractions Geographical Background Main study area: National Parc Cabo de Gata, Province Almeria, SE Spain Small-scaled mosaic of vegetation patches (tussock and annual grasses, palms & small bushes) and bare soil / rubble areas. Various vegetation conditions (green dry dead plants & plant parts) Type, degree and distribution of ground cover define degradation potential, and indicate disturbance & dangers. Thus, cover-percentage is a frequent parameter in land degradation models. Folie 20
21 Main Testsite: ational Parc Cabo de Gata, Province Almeria, SE Spain Folie 21
22 Application Example Ground Cover Fractions Database: HyMap-Datasets System corrected to at-sensor radiance Geocoded using ORTHO Atmospheric correction, terrain correction & empirical BRDFcorrection using ATCOR Field Spectra Field measurements of ground cover Folie 22
23 Spectral Variability in Dryland Vegetation Folie 23
24 EM Derivation & Selection Hyperspectral Data (atm., terrain & BRDF corrected) Image EM derived by two-stage approach SMACC (Sequential max. angle convex cone) on masked image Masking (classes, bad pixels) Optional: Additional generalized EM SMACC EM-Measures: EAR, myear, CAR, SID Collinearity & Condition of EMs Spectral Identification Scene- Specific EM Manual Check Prozess Additional EM after 1st unmixing iteration (like IEA) EM Selection Subprocess Optional (Manual) Subprocess... In simulations: ~70% of all EM could be automatically retrieved But: what are EM in reality? Spectral Identification Manual Check Additional EM Candidates 1. EM-Set Parametrization of EMs 1st Unmixing Iteration EM-Measures: EAR, myear, CAR, SID Hyperspectral Data (atm., terrain & BRDF corrected) Collinearity & Condition of EMs EM Selection Final EM-Set Hyperspectral Data 2nd & 3rd Unmixing (atm., terrain & BRDF Folie 24 Institut für Methodik Iteration corrected) der Fernerkundung bzw. Deutsches Fernerkundungsdatenzentrum
25 Iterative MESMA Parameterization of all EM spectra (diagnostic absorption bands related to bio-/geophysical properties, e.g., 2.2µm) Select new PV-EM, NPV-EM, Soil-EM Unmixing using current EM-Model Calculate model RMSE and abundances Check for physically unrealistic abundances Suggest potentially better PV-EM, NPV-EM, Soil-EM Residual analysis: identification & parameterization of absorption features (e.g., over- / under-determination of clay) Next iteration: select EM with desired properties (e.g., more / less clay) Yes Use current abundances as preliminary model Identify spectral features in residuum Calculate combined error score (based on weighted RMSE, residual features, unrealistic abundances) Test, if current error score is lower than preliminary error score Test, if a potentially better EM- Set exists (based on residual analysis) No Keep preliminary model Yes No Use preliminary result as final Folie 25 result. Continue with next pixel
26 Inclusion of Spatial Neighborhood Information 3 MESMA Iterations Unconstrained Unmixing & Residual Analysis Pixel properly modeled? No Current EM-Set Classify spectrum & check, if new EM EM Determination Iteration Constrained Unmixing & Residual Analysis Improved EM-Set Main Iteration Without / With eighborhood Iteration Modal Soil-EM selected in neighborhood? No Constrained Unmixing & Residual Analysis Neighborhood Iteration Check, if model only slightly detoriated Folie 26
27 Influence of View Angle Effects Local Incidence Angle Transition to Rough Terrain Maximum Underestimation of Soil [%] Influence of Local Incidence Angle Soil Abundances Local Incidence Angle [ ] without correction Folie 27 with empirical correction
28 Reliability Measure Linear Spectral Unmixing already offers measure for goodness of fit, i.e. the model RMSE Improved detection of pixels which are likely error-prone based on: Residual analysis & model RMSE Empirical regression model between cover and band indices Local incidence angle L2 data quality flags (from pre-processing) Folie 28
29 Results I HyMap (true colour) Soil EM (areas with soil abundance <30% masked out)) Soil EM spectra Soil abundance [%] NPV abundance [%] Folie 29
30 Results II Unmixing Abundance Sim. Model Nr. Reference Abundance Abundance error - PV Abundance error - NPV Abundance error per single simulation model Abundance error - Soil Histogram abundance error NPV Folie 30
31 Results III Simulations: Mean error: 5% - 10% abundance absolute (depending on scenario), R 2 between 0.65 and 0.85 (significance p < ) 50% of simulated models with error <3% abundance absolute, single errors up to 60% abundance absolute Contribution of errors caused by EMs: 60% - 80% Real-world data: Mean error: 9.6% abundance absolute, R 2 between 0.76 and 0.86 (significance p < 0.005), single errors up to 20% abundance absolute Literature values for MESMA: Mean errors: 5 15% abundance absolute, depending on applications & methods Folie 31
