344
R. Arnone et al.
variability in the physical processes and secondarily to the monthly bio-optical processes. Notice that regions with the highest standard deviation of backscattering
(Fig. 19.6b) are locations associated with physical forcing in strong tidal regimes
and coastal plumes.
We define the 24 h “persistence error” of the backscattering product as the difference within a 24 h change in backscattering. High changes indicate high persistence
error, or lack of persistence. A lower error suggests that small changes occur and
a similar backscattering coefficient can be used from 1 day to the next to represent
the forecast. The persistence error is computed by differencing locations using two
sequential (within 24 h) satellite backscattering fields.
The difference is computed daily and then averaged over the month. The persistence error represents change resulting from both physical and bio-optical processes,
since it is based only on observed satellite retrievals of backscattering products and
not on the physical circulation model. We computed this monthly persistence error
field for October 2009, as mean and standard deviation (Fig. 19.7a, b).
To evaluate the representativeness of the monthly “persistence errors” of the
mean and standard deviation, the number of match up samples (or satellite observed
pairs) for each grid point is shown in Fig. 19.7c. This is the number of data points
Fig. 19.7 The backscattering coefficient “persistence error” was computed as difference between
two satellite backscattering retrievals within a 24 h period and statistics assembled for October
2009: (a) monthly mean (black areas indicate no pair of 24 h difference were observed); (b)
monthly standard deviation; and (c) number of observational pairs used to generate the monthly
statistics
R. Arnone et al.
variability in the physical processes and secondarily to the monthly bio-optical processes. Notice that regions with the highest standard deviation of backscattering
(Fig. 19.6b) are locations associated with physical forcing in strong tidal regimes
and coastal plumes.
We define the 24 h “persistence error” of the backscattering product as the difference within a 24 h change in backscattering. High changes indicate high persistence
error, or lack of persistence. A lower error suggests that small changes occur and
a similar backscattering coefficient can be used from 1 day to the next to represent
the forecast. The persistence error is computed by differencing locations using two
sequential (within 24 h) satellite backscattering fields.
The difference is computed daily and then averaged over the month. The persistence error represents change resulting from both physical and bio-optical processes,
since it is based only on observed satellite retrievals of backscattering products and
not on the physical circulation model. We computed this monthly persistence error
field for October 2009, as mean and standard deviation (Fig. 19.7a, b).
To evaluate the representativeness of the monthly “persistence errors” of the
mean and standard deviation, the number of match up samples (or satellite observed
pairs) for each grid point is shown in Fig. 19.7c. This is the number of data points
Fig. 19.7 The backscattering coefficient “persistence error” was computed as difference between
two satellite backscattering retrievals within a 24 h period and statistics assembled for October
2009: (a) monthly mean (black areas indicate no pair of 24 h difference were observed); (b)
monthly standard deviation; and (c) number of observational pairs used to generate the monthly
statistics