X. Li et al., Validation of coastal sea and lake surface temperature measurements derived from NOAA/AVHRR data, INT J REMOT, 22(7), 2001, pp. 1285-1303
An interactive validation monitoring system is being used at the NOAA/NESDI
S to validate the sea surface temperature (SST) derived from the NOAA-12 an
d NOAA-14 polar orbiting satellite AVHRR sensors for the NOAA CoastWatch pr
ogram. In 1997, we validated the SST in coastal regions of the Gulf of Mexi
co, Southeast US and Northeast US and the lake surface temperatures in the
Great Lakes every other month. The in situ temperatures measured by 24 NOAA
moored buoys were used as ground data. The non-linear SST (NLSST) algorith
m was used for all AVHRR SST estimations except during the day in the Great
Lakes where the linear multichannel SST (MCSST) algorithm was used. The bu
oy-satellite matchups were made within one image pixel in space (1.1 km at
nadir) and +/-1 h in time.
For the NOAA-12 satellite, the validation results for the three coastal reg
ions (Gulf of Mexico, Southeast US and Northeast US) showed that the mean t
emperature difference between satellite and buoy surface temperature (bias)
was about 0.4 degreesC during the day and 0.2 degreesC at night. The stand
ard deviation was about 1.0 degreesC. Great Lakes validation results showed
a bias less than 0.3 degreesC during the day. However, due to the early mo
rning fog situation in the summer months in the Great Lakes region, the NLS
ST night algorithm yielded a fairly large bias of about 1.5 degreesC.
The same statistics were computed for the NOAA-14 satellite measurements. F
or the coastal regions, the bias was less than 0.2 degreesC with a standard
deviation about 1.0 degreesC. For the Great Lakes region, the bias was abo
ut 0.4 degreesC for both day and night with a standard deviation about 1.0
degreesC.
Our study also showed that the NLSST algorithm provides the same order of S
ST accuracy over all study regions and under a wide range of environmental
conditions.