A long-standing issue in dendrohydrology is measuring and
expressing uncertainty in the inference of water levels from treering
proxies. Standard practice is to derive a single best reconstruction
from a small pool of predictor chronologies from sites in or near
the watershed.