4.4. Forest vulnerability index (FVI)
The FVI is based on the FSI (Fig. 1) calculated at the pixel level for
each month, from April through October, from the first full year of
Aqua LST data (2003) through 2012. This seasonal period captures the
primary physiologically active months for plants during which thermal
and water stress tend to occur in our study region, and includes the
onset and alleviation of these seasonal stress cycles. The ET andmonthly
mean LST were resampled using bilinear interpolation from 1-km to
800-m to match the precipitation resolution. The FTG data were
resampled to 800-m using a majority rule and used as mask to isolate
forests in the LST and WB datasets. To contrast LST and WB, as per our
conceptual model, it was necessary to normalize those datasets, which
was done at the FTG level. The normalized LST andWB were calculated
as:
Zi ¼ yi–y̅
SDy
ð1Þ
where yi is either the LST or WB for pixel i for a given month in a
given year, and ȳ and SDy are the average and standard deviation of
LST or WB, respectively, across months (April–October) and years
(2003–2012). The FSI was calculated as
FSI ¼ LSTz WBz ð2Þ