32 Results IV Factors influencing the unmixing accuracy: MESMA (~4% abundance absolute, 30% - 50% relative) EM selection (~4% abundance absolute, 20-40%% relative, might significantly increase if wrong EM) Model selection criterion (~3% abundance absolute, 20-25% relative, important when not all EM were found) Solving algorithm / constraints (up to 20%) Empirical correction of local view angle effects (~2% abundance absolute, 20% relative) Shading (<2% abundance absolute, <10% relative) but these numbers are case-specific! Folie 32
33 MESMA Summary When to use MESMA? Heterogeneous scenes Large number of materials to be mapped High spectral variability within material classes No fast / real-time processing is required Why use MESMA? Increased unmixing accuracy High number of EM possible Spectral variability explicitly modeled Appropriate EM for each pixel are automatically selected Reduced problems with EM collinearity thus (often) stable mixing model Folie 33
34 Selected References on MESMA: ROBERTS, D.A. ; GARDNER, M. ; CHURCH, R. ; USTIN, S. ; SCHEER, G. ; GREEN, R.O.: Mapping Chaparral in the Santa Monica Mountains Using Multiple Endmember Spectral Mixture Models. In: Remote Sensing of Environment 65 (1998), S DENNISON, P.E. ; HALLIGAN, K.Q. ; ROBERTS, D.A.: A Comparison of Error Metrics and Constraints for Multiple Endmember Spectral Mixture Analysis and Spectral Angle Mapper. In: Remote Sensing of Environment 93 (2004), S BALLANTINE, J.-A. ; OKIN, G.S. ; PRENTISS, D.E. ; ROBERTS, D.A.: Mapping North African Landforms Using Continental Scale Unmixing of MODIS Imagery. In: Remote Sensing of Environment 97 (2005), S ASNER, G.P. ; LOBELL, D.B.: A Biogeophysical Approach for Automated SWIR Unmixing of Soils and Vegetation. In: Remote Sensing of Environment 74 (2000), S ASNER, G.P. ; ELMOR, A.J. ; HUGES, R.F. ; WARNER, A.S. ; VITOUSEK, P.M.: Ecosystem Structure along Bioclimatic Gradients in Hawaii from Imaging Spectroscopy. In: Remote Sensing of Environment 96 (2005), S SEGL, K. ; ROESSNER, S. ; HEIDEN, U. ; KAUFMANN, H.: Fusion of Spectral and Shape Features for Identification of Urban Surface Cover Types Using Reflective and Thermal Hyperspectral Data. In: ISPRS Journal of Photogrammetry & Remote Sensing 58 (2003), S GARCIA-HARO, F. ; SOMMER, S. ; KEMPER, T.: A New Tool for Variable Multiple Endmember Spectral Mixture Analysis (VMESMA). In: International Journal of Remote Sensing 26 (2005), Nr. 10, S BACHMANN, M.; MÜLLER, A.; HABERMEYER, M.; SCHMIDT, M.; DECH, S.: Iterative MESMA Unmixing for Fractional Cover Estimates - Evaluating the Portability. In: Proceedings of the 4th EARSeL Workshop on Imaging Spectroscopy. Warsaw, 2005 Folie 34
35 Vacant Hyper-I-Net Position: ESR2 - Improved hyperspectral sensor design Topic: Design & implementation of a modular sensor simulation tool. Incl. spectral, radiometric, and geometric instrument characterization. Background: Degree in Physics, Meteorology or Engineering Sciences Institutions: German Aerospace Center (DLR), Oberpfaffenhofen, Germany Kayser-Threde GmbH (KT), Munich, Germany. Contact: Andreas Müller Stefan Hofer Rudolf.Richter@dlr.de Folie 35
36 Folie 36
37 MESMA Parameters Red line: reference soil spectrum Green line: same soil with clay absorption (abs. depth 4% reflectance absolute) Black line: reduced overall albedo by 1% absolute Differences Red-Green: 0.3% of overall albedo Difference Red-Black: 2.9% of overall albedo Folie 37
38 MESMA Approaches Overview Folie 38
39 Shade Component Shade component Not necessary in MESMA since EM with different overall brightness can be used Flat shade spectrum is oversimplified! Ratio direct / diffuse irradiation and/or leaf transmission spectra are required (=> blue XXX) Or: radiative transfer model in order to calculate shaded surfaces incl. multiple scattering effects Inclusion of a flat shade component may result in an ill-conditioned mixing model When using shade component: Better: Check that spatial distribution of shade-em (mostly) corresponds to terrain patterns! Terrain correction to remove topograhic illumination effects Image EM implicitly include canopy shade & multple scattering effects Simulation of (partially) shaded soil-ems (ADLER-GOLDEN for SAM) Then unmixing without shade Or: normalization of image & EM Folie 39
A Quantitative and Comparative Analysis of Endmember Extraction Algorithms From Hyperspectral Data
650 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 42, NO. 3, MARCH 2004 A Quantitative and Comparative Analysis of Endmember Extraction Algorithms From Hyperspectral Data Antonio Plaza, Pablo
More informationData Processing Developments at DFD/DLR. Stefanie Holzwarth Martin Bachmann, Rudolf Richter, Martin Habermeyer, Derek Rogge
Data Processing Developments at DFD/DLR Stefanie Holzwarth Martin Bachmann, Rudolf Richter, Martin Habermeyer, Derek Rogge EUFAR Joint Expert Working Group Meeting Edinburgh, April 14th 2011 Conclusions
More informationIntegrating Airborne Hyperspectral Sensor Data with GIS for Hail Storm Post-Disaster Management.
