< p >我不是专家,我刚处理各种各样的gfs格丽有一段时间了。我知道他们是我唯一读过,我引用了。我不知道你在哪里得到这些文件。它看起来像你在看文件的子领域,但在通用文件格式是< a href = " https://www.unidata.ucar.edu/data/conduit/ldm_idd/gfs_files.html " rel = " noreferrer " >更清楚地显示在这个网站上< / >:< / p > < pre > <代码>数据/ nccf /com/gfs/prod/gfs.YYYYMMDDHH/gfs.tHHz。formidffhour ! GRIB_version / ncep /模型/ # 000 / YYYYMMDDHHNNFfff /改/水平!# # # # # # < /代码> < / pre > < p >网格# 004,0.5 x0.5度纬度/经度网格,formid = pgrb2(小时F000-F180) GRIB_version = grib2(网格编号显示为# 000)注:fhour:退火,f00, f06, f12 ....f384(注:2或3位数fhour ftp服务器上)GRIB2注:“模型日期”和“预计一小时”字段不是由“/”< / p > < p >我建议你把你的文件从< a href = " http://www.nco.ncep.noaa.gov/pmb/products/gfs/ " rel = " noreferrer " > < / >开始。< / p > < p >退火和fxxx fxxx中间模拟运行,通常包含未初始化的数据。< / p > < blockquote > < p >预测包含相同的参数分析和预测变量如沉淀积累起来的。分析文件,如果他们包含预计的字段,设置为零。例如,新兵和GFS直接从摘要获得的文件,包含了克雷恩变量。但在新兵文件,所有的科伦字段将是零。 In summary, the FNL analysis incorporates ~10% more data than the GFS by waiting for ~3 hours longer for the data stragglers. GFS forecasts contain accumulated fields, such as precipitation, that are missing or set to zero in the analysis fields. - link
This analysis-forecast cycle is complex and iterative. Here's a bit more detail.
And here is the current status in real time of models running.