Preliminary Results for Retrievals Using GFS Forecast as First Guess

Susskind, Joel ; Blaisdell, John

关键词: ANOMALIES;    BIAS;    CLIMATE CHANGE;    NEURAL NETS;    OZONE DEPLETION;    SURFACE TEMPERATURE;    TEMPERATURE PROFILES;    WEATHER FORECASTING;    ATMOSPHERIC MOISTURE;    CLIMATOLOGY;    COMPUTER NETWORKS;    TEMPERATURE GRADIENTS;    TIME SERIES ANALYSIS;    WATER VAPOR;   

RP-ID  :  GSFC-E-DAA-TN62338

学科分类：地球科学(综合)

美国|英语

来源: NASA Technical Reports Server

【 摘 要 】

Four days of retrievals in different years were performed using both a Neural Network first guess and also using the GFS forecast temperature profile, water vapor profile, and surface temperature as the first guess.Yield and temperature and water profile plots for days early and late in the AIRS mission follow.Yield is higher near the surface for forecast based retrievals.Statistical comparisons show that forecast based retrieval temperature RMS differences from ECMWF are smaller than those of the Neural Net based system. AIRS retrievals improve over the Neural Net but degrade the forecast.Statistical comparisons with ECMWF show that the bias structure of the forecast guess retrievals varies with time during the AIRS mission much more than those of the Neural Network based retrievals for both temperature and water vapor profiles.