Integrating Airborne Hyperspectral Sensor Data with GIS for Hail Storm Post-Disaster Management. *Sunil BHASKARAN, *Bruce FORSTER, **Trevor NEAL *School of Surveying and Spatial Information Systems, Faculty
More informationCROP CLASSIFICATION WITH HYPERSPECTRAL DATA OF THE HYMAP SENSOR USING DIFFERENT FEATURE EXTRACTION TECHNIQUES
Proceedings of the 2 nd Workshop of the EARSeL SIG on Land Use and Land Cover CROP CLASSIFICATION WITH HYPERSPECTRAL DATA OF THE HYMAP SENSOR USING DIFFERENT FEATURE EXTRACTION TECHNIQUES Sebastian Mader
More informationUsing Remote Sensing Imagery to Evaluate Post-Wildfire Damage in Southern California
Graham Emde GEOG 3230 Advanced Remote Sensing February 22, 2013 Lab #1 Using Remote Sensing Imagery to Evaluate Post-Wildfire Damage in Southern California Introduction Wildfires are a common disturbance
More information2.3 Spatial Resolution, Pixel Size, and Scale
Section 2.3 Spatial Resolution, Pixel Size, and Scale Page 39 2.3 Spatial Resolution, Pixel Size, and Scale For some remote sensing instruments, the distance between the target being imaged and the platform,
More informationANALYSIS OF AVIRIS DATA: A COMPARISON OF THE PERFORMANCE OF COMMERCIAL SOFTWARE WITH PUBLISHED ALGORITHMS. William H. Farrand 1
ANALYSIS OF AVIRIS DATA: A COMPARISON OF THE PERFORMANCE OF COMMERCIAL SOFTWARE WITH PUBLISHED ALGORITHMS William H. Farrand 1 1. Introduction An early handicap to the effective use of AVIRIS data was
More informationTHE SPECTRAL DIMENSION IN URBAN LAND COVER MAPPING FROM HIGH - RESOLUTION OPTICAL REMOTE SENSING DATA *
THE SPECTRAL DIMENSION IN URBAN LAND COVER MAPPING FROM HIGH - RESOLUTION OPTICAL REMOTE SENSING DATA * Martin Herold 1, Meg Gardner 1, Brian Hadley 2 and Dar Roberts 1 1 Department of Geography, University
More informationy = Xβ + ε B. Sub-pixel Classification
Sub-pixel Mapping of Sahelian Wetlands using Multi-temporal SPOT VEGETATION Images Jan Verhoeye and Robert De Wulf Laboratory of Forest Management and Spatial Information Techniques Faculty of Agricultural
More informationDigital Remote Sensing Data Processing Digital Remote Sensing Data Processing and Analysis: An Introduction and Analysis: An Introduction
Digital Remote Sensing Data Processing Digital Remote Sensing Data Processing and Analysis: An Introduction and Analysis: An Introduction Content Remote sensing data Spatial, spectral, radiometric and
More informationIntroduction to Hyperspectral Image Analysis
Introduction to Hyperspectral Image Analysis Background Peg Shippert, Ph.D. Earth Science Applications Specialist Research Systems, Inc. The most significant recent breakthrough in remote sensing has been
More informationMultiscale Object-Based Classification of Satellite Images Merging Multispectral Information with Panchromatic Textural Features
Remote Sensing and Geoinformation Lena Halounová, Editor not only for Scientific Cooperation EARSeL, 2011 Multiscale Object-Based Classification of Satellite Images Merging Multispectral Information with
More informationComparison of ALOS-PALSAR and TerraSAR-X Data in terms of Detecting Settlements First Results
ALOS 2008 Symposium, 3-7 November Rhodes, Greece Comparison of ALOS-PALSAR and TerraSAR-X Data in terms of Detecting Settlements First Results Thomas Esch*, Achim Roth*, Michael Thiel, Michael Schmidt*,
More informationState of the Urban Forest: San Francisco Bay Area- Progress Report
State of the Urban Forest: San Francisco Bay Area- Progress Report Jim Simpson, Greg McPherson, Chad Delany Center for Urban Forest Research USDA Forest Service, PSW Research Station Davis, CA June 20,
More informationTowards agreed data quality layers for airborne hyperspectral imagery
Towards agreed data quality layers for airborne hyperspectral imagery M. Bachmann, DLR M. Bachmann, DLR, S. Adar, TAU; E. Ben-Dor, TAU; J. Biesemans, VITO; X. Briottet, ONERA; M. Grant, PML; J. Hanus,
More informationSAMPLE MIDTERM QUESTIONS
Geography 309 Sample MidTerm Questions Page 1 SAMPLE MIDTERM QUESTIONS Textbook Questions Chapter 1 Questions 4, 5, 6, Chapter 2 Questions 4, 7, 10 Chapter 4 Questions 8, 9 Chapter 10 Questions 1, 4, 7
More informationENVI Classic Tutorial: Atmospherically Correcting Multispectral Data Using FLAASH 2
ENVI Classic Tutorial: Atmospherically Correcting Multispectral Data Using FLAASH Atmospherically Correcting Multispectral Data Using FLAASH 2 Files Used in this Tutorial 2 Opening the Raw Landsat Image
