156
Chapter 17
for national and global inventories, full image coverage occurs once
every 1–5 years, in practice all areas should be covered annually
for operative inventories, once per year
clouds often prevent optical imaging with sufficient frequency
spatial resolution
for national and continental inventories, 10–30 m
for stand-level inventories, 5–10 (20) m
availability of ancillary data
DTM is almost always necessary
digital map data improve the accuracy, sometimes necessary
phenological and soil data improve the accuracy
spectral range and resolution
the range 450–680
is important
the range 680–750
is very important
the range 780–810
is very important
the range beyond 850
is very important, i.e., Landsat TM bands,
3, 4, 5 (all bands should be used)
radiometric resolution
in today’s satellites, it is usually poor compared to the requirements
signal-to-noise ratio should be improved, especially at the low end of
the spectrum
important especially for small scale estimates
requirements for methods
define the inventory units
all important variables should be estimated for all units
pixel level errors may sometimes be high, in spite of low inventory
units level errors
One of the crucial problems is image availability. Recent microwave data
cannot compensate the lack of optical area remote sensing data (Tomppo et
al. 1996).
5.
ADVANTAGES OF REMOTE SENSING BASED
INVENTORIES
There are some obvious advantages of remote sensing data for forest
inventories:
information can be produced with small additional costs for much larger
areas than is possible with field measurements only
information is more accurately geo-referenced than with field
measurement-based inventories