GEOGG121: Methods Inversion I: linear approaches Dr. Mathias (Mat) Disney UCL Geography Office: 113, Pearson Building Tel: 7670 0592 Email: mdisney@ucl.geog.ac.uk www.geog.ucl.ac.uk/~mdisney Lecture outline • Linear models and inversion – Least squares revisited, examples – Parameter estimation, uncertainty – Practical examples • Spectral linear mixture models • Kernel-driven BRDF models and change detection Reading • Linear models and inversion – Linear modelling notes: Lewis, 2010 – Chapter 2 of Press et al. (1992) Numerical Recipes in C (online version http://apps.nrbook.com/c/index.html) – http://en.wikipedia.org/wiki/Linear_model – http://en.wikipedia.org/wiki/System_of_linear_equations Linear Models • For some set of independent variables x = {x0, x1, x2, … , xn} have a model of a dependent variable y which can be expressed as a linear combination of the independent variables. y a0 a1 x1 y a 0 a1 x1 a 2 x 2 y in a x i 0 i i y ax y ax Linear Models? i n y a 0 a i sin x i bi i n i 1 y a 0 a i sin x i bi cos x i i 1 y in a x i 0 i i 0 a 0 a1 x 0 a 2 x ... a n x 2 0 y a0 e a1 x n 0 Linear Mixture Modelling • Spectral mixture modelling: – Proportionate mixture of (n) end-member spectra r i 0 i Fi i n 1 i n 1 i 0 r F Fi 1 – First-order model: no interactions between components Linear Mixture Modelling • r = {rl0, rl1, … rlm, 1.0} – Measured reflectance spectrum (m wavelengths) • nx(m+1) matrix: rl 0 l0 0 rl1 l1 0 rlm 1 l m 1 0 1.0 1.0 l0 1 l1 0 l0 2 l1 2 lm 1 1 lm 1 21 1.0 1.0 r F l0 n 1 P0 l1 n 1 P1 l m 1 n 1 1.0 P 2 P n 1 Linear Mixture Modelling • n=(m+1) – square matrix r F 1 F r • Eg n=2 (wavebands), m=2 (end-members) 2 Reflectance Band 2 r 3 1 Reflectance Band 1 Linear Mixture Modelling • • as described, is not robust to error in measurement or end-member spectra; Proportions must be constrained to lie in the interval (0,1) – • • - effectively a convex hull constraint; m+1 end-member spectra can be considered; needs prior definition of end-member spectra; cannot directly take into account any variation in component reflectances – e.g. due to topographic effects Linear Mixture Modelling in the presence of Noise r F e • Define residual vector e e • minimise the sum of the squares of the error e, i.e. r F r F l m1 l 0 r l l F 2 Method of Least Squares (MLS) ee r F r F l m1 l 0 Error Minimisation • Set (partial) derivatives to zero 2 l m 1 l 0 rl l F l F l m 1 2l 0 rl l F Pi Fi 0 l F Fi l i 0 l m 1 l 0 rl l i 2l 0 l m 1 r l m 1 l 0 l l F l i l F l i r l l F 2 ee l m1 Error Minimisation l 0 rl l i ll 0m1 l F l i OMP • Can write as: rl ll 0 ll 0 ll 0 l m 1 l m 1 rl ll 1 ll 0 ll 1 l 0 l 0 r l l l l n 1 l 0 l n 1 l ll 1 ll 0 ll 1 ll 1 ll 1 ll n1 Solve for P by matrix inversion ll n 1 ll 0 F0 ll n 1 ll 1 F1 ll n 1 ll n 1 Fn 1 e.g. Linear Regression y c mx yl l n1 1 l 0 y l xl l 0 xl O MP y 1 yx x M 1 xl c 2 xl m l n 1 1 x 2 x 2 xx x 1 x x 2 xx 2 x c 2 x m y y 2 2 xx x xy2 2 xx 2 xy 2 xx x RMSE e 2 l n 1 2 y c mx i i l 0 RMSE 2 nm y x2 x x1 x Weight of Determination (1/w) • Calculate uncertainty at y(x) 1 c y x Q P x m 1 T 1 Q M Q w 1 e w 1 xx 1 w xx2 2 P1 RMSE P0 P1 RMSE P0 Issues • • • • Parameter transformation and bounding Weighting of the error function Using additional information Scaling Parameter transformation and bounding • Issue of variable sensitivity – E.g. saturation of LAI effects – Reduce by transformation • Approximately linearise parameters • Need to consider ‘average’ effects Weighting of the error function • Different wavelengths/angles have different sensitivity to parameters • Previously, weighted all equally – Equivalent to assuming ‘noise’ equal for all observations iN 2 i i measured modelled RMSE i 1 iN 1 i 1 Weighting of the error function • Can ‘target’ sensitivity – E.g. to chlorophyll concentration – Use derivative weighting (Privette 1994) i i modelled P measured i 1 2 iN i 1 P iN RMSE 2 Using additional information • Typically, for Vegetation, use canopy growth model – See Moulin et al. (1998) • Provides expectation of (e.g.) LAI – Need: • planting date • Daily mean temperature • Varietal information (?) • Use in various ways – Reduce parameter search space – Expectations of coupling between parameters Scaling • Many parameters scale approximately linearly – E.g. cover, albedo, fAPAR • Many do not – E.g. LAI • Need to (at least) understand impact of scaling Crop Mosaic LAI 1 LAI 4 LAI 0 L A I 1 Crop Mosaic • 20% of LAI 0, 