More informationSelecting the appropriate band combination for an RGB image using Landsat imagery
Selecting the appropriate band combination for an RGB image using Landsat imagery Ned Horning Version: 1.0 Creation Date: 2004-01-01 Revision Date: 2004-01-01 License: This document is licensed under a
More informationCollaborative research project pre agro
Collaborative research project pre agro Extraction of phenology-dependent structural information from hyperspectral, directional CHRIS data for a better derivation of canopy parameters of winter-wheat
More informationCalculation of Minimum Distances. Minimum Distance to Means. Σi i = 1
Minimum Distance to Means Similar to Parallelepiped classifier, but instead of bounding areas, the user supplies spectral class means in n-dimensional space and the algorithm calculates the distance between
More informationTerraSAR-X Interferometry. Michael Eineder, Nico Adam Remote Sensing Technology Institute
TerraSAR-X Interferometry Michael Eineder, Nico Adam Remote Sensing Technology Institute TerraSAR-X Contribution to Commissioning Phase: verify phase and geometric stability of instrument and SAR processor
More informationWATER BODY EXTRACTION FROM MULTI SPECTRAL IMAGE BY SPECTRAL PATTERN ANALYSIS
WATER BODY EXTRACTION FROM MULTI SPECTRAL IMAGE BY SPECTRAL PATTERN ANALYSIS Nguyen Dinh Duong Department of Environmental Information Study and Analysis, Institute of Geography, 18 Hoang Quoc Viet Rd.,
More informationCloud Masking and Cloud Products
Cloud Masking and Cloud Products MODIS Operational Algorithm MOD35 Paul Menzel, Steve Ackerman, Richard Frey, Kathy Strabala, Chris Moeller, Liam Gumley, Bryan Baum MODIS Cloud Masking Often done with
More informationUse of hyperspectral data for deriving vegetation types in Savannahs in Central Namibia
Use of hyperspectral data for deriving vegetation types in Savannahs in Central Namibia Lena Lieckfeld a*, Jens Oldeland b, Bettina Weber, Christoph Schultz c, Andreas Müller a, Michael Schmidt a, Norbert
More informationCrop Drought Stress Monitoring by Remote Sensing (DROSMON) Overview. Werner Schneider
Crop Drought Stress Monitoring by Remote Sensing (DROSMON) Overview Werner Schneider Institut of Surveying, Remote Sensing and Land Information Department of Landscape, Spatial and Infrastructure Sciences
More informationAdaptive HSI Data Processing for Near-Real-time Analysis and Spectral Recovery *
Adaptive HSI Data Processing for Near-Real-time Analysis and Spectral Recovery * Su May Hsu, 1 Hsiao-hua Burke and Michael Griffin MIT Lincoln Laboratory, Lexington, Massachusetts 1. INTRODUCTION Hyperspectral
More informationMonitoring Soil Moisture from Space. Dr. Heather McNairn Science and Technology Branch Agriculture and Agri-Food Canada heather.mcnairn@agr.gc.
Monitoring Soil Moisture from Space Dr. Heather McNairn Science and Technology Branch Agriculture and Agri-Food Canada heather.mcnairn@agr.gc.ca What is Remote Sensing? Scientists turn the raw data collected
More informationSLC-off Gap-Filled Products Gap-Fill Algorithm Methodology
SLC-off Gap-illed roducts Gap-ill Algorithm Methodology Background The U.S. Geological Survey (USGS) Earth Resources Observation Systems (EROS) Data Center (EDC) has developed multi-scene (same path/row)
More informationEnvironmental Remote Sensing GEOG 2021
Environmental Remote Sensing GEOG 2021 Lecture 4 Image classification 2 Purpose categorising data data abstraction / simplification data interpretation mapping for land cover mapping use land cover class
More informationCOMPARISON OF OBJECT BASED AND PIXEL BASED CLASSIFICATION OF HIGH RESOLUTION SATELLITE IMAGES USING ARTIFICIAL NEURAL NETWORKS
COMPARISON OF OBJECT BASED AND PIXEL BASED CLASSIFICATION OF HIGH RESOLUTION SATELLITE IMAGES USING ARTIFICIAL NEURAL NETWORKS B.K. Mohan and S. N. Ladha Centre for Studies in Resources Engineering IIT
More informationSupervised Classification workflow in ENVI 4.8 using WorldView-2 imagery
Supervised Classification workflow in ENVI 4.8 using WorldView-2 imagery WorldView-2 is the first commercial high-resolution satellite to provide eight spectral sensors in the visible to near-infrared
More informationAPPLICATION OF TERRA/ASTER DATA ON AGRICULTURE LAND MAPPING. Genya SAITO*, Naoki ISHITSUKA*, Yoneharu MATANO**, and Masatane KATO***
APPLICATION OF TERRA/ASTER DATA ON AGRICULTURE LAND MAPPING Genya SAITO*, Naoki ISHITSUKA*, Yoneharu MATANO**, and Masatane KATO*** *National Institute for Agro-Environmental Sciences 3-1-3 Kannondai Tsukuba
