06.03.2013 Views

7th Workshop on Forest Fire Management - EARSeL, European ...

7th Workshop on Forest Fire Management - EARSeL, European ...

7th Workshop on Forest Fire Management - EARSeL, European ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Correcti<strong>on</strong> of topographic effects influencing the differenced normalized burn ratio’s optimality for estimating fire severity 273<br />

2.3 - Correcting for illuminati<strong>on</strong> effects<br />

Methods that correct for illuminati<strong>on</strong> effects are based <strong>on</strong> the cosine of the<br />

incidence angle, which is the angle between the normal to the ground and<br />

the sun rays (Teillet et al., 1982):<br />

cos γ i = cos θ p cos θ z + sin θ p sin θ z cos (φ a - φ o ) (3)<br />

where γ i is the incident angle; θ p is the slope angle; θ z is the solar zenith<br />

angle; φ a is the solar azimuth angle; and φ o is the aspect angle.<br />

In the c-correcti<strong>on</strong> method terrain corrected reflectance ρ t is defined as:<br />

( cos θ ) z + ck ρt = ρa ---------------cos<br />

γt + ck where c k is a band specific parameter.<br />

We propose a modified c-correcti<strong>on</strong> method that corrects reflectance to a<br />

maximum illuminati<strong>on</strong> cos γ t of <strong>on</strong>e, instead of normalizing as a functi<strong>on</strong><br />

of the solar zenith angle:<br />

( )<br />

1 + c k<br />

ρ t = ρ a ----------------cos<br />

γ t + c k<br />

Index optimality was compared am<strong>on</strong>g eight aspect classes and am<strong>on</strong>g different<br />

classes of bi-temporally averaged illuminati<strong>on</strong>.<br />

3 - Results and discussi<strong>on</strong><br />

Figures 2A-C depict the topographically uncorrected, c-corrected and modified<br />

c-correcti<strong>on</strong> dNBR optimality maps. The modified c-correcti<strong>on</strong> dNBR<br />

optimality (mean = 0.66) outperformed c-corrected and uncorrected optimality<br />

(means of respectively 0.59 and 0.57), whereas c-corrected optimality<br />

provided slightly better results than uncorrected optimality. This is also<br />

reflected when the respective histograms are inspected (see figures 2D-F).<br />

(4)<br />

(5)

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!