A study of predictability of SST at different time scales based on satellite time
2008
Sea surface temperature (SST) is both an important variable for weather and ocean forecasting, but also a key indicator
of climate change. Predicting future SST at different time scales constitutes an important scientific problem. The
traditional approach to prediction is achieved through numerical simulation, but it is difficult to obtain a detailed
knowledge of ocean initial conditions and forcing. This paper proposes a improved prediction system based on SOFT
proposed by Alvarez et al and studies the predictability of SST at different time scales, i.e., 5 day, 10 day, 15 day, 20 day
and month ahead. This method is used to forecast the SST in the Yangtze River estuary and its adjacent areas. The period
of time ranging from Jan 1st 2000 to Dec 31st 2005 is employed to build the prediction system and the period of time
ranging from Jan 1st 2006 to Dec 31st 2007 is employed to validate the performance of this prediction system. Results
indicate: The prediction errors of 5 day,10 day,15 day, 20 day and monthly ahead are 0.78°C,0.86°C,0.90°C,1.00°C and
1.45°C respectively. The longer of time scales prediction, the worse of prediction capability. Compared with the SOFT
system proposed by Alvarez et al, the improved prediction system is more robust. Merging more satellite data and trying
to better reflect the real state of ocean variables, we can greatly improve the predictive precision of long time scale.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
7
References
1
Citations
NaN
KQI