More informationADVANTAGES AND DISADVANTAGES OF THE HOUGH TRANSFORMATION IN THE FRAME OF AUTOMATED BUILDING EXTRACTION
ADVANTAGES AND DISADVANTAGES OF THE HOUGH TRANSFORMATION IN THE FRAME OF AUTOMATED BUILDING EXTRACTION G. Vozikis a,*, J.Jansa b a GEOMET Ltd., Faneromenis 4 & Agamemnonos 11, GR - 15561 Holargos, GREECE
More informationLectures Remote Sensing
Lectures Remote Sensing ATMOSPHERIC CORRECTION dr.ir. Jan Clevers Centre of Geo-Information Environmental Sciences Wageningen UR Atmospheric Correction of Optical RS Data Background When needed? Model
More informationCoarse Resolution Image Analysis and Cleaning UpCloud Radiation
This article was downloaded by:[university of Tokyo/TOKYO DAIGAKU] On: 3 March 2008 Access Details: [subscription number 778576937] Publisher: Taylor & Francis Informa Ltd Registered in England and Wales
More informationMODIS IMAGES RESTORATION FOR VNIR BANDS ON FIRE SMOKE AFFECTED AREA
MODIS IMAGES RESTORATION FOR VNIR BANDS ON FIRE SMOKE AFFECTED AREA Li-Yu Chang and Chi-Farn Chen Center for Space and Remote Sensing Research, National Central University, No. 300, Zhongda Rd., Zhongli
More informationResolutions of Remote Sensing
Resolutions of Remote Sensing 1. Spatial (what area and how detailed) 2. Spectral (what colors bands) 3. Temporal (time of day/season/year) 4. Radiometric (color depth) Spatial Resolution describes how
More informationModelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches
Modelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches PhD Thesis by Payam Birjandi Director: Prof. Mihai Datcu Problematic
More informationDevelopment of Method for LST (Land Surface Temperature) Detection Using Big Data of Landsat TM Images and AWS
Development of Method for LST (Land Surface Temperature) Detection Using Big Data of Landsat TM Images and AWS Myung-Hee Jo¹, Sung Jae Kim², Jin-Ho Lee 3 ¹ Department of Aeronautical Satellite System Engineering,
More informationAdvanced Computer Graphics. Rendering Equation. Matthias Teschner. Computer Science Department University of Freiburg
Advanced Computer Graphics Rendering Equation Matthias Teschner Computer Science Department University of Freiburg Outline rendering equation Monte Carlo integration sampling of random variables University
More informationPHOTOGRAMMETRIC TECHNIQUES FOR MEASUREMENTS IN WOODWORKING INDUSTRY
PHOTOGRAMMETRIC TECHNIQUES FOR MEASUREMENTS IN WOODWORKING INDUSTRY V. Knyaz a, *, Yu. Visilter, S. Zheltov a State Research Institute for Aviation System (GosNIIAS), 7, Victorenko str., Moscow, Russia
More informationP.M. Rich, W.A. Hetrick, S.C. Saving Biological Sciences University of Kansas Lawrence, KS 66045
USING VIEWSHED MODELS TO CALCULATE INTERCEPTED SOLAR RADIATION: APPLICATIONS IN ECOLOGY by P.M. Rich, W.A. Hetrick, S.C. Saving Biological Sciences University of Kansas Lawrence, KS 66045 R.O. Dubayah
More informationData Processing Flow Chart
Legend Start V1 V2 V3 Completed Version 2 Completion date Data Processing Flow Chart Data: Download a) AVHRR: 1981-1999 b) MODIS:2000-2010 c) SPOT : 1998-2002 No Progressing Started Did not start 03/12/12
More informationSome elements of photo. interpretation
Some elements of photo Shape Size Pattern Color (tone, hue) Texture Shadows Site Association interpretation Olson, C. E., Jr. 1960. Elements of photographic interpretation common to several sensors. Photogrammetric
More information1. Theoretical background
1. Theoretical background We consider the energy budget at the soil surface (equation 1). Energy flux components absorbed or emitted by the soil surface are: net radiation, latent heat flux, sensible heat
More informationReview for Introduction to Remote Sensing: Science Concepts and Technology
Review for Introduction to Remote Sensing: Science Concepts and Technology Ann Johnson Associate Director ann@baremt.com Funded by National Science Foundation Advanced Technological Education program [DUE
More informationA KNOWLEDGE-BASED APPROACH FOR REDUCING CLOUD AND SHADOW ABSTRACT
A KNOWLEDGE-BASED APPROACH FOR REDUCING CLOUD AND SHADOW Mingjun Song, Graduate Research Assistant Daniel L. Civco, Director Laboratory for Earth Resources Information Systems Department of Natural Resources
More informationDevelopment of a remote sensing based land barometer for sustainable land use management
Bilateral German-Polish REFINA workshop Research and model projects on suburbanization and land consumption Development of a remote sensing based land barometer for sustainable land use management Thomas
