Although theories of thoroughly mixed and unmixed wellbore flow have been d
eveloped, recent studies suggest that ground water data variability and bia
s can result from various types of partial mixing. In this study, digital i
mages of dye distributions within a vertical cross-sectional model of a scr
eened monitoring well allowed mixing processes to be recorded and character
ized. Given the conditions represented by this study, two scales of well wa
ter mixing are evident; small-scale mixing adjacent to the well screen and
larger-scale mixing with the water from the casing above the screen section
. Evidence suggested that the small-scale mixing initially caused concentra
tions at the pump to change more rapidly than the aforementioned theories p
redicted and to stabilize at biased concentrations. Larger-scale mixing of
casing water and screen section water resulted from either turbulent/viscou
s entrainment or, more likely, a slightly elevated casing water density. If
this mixing occurs due to wellhead cooling where the casing water is signi
ficantly chemically altered, bias would result. These results suggest that
concentration variability and bias before and during pumping are, in part,
controlled by the characteristics and prevalence of small and large scale p
artial mixing throughout the wellbore in the presence of wellbore-concentra
tion heterogeneities. The bias and variability could be on the order of tho
se concentration heterogeneities where concentration heterogeneities and pa
rtial mixing are present. Bias and variability of the type observed under t
hese experimental conditions call for investigations into conditions that l
ead to partial mixing in the field and into the possible magnitude of the i
nfluences of these processes.