40% LAI 4, 40% LAI 1. • ‘real’ total value of LAI: – 0.2x0+0.4x4+0.4x1=2.0. (1 exp( LAI / 2)) s exp( LAI / 2) visible: NIR 0.9; s 0.3 0.2; s 0.1 L A I 4 L A I 0 canopy reflectance 0.9 0.8 0.7 reflectance 0.6 0.5 visible NIR 0.4 0.3 0.2 0.1 49 46 43 40 37 34 31 28 25 22 19 16 13 10 7 4 1 0 LAI canopy reflectance over the pixel is 0.15 and 0.60 for the NIR. • If assume the model above, this equates to an LAI of 1.4. • ‘real’ answer LAI 2.0 Linear Kernel-driven Modelling of Canopy Reflectance • Semi-empirical models to deal with BRDF effects – Originally due to Roujean et al (1992) – Also Wanner et al (1995) – Practical use in MODIS products • BRDF effects from wide FOV sensors – MODIS, AVHRR, VEGETATION, MERIS Satellite, Day 1 Satellite, Day 2 X 0.45 0.4 0.35 0.25 0.2 0.15 0.1 0.05 Julian Day original NDVI MVC BRDF normalised NDVI AVHRR NDVI over Hapex-Sahel, 1992 282 275 268 261 254 247 240 233 226 218 206 199 192 185 178 171 164 157 150 143 0 136 NDVI 0.3 Linear BRDF Model • of form: l , , f iso l f vol l k vol , f geo l k geo , Model parameters: Isotropic Volumetric Geometric-Optics Linear BRDF Model • of form: l , , f iso l f vol l k vol , f geo l k geo , Model Kernels: Volumetric Geometric-Optics Volumetric Scattering • Develop from RT theory – – – – Spherical LAD Lambertian soil Leaf reflectance = transmittance First order scattering • Multiple scattering assumed isotropic sin cos 2 2 1 , l 1 eX seX 3 L X 2 Volumetric Scattering e sin cos 2 2 l 1 , 1 eX seX 3 X X L 2 1 X • If LAI small: sin cos 2 l L 2 L 1 s 1 , 3 2 2 sin cos 2 l 2 L 1 s , 3 2 sin cos 2 2 L s 1 , l 3 2 Volumetric Scattering thin l , , a0 l a1 l k vol , • Write as: sin cos 2 k vol , 2 L l a0 l s 6 L l a1 l 3 RossThin kernel Similar approach for RossThick L exp LB exp 2 Geometric Optics • Consider shadowing/protrusion from spheroid on stick (Li-Strahler 1985) shadowed crown Sunlit crown r Projection (shadowed) b h A() shadowed ground Geometric Optics • Assume ground and crown brightness equal • Fix ‘shape’ parameters • Linearised model – LiSparse – LiDense RossThick Kernels LiSparse 1 0.5 0 kernel value -75 -60 -45 -30 -15 -0.5 0 15 30 45 60 -1 -1.5 -2 -2.5 Retro reflection (‘hot spot’) -3 view angle / degrees Volumetric (RossThick) and Geometric (LiSparse) kernels for viewing angle of 45 degrees 75 Kernel Models • Consider proportionate (a) mixture of two scattering effects l , , 1 a a l aa l l 0 vol 0 geo mult 1 a a1vol l kvol , aa1geo l k geo , Using Linear BRDF Models for angular normalisation • Account for BRDF variation • Absolutely vital for compositing samples over time (w. different view/sun angles) • BUT BRDF is source of info. too! MODIS NBAR (Nadir-BRDF Adjusted Reflectance MOD43, MCD43) http://www-modis.bu.edu/brdf/userguide/intro.html MODIS NBAR (Nadir-BRDF Adjusted Reflectance MOD43, MCD43) http://www-modis.bu.edu/brdf/userguide/intro.html BRDF Normalisation • Fit observations to model • Output predicted reflectance at standardised angles – E.g. nadir reflectance, nadir illumination • Typically not stable – E.g. nadir reflectance, SZA at local mean l, , P K f iso l P f vol l f l geo 1 K k vol , k , geo And uncertainty via 1 T 1 Q M Q w Linear BRDF Models to track change • Examine change due to burn (MODIS) 220 days of Terra (blue) and Aqua (red) sampling over point in Australia, w. sza (T: orange; A: cyan). Time series of NIR samples from above sampling FROM: http://modis-fire.umd.edu/Documents/atbd_mod14.pdf MODIS Channel 5 Observation DOY 275 MODIS Channel 5 Observation DOY 277 Detect Change • Need to model BRDF effects • Define measure of dis-association observed predicted e 2 2 observed predicted e 1 1 w MODIS Channel 5 Prediction DOY 277 MODIS Channel 5 Discrepency DOY 277 MODIS Channel 5 Observation DOY 275 MODIS Channel 5 Prediction DOY 277 MODIS Channel 5 Observation DOY 277 So BRDF source of info, not JUST noise! • Use model expectation of angular reflectance behaviour to identify subtle changes Dr. Lisa Maria Rebelo, IWMI, CGIAR. 54 Detect Change • Burns are: – negative change in Channel 5 – Of ‘long’ (week’) duration • Other changes picked up – – – – E.g. clouds, cloud shadow Shorter duration or positive change (in all channels) or negative change in all channels Day of burn http://modis-fire.umd.edu/Burned_Area_Products.html Roy et al. (2005) Prototyping a global algorithm for systematic fire-affected area mapping using MODIS time series data, RSE 97, 137-162.