More informationMyths and misconceptions about remote sensing
Myths and misconceptions about remote sensing Ned Horning (graphics support - Nicholas DuBroff) Version: 1.0 Creation Date: 2004-01-01 Revision Date: 2004-01-01 License: This document is licensed under
More informationPATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION
PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION Introduction In the previous chapter, we explored a class of regression models having particularly simple analytical
More informationTerraSAR X and TanDEM X satellite missions update & other activities Dana Floricioiu German Aerospace Center (DLR), Remote Sensing Technology
TerraSAR X and TanDEM X satellite missions update & other activities Dana Floricioiu German Aerospace Center (DLR), Remote Sensing Technology Institute, Oberpfaffenhofen, Germany PSTG 2 12 14 June 2012
More informationIntroduction to Imagery and Raster Data in ArcGIS
Esri International User Conference San Diego, California Technical Workshops July 25, 2012 Introduction to Imagery and Raster Data in ArcGIS Simon Woo slides Cody Benkelman - demos Overview of Presentation
More informationLiDAR for vegetation applications
LiDAR for vegetation applications UoL MSc Remote Sensing Dr Lewis plewis@geog.ucl.ac.uk Introduction Introduction to LiDAR RS for vegetation Review instruments and observational concepts Discuss applications
More informationImpact of sensor s point spread function on land cover characterization: assessment and deconvolution
Remote Sensing of Environment 80 (2002) 203 212 www.elsevier.com/locate/rse Impact of sensor s point spread function on land cover characterization: assessment and deconvolution Chengquan Huang a, *, John
More informationExtraction of Satellite Image using Particle Swarm Optimization
Extraction of Satellite Image using Particle Swarm Optimization Er.Harish Kundra Assistant Professor & Head Rayat Institute of Engineering & IT, Railmajra, Punjab,India. Dr. V.K.Panchal Director, DTRL,DRDO,
More informationReceived in revised form 24 March 2004; accepted 30 March 2004
Remote Sensing of Environment 91 (2004) 237 242 www.elsevier.com/locate/rse Cloud detection in Landsat imagery of ice sheets using shadow matching technique and automatic normalized difference snow index
More informationCLASSIFICATION ACCURACY INCREASE USING MULTISENSOR DATA FUSION
CLASSIFICATION ACCURACY INCREASE USING MULTISENSOR DATA FUSION Aliaksei Makarau, Gintautas Palubinskas, and Peter Reinartz German Aerospace Center (DLR) German Remote Sensing Data Center (DFD) bzw. Remote
More informationSimilarity-Based Unsupervised Band Selection for Hyperspectral Image Analysis
564 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 5, NO. 4, OCTOBER 2008 Similarity-Based Unsupervised Band Selection for Hyperspectral Image Analysis Qian Du, Senior Member, IEEE, and He Yang, Student
More informationImage Analysis CHAPTER 16 16.1 ANALYSIS PROCEDURES
CHAPTER 16 Image Analysis 16.1 ANALYSIS PROCEDURES Studies for various disciplines require different technical approaches, but there is a generalized pattern for geology, soils, range, wetlands, archeology,
More informationENVI Classic Tutorial: Classification Methods
ENVI Classic Tutorial: Classification Methods Classification Methods 2 Files Used in this Tutorial 2 Examining a Landsat TM Color Image 3 Reviewing Image Colors 3 Using the Cursor Location/Value 4 Examining
More informationDETECTION OF URBAN FEATURES AND MAP UPDATING FROM SATELLITE IMAGES USING OBJECT-BASED IMAGE CLASSIFICATION METHODS AND INTEGRATION TO GIS
Proceedings of the 4th GEOBIA, May 79, 2012 Rio de Janeiro Brazil. p.315 DETECTION OF URBAN FEATURES AND MAP UPDATING FROM SATELLITE IMAGES USING OBJECTBASED IMAGE CLASSIFICATION METHODS AND INTEGRATION
More informationHow to calculate reflectance and temperature using ASTER data
How to calculate reflectance and temperature using ASTER data Prepared by Abduwasit Ghulam Center for Environmental Sciences at Saint Louis University September, 2009 This instructions walk you through
More informationAn Assessment of the Effectiveness of Segmentation Methods on Classification Performance
An Assessment of the Effectiveness of Segmentation Methods on Classification Performance Merve Yildiz 1, Taskin Kavzoglu 2, Ismail Colkesen 3, Emrehan K. Sahin Gebze Institute of Technology, Department
More informationGraduated Student: José O. Nogueras Colón Adviser: Yahya M. Masalmah, Ph.D.
