We determined for every pixel in each of the three sites the strongest correlations between the FMSI over all possible ranges of biweekly periods between March and July, and pre- cipitation in the preceding winter months (December through March)� We also deter- mined the strongest correlations between the FMSI over all possible ranges of biweekly periods between March and July, and the temperature over all possible months between February and July that precede or are concurrent with each range of the FMSI periods tested� Figure 13�5 summarizes the results�
The results showed significant (95% confidence) correlations between the FMSI and one or both climate variables for most pixels in each of the three sites� However, correlating ranges of time are often too short to be considered reasonable fire season approximations, and correlations between the FMSI and the two climate drivers often exist over differ- ent ranges of biweekly periods� Precipitation correlations can extend from early March through July, and some number of precipitation correlations end before the earliest of monsoon rains, which is feasible� Nearly all temperature-correlating FMSI period ranges are concentrated between May and June, and some continue through July�
(a)
240,000 500450 400350 300
Fire count
250200 150100 500
3. 4. 5.
Month 6. 7.
220,000 200,000 180,000 160,000 140,000 120,000 100,000 80,000 60,000 40,000 20,000
0 3. 4. 5. 6.
Month
Project area Fires by month Acres burned by month
Fires by month
600–1,131.4 3 4 5 6 7 8 9 12 1,131.400001-1,662.8 1,662.800001-2,194.2
2,725.600001-3,257 2,194.200001-2,275.6 Elevation
m
Acres burned
7.
1,593.25 3.
112 3.
199 4.
363 5.
467 6.
541 7.
52,158.45 4.
54,253.65 5.
250,403.15 6.
85,817.4 7.
Fires by month and elevation
(b)
Elevation (m)
3,000 2,900 2,800 2,700 2,600 2,500 2,400 2,300 2,200 2,100 2,000 1,900 1,800 1,700 1,600 1,500 1,400 1,300 1,200 1,100 1,000 900800
April
March May
Month June July
FIgure 13.4
(See color insert following page 426.) (a) Actual fires for the period of record covered by the FMSI� Earliest fires (e�g�, March) occur at lower elevations in the Catalina-Rincons� The FMSI shows lower elevations remain vulnerable as the fire season “progresses” in elevation� By July, all elevations are vulnerable; this is evident in Figure 13�4b, which shows July fires span high and low elevations; and in Figure 13�4c, where the FMSI shows the early, sustained vulnerability of low-elevation fuels� (b) Box plot showing fire counts by elevation per month�
The medians are the dotted lines within the box� Each box contains 50% of the values� The “whiskers” denote minimum and maximum values� Median fire counts tend to increase in elevation by month� (c) The length of fire season (LOFS) as determined by the FMSI, which shows live fuels cure later at higher elevations; fire season thus is later at higher elevations, and this is consistent with Figures 13�4a and 13�4b�
When the FMSI begins to correlate with one or both climate drivers, we interpret this to mean that moisture resources have begun to be tapped significantly, and vegetation has entered its annual season of arid fore-summer survival, with increased vulnerabil- ity to combustion� We designated the FMSI period associated with the earliest signifi- cant correlation to either climate variable as the first period in a given pixel’s fire season�
Correlations to the second climate variable, when they occur later in the season, could
Winter precipitation
Arid fore-summer
Determining the temporal boundaries of one pixel’s fire season
Maximum possible fire season length for a pixel in this climate division First good correlation
between FMS and antecedent season precipitation or current
season temperatures
Average difference between FMS in periods 15 and 14 indicates annual drop in FMS at this time This pixel’s
fire season
Monsoon Yikes!
D
Month: J
Biweekly period:
F M A
6 7 8 9 10 11 12 13 14 15
M J J A S
5
FIgure 13.5
For this pixel, the earliest significant correlation occurs in period 8, indicating the beginning of the fire season�
The average difference between FMSI in periods 14 and 15 indicates FMSI characteristically drops in period 15 each year, indicating monsoonal moistures usually begins alleviating fuel moisture stress by this time� Thus, period 14 is selected to be the last period of this pixel’s fire season�
(c)
No corr 2 4 6 8 10 12 14 16 18 20 22
4 189 461 513 610 566 728 988 880 683 274 110 Histogram LOFS(wks)
Color
FIgure 13.4 (Continued)
indicate an exhaustion stage� The earliest correlating climate driver is most often the pre- cipitation, except in upper elevations and urbanized areas, where temperature is often the earliest correlating climate driver� The maps in Figure 13�3 show the earliest correlating climate driver for each of the three sites�
The end of the fire season should be tied closely to the arrival of the monsoon rains (typi- cally between periods 13 and 15)� We designated the last period of each pixel’s fire season as that period between 13 and 15 that precedes the period when the average (taken over all years in the dataset) FMSI difference between periods is negative and stays negative through period 15� This average drop in FMSI suggests that monsoonal moisture is usu- ally effective by this time for this pixel� When no drop in value is detected, period 15 is assumed to be the end of the season� Lack of a drop in the FMSI could possibly be due to a characteristic longer lag in vegetative response to monsoon moisture and characteristi- cally later rains in these locations� Figure 13�6 illustrates how one pixel’s fire season would be modeled