Figure
9: Illustration of the relationship between the retrieved values and
the MCD prior. The deviance of the synergy from the MCD estimates are
calculated as a relative difference such that a value of 0 indicates
where the synergy and the MCD are equal, and instances where the synergy
yields larger values are positive. The left column shows the ratio of
the posterior to the prior full water vapor column, while the right
column shows the ratio of the posterior to the prior partitioning
indices. The top row visualizes the data on a latitude by Ls grid, while
the bottom row distributes the data on a latitude-longitude map. The
data has been averaged in bins of 2° latitude, 2° Ls and 4° longitude.
The vertical confinement displays the opposite behavior, with the
synergy often finding a PI comparable to the MCD at mid and high
latitudes. Note that during the sublimation season the MCD quite
accurately reproduces the observations, indicating that the sublimation
processes of the NPC are quite well understood in terms of vertical
distribution. The PI difference is highest at low and middle latitudes
in late summer, when large amounts of water vapor are being transported
from the NH and across the equator. At most, the synergy PI at
low-latitudes in late summer is almost twice as strong as model
predictions. This is in general a fairly dry area, with a CIA of 10-15
pr-μm, where the synergy indicates that roughly 60% of the column is
confined below 5 km. The atmospheric behavior in this region is less
dominated by temperature and more affected by wind. The details of local
air flow patterns are typically known with less certainty than
temperature variations, which could explain why the model deviates the
most from the observations at low and mid-latitudes.
4.5 Seasonal evolution of water distribution with latitude
The seasonal variations of the CIA and PI can be visualized by zonal
averages plotted as a function of latitude. All data points in Figures
11 and 12 illustrate the CIA and PI averaged in bins of 2° latitude and
15° intervals of Ls, and the curves are smoothed using a Savitzky-Golay
low-pass filter with a second order polynomial and a window
corresponding to 20° latitude. Curves covering the same seasonal periods
have identical colors for both hemispheres to aid cross-hemispherical
comparison (for example, the red curve corresponds to mid-spring for
both hemispheres; Ls=45°-60° in the NH and Ls=225°-240° in the SH).
Both hemispheres are fairly dry from the equator to mid-latitudes during
the spring-summer season. The SH displays a smaller spread in seasonal
variation and a smaller increase with latitude compared to the NH,
remaining at around 10 pr-μm from the equator to 40°S. From there, the
water column starts to increase steadily. Overall, the synergy and MCD
agree very well in the SH, with the most noticable difference being the
degree of seasonal spread, distinguishable at all latitudes in the
synergy, while only becoming distinguishable after 40°S for the MCD. All
synergistically retrieved seasonal curves show a southern maximum which
is migrating poleward with season, matched well by the MCD. The
exception is the first seasonal average in mid-spring (Ls=225°-240°)
which displays a continuously decreasing curve, with the highest value
at equator for the synergy, while the MCD finds a weak maximum of 15
pr-μm at 70°S for the same season. The SH sublimation season maximum
occurs during Ls=285°-300°, with a maximum value of 34 pr-μm near 87°S.