Graduated Student: José O. Nogueras Colón Adviser: Yahya M. Masalmah, Ph.D. Introduction Problem Statement Objectives Hyperspectral Imagery Background Grid Computing Desktop Grids DG Advantages Green Desktop
More informationDesign of a High Resolution Multispectral Scanner for Developing Vegetation Indexes
Design of a High Resolution Multispectral Scanner for Developing Vegetation Indexes Rishitosh kumar sinha*, Roushan kumar mishra, Sam jeba kumar, Gunasekar. S Dept. of Instrumentation & Control Engg. S.R.M
More informationValidating MOPITT Cloud Detection Techniques with MAS Images
Validating MOPITT Cloud Detection Techniques with MAS Images Daniel Ziskin, Juying Warner, Paul Bailey, John Gille National Center for Atmospheric Research, P.O. Box 3000, Boulder, CO 80307 ABSTRACT The
More informationLeast-Squares Intersection of Lines
Least-Squares Intersection of Lines Johannes Traa - UIUC 2013 This write-up derives the least-squares solution for the intersection of lines. In the general case, a set of lines will not intersect at a
More informationRemote Sensing of Environment
Remote Sensing of Environment 115 (2011) 1145 1161 Contents lists available at ScienceDirect Remote Sensing of Environment journal homepage: www.elsevier.com/locate/rse Per-pixel vs. object-based classification
More informationQuantifying Seasonal Variation in Cloud Cover with Predictive Models
Quantifying Seasonal Variation in Cloud Cover with Predictive Models Ashok N. Srivastava, Ph.D. ashok@email.arc.nasa.gov Deputy Area Lead, Discovery and Systems Health Group Leader, Intelligent Data Understanding
More informationVCS REDD Methodology Module. Methods for monitoring forest cover changes in REDD project activities
1 VCS REDD Methodology Module Methods for monitoring forest cover changes in REDD project activities Version 1.0 May 2009 I. SCOPE, APPLICABILITY, DATA REQUIREMENT AND OUTPUT PARAMETERS Scope This module
More informationMineral Exploration Using GIS and Processed Aster Images
Mineral Exploration Using GIS and Processed Aster Images Carlos A. Torres Advance GIS EES 6513 (Spring 2007) University of Texas at San Antonio Abstract The risks of developing mineral resources need to
More informationDigital image processing
746A27 Remote Sensing and GIS Lecture 4 Digital image processing Chandan Roy Guest Lecturer Department of Computer and Information Science Linköping University Digital Image Processing Most of the common
More informationIRS Level 2 Processing Concept Status
IRS Level 2 Processing Concept Status Stephen Tjemkes, Jochen Grandell and Xavier Calbet 6th MTG Mission Team Meeting 17 18 June 2008, Estec, Noordwijk Page 1 Content Introduction Level 2 Processing Concept
More informationUsing D2K Data Mining Platform for Understanding the Dynamic Evolution of Land-Surface Variables
Using D2K Data Mining Platform for Understanding the Dynamic Evolution of Land-Surface Variables Praveen Kumar 1, Peter Bajcsy 2, David Tcheng 2, David Clutter 2, Vikas Mehra 1, Wei-Wen Feng 2, Pratyush
More informationThe empirical line method for the atmospheric correction of IKONOS imagery
INT. J. REMOTE SENSING, 2003, VOL. 24, NO. 5, 1143 1150 The empirical line method for the atmospheric correction of IKONOS imagery E. KARPOUZLI* and T. MALTHUS Department of Geography, University of Edinburgh,
More informationA STRATEGY FOR ESTIMATING TREE CANOPY DENSITY USING LANDSAT 7 ETM+ AND HIGH RESOLUTION IMAGES OVER LARGE AREAS
A STRATEGY FOR ESTIMATING TREE CANOPY DENSITY USING LANDSAT 7 ETM+ AND HIGH RESOLUTION IMAGES OVER LARGE AREAS Chengquan Huang*, Limin Yang, Bruce Wylie, Collin Homer Raytheon ITSS EROS Data Center, Sioux
More informationSub-pixel mapping: A comparison of techniques
Sub-pixel mapping: A comparison of techniques Koen C. Mertens, Lieven P.C. Verbeke & Robert R. De Wulf Laboratory of Forest Management and Spatial Information Techniques, Ghent University, 9000 Gent, Belgium
More informationTerraColor White Paper
TerraColor White Paper TerraColor is a simulated true color digital earth imagery product developed by Earthstar Geographics LLC. This product was built from imagery captured by the US Landsat 7 (ETM+)
More informationRadiation Transfer in Environmental Science
Radiation Transfer in Environmental Science with emphasis on aquatic and vegetation canopy media Autumn 2008 Prof. Emmanuel Boss, Dr. Eyal Rotenberg Introduction Radiation in Environmental sciences Most
More informationMOD09 (Surface Reflectance) User s Guide
MOD09 (Surface ) User s Guide MODIS Land Surface Science Computing Facility Principal Investigator: Dr. Eric F. Vermote Web site: http://modis-sr.ltdri.org Correspondence e-mail address: mod09@ltdri.org
More informationThe USGS Landsat Big Data Challenge
The USGS Landsat Big Data Challenge Brian Sauer Engineering and Development USGS EROS bsauer@usgs.gov U.S. Department of the Interior U.S. Geological Survey USGS EROS and Landsat 2 Data Utility and Exploitation
More informationASSESSMENT OF FOREST RECOVERY AFTER FIRE USING LANDSAT TM IMAGES AND GIS TECHNIQUES: A CASE STUDY OF MAE WONG NATIONAL PARK, THAILAND
ASSESSMENT OF FOREST RECOVERY AFTER FIRE USING LANDSAT TM IMAGES AND GIS TECHNIQUES: A CASE STUDY OF MAE WONG NATIONAL PARK, THAILAND Sunee Sriboonpong 1 Yousif Ali Hussin 2 Alfred de Gier 2 1 Forest Resource
More informationA Short Introduction to Computer Graphics
A Short Introduction to Computer Graphics Frédo Durand MIT Laboratory for Computer Science 1 Introduction Chapter I: Basics Although computer graphics is a vast field that encompasses almost any graphical
More informationImproved predictive modeling of white LEDs with accurate luminescence simulation and practical inputs
Improved predictive modeling of white LEDs with accurate luminescence simulation and practical inputs TracePro Opto-Mechanical Design Software s Fluorescence Property Utility TracePro s Fluorescence Property
More informationIMPERVIOUS SURFACE MAPPING UTILIZING HIGH RESOLUTION IMAGERIES. Authors: B. Acharya, K. Pomper, B. Gyawali, K. Bhattarai, T.
