Long-Range Forecasting of Surface Air Temperature and Precipitation for the Korean Peninsula
Abstract
We have designed, developed, and tested a long-range forecasting system for producing forecasts of surface air temperatures and precipitation rates in the Korean Peninsula region at leads of two months for each calendar month. We tested predictors based on the following: (1) indices of the Arctic Oscillation, El Nino/La Nina, North Atlantic Oscillation, Pacific/North American Pattern, and the West Pacific Pattern; (2) 850 hectopascal geopotential heights and sea surface temperatures in specific regions; (3) persistence; and (4) year (to represent long-term trends). Our forecasting system includes 24 multiple linear regression models, one for temperature and one for precipitation for each month. Each model uses a unique set of predictors. We tested each model by conducting 43 years of cross-validated hindcasting for our 1970-2012 study period. The hindcast results showed that, overall, the models had skill in predicting above normal, near normal, and below normal temperatures and precipitation rates for the Korean Peninsula (e.g., Heidke skill scores > 0). We used our January models to successfully forecast temperatures and precipitation for January 2013. We also developed a series of forecaster worksheets to be used to produce forecasts for the Korean Peninsula.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2013
- Accession Number
- ADA579958
Entities
People
- Nicholas J. Reitz
Organizations
- Naval Postgraduate School