Fire intensity is considered an important component of our understanding of savannas. Fire intensity can be used to predict probability of top kill of woody stems, and thus can deterministically link fire to demographic bottlenecks observed in savanna woody species. However, our understanding of the spatial processes of fire intensity is limited: What tree neighborhood effects on fire intensity exist? How does intensity vary across woody cover gradients? Is fire intensity really the best metric for understanding impact of fire on savanna plant communities? My research hopes to expand on these topics while suggesting future directions for research and applications in fire management.
4. For a given interaction of fuel + oxygen +
heat…
How much energy is released?
5. Intensity explains p(topkill)
Demographic bottlenecks
Structurally defines savannas and allows for
coexistence of trees, grass
Although not the only factor…
Fire limits tree establishment
6. If fire is a strong factor in limiting tree
establishment in savannas…
In areas of high tree cover, you could expect
either infrequent or low intensity fires
In areas of low tree cover, you could expect
either frequent or high intensity fires.
7. I = Hwr
I – Fireline Intensity (kW m-1)
H – Heat content (kJ g-1)
w – Mass of combusted fuel (g)
r – Rate of Spread (m s-1)
8.
9.
10. Why do we see increasing woody cover at the
same time as increasing p(top kill)?
Confounding effects of granitic vs. basaltic?
Fire doesn’t limit as much as we think
Fire intensity as we measure it doesn’t
capture the effects on woody cover that we
think it should
11. Severity is the ecological impact of a fire
Plants
Animals
Soil
Intensity is energy released
Intensity != Severity…
…but no standard metric for severity in
savannas
12. Remote Sensing using MODIS Fire Product
Modeled fireline intensity
Analysis of fire scar reports from KNP
Local-scale intensity variation
13.
14. Ten years of data (1 Jan 2004 – 1 Jan 2014
Subset to only include fires of > 90% accuracy
15.
16.
17. No clear relationship with MAP
No clear relationship with woody cover
FRP bad metric for savanna fires?
18.
19. Mean
Low
RH
Wind
Speed
Range
MAP
MFRI
Fuel Load:
f(Time Since
Fire, Rainfall)
Fire Intensity:
f(Fuel Load, RH, Fuel
Moisture, Wind Speed)
Fuel
Moisture
Range
Probability of Top Kill
f(Intensity, Height)
20.
21.
22.
23.
24. Intuitive: hotter fires near PRET
Clear relationship with MAP…but that’s
expected given the model.
No clear relationship with woody cover,
though differences in intensity between
geologic material.
25.
26. Date
Location
Size
Impact on Woody Cover
Impact on Herbaceous Material
27. Impact on Woody Cover
Very severe – None
Impact on Herbaceous Material
Very patchy – Clean
28.
29.
30.
31.
32. Most of this is intuitive
To do:
Link to previous summer’s rainfall
Use as training samples for remotely sensed data
33.
34. How “spatial” is fire intensity?
Fire intensity has deterministic drivers (fuel
moisture, available oxygen, fuel amount)
Do these vary enough around trees to drive
potential effects on p(top kill)?
38. Classic point clouds.
Similar to earlier studies (Moustakas et al.
2013), at wetter end, grass biomass is
decoupled from canopy coverage
39. Variation in fuel moisture content ~ Canopy
Coverage.
Better assessment of intensity from
temperature curves
Canopy opening patch size effect on fire
intensity, controlling for RH%
40.
41. When measured or modeled, intensity
doesn’t appear to predict woody cover.
…so does it matter as much as we think?
Maybe we need to shift away from intensity
as a metric..
And what, exactly, does fire severity look like
in savannas?
42.
43.
44. LAL (lighting activity level)
Dispersion (atmospheric stability)
Mixing Height (“ceiling”)
Haines Index (atmospheric driven large fire
growth)
45. ERC is the potential energy released from a
headfire
Calculated based on dynamic fuel models
Fuel moisture calculated based on
magnitude/intensity of rain, RH%, live and dead
fuel moisture
Extensive monitoring of live and dead fuel
moisture
46. Oak savanna research: 10 hr live fuel moisture
content key to top kill
Haines Index: Likelihood of “firestorm”
Editor's Notes
So Byram’s Fireline Intensity assumes a trade off between amount combusted and rate of spread. Each of these lines is a fire of equal intensity. We thus assume that low rate of spread will consume more material, and vice versa.
Thus if we plug this into our
So I modeled some broad scale relationships. A) here is average woody cover % by MAP from the Bucini data set.
B. Is fuel load by MAP, as described by Navashni in 2006.
C. Is the relationship between fireline intensity and fuel load. Linking B and C, we get D. E, then, takes the Higgins et al.’s relationship between p(topkill) and links it to D.
Product of MODIS 4 and 11 micrometer radiances
Detects thermal anomalies and calculates megawatts / km^2
MFRI came from Izak’s work. MAP is from BioClim. Fuel Moisture and wind speed values are from published values from Winston Trollope. Mean Low RH is from a gridded South African ag dataset. Fuel load function comes from Govender et al. 2006, as does fire intensity. P(top kill) is Higgins most recent one that used a likelihood estimation framework.
South African Atlas of Climatology and Agrohydrology
Lots of noise, no clear signal. Not really seeing otter fires in high woody cover or low woody cover areas.
A. Is high, B is avg
N=75
For a system for measuring maxima ~ biomasss…sort of works.
Bit of a point cloud still. Need to account for RH in this. May integrate under the curve for each of these instead of max temp.
Trend is non-significant and is weighted by outliers. This doesn’t include this year’s data, which amounts to ~ 300.
RAWS are scattered throughout the country – over 2200 of them