IMPERVIOUS SURFACE MAPPING UTILIZING HIGH RESOLUTION IMAGERIES Authors: B. Acharya, K. Pomper, B. Gyawali, K. Bhattarai, T. Tsegaye ABSTRACT Accurate mapping of artificial or natural impervious surfaces
More informationPassive Remote Sensing of Clouds from Airborne Platforms
Passive Remote Sensing of Clouds from Airborne Platforms Why airborne measurements? My instrument: the Solar Spectral Flux Radiometer (SSFR) Some spectrometry/radiometry basics How can we infer cloud properties
More informationJava Modules for Time Series Analysis
Java Modules for Time Series Analysis Agenda Clustering Non-normal distributions Multifactor modeling Implied ratings Time series prediction 1. Clustering + Cluster 1 Synthetic Clustering + Time series
More informationAPPLICATION OF MULTITEMPORAL LANDSAT DATA TO MAP AND MONITOR LAND COVER AND LAND USE CHANGE IN THE CHESAPEAKE BAY WATERSHED
APPLICATION OF MULTITEMPORAL LANDSAT DATA TO MAP AND MONITOR LAND COVER AND LAND USE CHANGE IN THE CHESAPEAKE BAY WATERSHED S. J. GOETZ Woods Hole Research Center Woods Hole, Massachusetts 054-096 USA
More informationWeight of Evidence Module
Formula Guide The purpose of the Weight of Evidence (WoE) module is to provide flexible tools to recode the values in continuous and categorical predictor variables into discrete categories automatically,
More informationVIIRS-CrIS mapping. NWP SAF AAPP VIIRS-CrIS Mapping
NWP SAF AAPP VIIRS-CrIS Mapping This documentation was developed within the context of the EUMETSAT Satellite Application Facility on Numerical Weather Prediction (NWP SAF), under the Cooperation Agreement
More informationPredict the Popularity of YouTube Videos Using Early View Data
000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050
More informationRESOLUTION MERGE OF 1:35.000 SCALE AERIAL PHOTOGRAPHS WITH LANDSAT 7 ETM IMAGERY
RESOLUTION MERGE OF 1:35.000 SCALE AERIAL PHOTOGRAPHS WITH LANDSAT 7 ETM IMAGERY M. Erdogan, H.H. Maras, A. Yilmaz, Ö.T. Özerbil General Command of Mapping 06100 Dikimevi, Ankara, TURKEY - (mustafa.erdogan;
More informationMeasurement of the effect of biomass burning aerosol on inhibition of cloud formation over the Amazon
Supporting Online Material for Koren et al. Measurement of the effect of biomass burning aerosol on inhibition of cloud formation over the Amazon 1. MODIS new cloud detection algorithm The operational
More informationAnalysis of Landsat ETM+ Image Enhancement for Lithological Classification Improvement in Eagle Plain Area, Northern Yukon
Analysis of Landsat ETM+ Image Enhancement for Lithological Classification Improvement in Eagle Plain Area, Northern Yukon Shihua Zhao, Department of Geology, University of Calgary, zhaosh@ucalgary.ca,
More informationSOLAR RADIATION AND YIELD. Alessandro Massi Pavan
SOLAR RADIATION AND YIELD Alessandro Massi Pavan Sesto Val Pusteria June 22 nd 26 th, 2015 DEFINITIONS Solar radiation: general meaning Irradiation [Wh/m 2 ]: energy received per unit area Irradiance [W/m
More informationRealization of a UV fisheye hyperspectral camera
Realization of a UV fisheye hyperspectral camera Valentina Caricato, Andrea Egidi, Marco Pisani and Massimo Zucco, INRIM Outline Purpose of the instrument Required specs Hyperspectral technique Optical
More informationENVI Classic Tutorial: Atmospherically Correcting Hyperspectral Data using FLAASH 2
ENVI Classic Tutorial: Atmospherically Correcting Hyperspectral Data Using FLAASH Atmospherically Correcting Hyperspectral Data using FLAASH 2 Files Used in This Tutorial 2 Opening the Uncorrected AVIRIS
More informationIMAGINES_VALIDATIONSITESNETWORK ISSUE 1.00. EC Proposal Reference N FP7-311766. Name of lead partner for this deliverable: EOLAB
Date Issued: 26.03.2014 Issue: I1.00 IMPLEMENTING MULTI-SCALE AGRICULTURAL INDICATORS EXPLOITING SENTINELS RECOMMENDATIONS FOR SETTING-UP A NETWORK OF SITES FOR THE VALIDATION OF COPERNICUS GLOBAL LAND
More informationRULE INHERITANCE IN OBJECT-BASED IMAGE CLASSIFICATION FOR URBAN LAND COVER MAPPING INTRODUCTION
RULE INHERITANCE IN OBJECT-BASED IMAGE CLASSIFICATION FOR URBAN LAND COVER MAPPING Ejaz Hussain, Jie Shan {ehussain, jshan}@ecn.purdue.edu} Geomatics Engineering, School of Civil Engineering, Purdue University
More informationD.S. Boyd School of Earth Sciences and Geography, Kingston University, U.K.
PHYSICAL BASIS OF REMOTE SENSING D.S. Boyd School of Earth Sciences and Geography, Kingston University, U.K. Keywords: Remote sensing, electromagnetic radiation, wavelengths, target, atmosphere, sensor,
More information