一、ERA-Interim
1. R，预报场，选的0点，step 12，2010年数据，0.5x0.5
2. R，预报场，step 12，选的剩下的12点，2010年数据
与上面对比。
哇地一声在自习室忍住了哭，这两个咋对比啊？就是差了12个小时的预报值呀。
问题来了：预报值和实测值到底差多少？差了半天的预测值又要如何修正？某一天某一点的预报值能代表一个月的数值吗？公开的数据中心里到底有没有实测值来修正一下预报值？是不是径流这种数据，难以用卫星监测到，没有合适的模型计算全球范围的数据，so，只有预测值？
预测值划去，数据同化之后的数据产品，是模拟资料，可以用来分析、预测。
数据同化基本思想是不同源的实测资料、不同模型的模拟资料、不同的观测项目，通过一定的技术手段把这些数据综合在一起，实现序列拓展、精度提升或者是其他目的。
模拟资料，不一定是预测的；对无观测资料地区或者未来时段，叫预测
二、ERA-Interim/Land
surface runoff，好像是日数据，好大的数据啊…不知道又要下多久。。。
先去吃饭吧。。
-----------更新至2017/07/08-----------
2017.8.14 更新
1. 再下过一个runoff，大家一起来找茬。。
Monthly means of daily forecast accumulations ,step 0-12
apps.ecmwf.int/datasets/data/interim-mdfa/levtype=sfc/requests/netcdf/59911de7dd5e9faa93b74684/
英文好重要T^T
2. 在气象家园上看到这么个说法，
bbs.06climate.com/forum.php
【Synoptic Monthly Means和Monthly Means of Daily Means ，做研究一般是下后者吧，前者好像是用于业务的~
在ERA-40的Synoptic Monthly Means资料下有time select这个选项，这个选项下分成了四个时刻，假如我单选0:00这个时刻，那么得到的数据是每个月在0:00时刻的月平均值，单选12:00即每月该时刻的平均值。如果同时选0:00和12:00两个时刻，那么得到的数据就是这两个时刻所对应的月平均值之和的一半，以此类推。】
为此，同时选择了step 0-12 与12-24的下载……
更正，气象家园里的说法是time select，并不是step。
2017.8.14 更新
向李老师寻求帮助后，
time select 并不会过多影响数据，具体应该是预报的时间点。
step 的选项应该是用来数据合成的时长，越长一般越有代表性。
Monthly Means of Daily Forecast Accumulations(mdfa) are similar toMonthly Means of Daily Meansbut for accumulated fields (eg precipitation, radiation). For example for precipitation, the data gives the monthly mean of daily (24 hour) precipitation. In addition, the data is available for three forecast steps, 0-12, 12-24 and 24-36 hours. Note these are not synoptic periods (00:00 to 12:00, etc), but specifies the provenance of the accumulated data, e.g. from the 0-to-12 hour forecast, 12-to-24 hour forecast, etc. In any case the data gives the monthly mean of daily 24 hour accumulation. Most users will want to use step 0-12.
1. ERA-Interim，
monthly means of forecast accumulated，下载径流与蒸发数据，79年1月—16年12月，0.125°x0.125°。
unit：m ，月均值，需乘以当月天数。
问题：需处理单位，需乘以当月天数得月总量。
runoff： apps.ecmwf.int/datasets/data/interim-mdfa/levtype=sfc/selectors/netcdf/1000302/
2. 来个日数据试试水。
ERA-Interim，Daily，
Runoff， time：12:00 ， step = 12， 2016年1月数据，0.125°，
apps.ecmwf.int/datasets/data/interim-full-daily/levtype=sfc/selectors/netcdf/999398/
3.ERA-Interim，Monthly means of daily means
2m temperature 与 skin temperature 的比较
Skin temperature is defined as the temperature of the surface at radiative equilibrium. It forms the interface between soil, snow or ice and the atmosphere. Skin temperature in ERA-Interim (code 235) is derived from the surface energy balance. Over ocean, the skin temperature is set to the surface temperature.
You should use 2 meter temperature to compare with station observations, since "surface" air temperature observations are taken at 2 m above ground. Skin temperature in the model is the ground surface temperature (or SST over water) which can be quite different from the 2-m air temperature. Over land, skin temperature is calculated from the surface energy balance.
pass skin temperature
Monthly Means of Daily Means(moda) is available only for analysis and instantaneous forecast data. It is calculated like this, for example for temperature: You take the temperature readings for every day at 00:00, 06:00, 12:00, and 18:00 UTC for a month, and you average them. The result is a single mean temperature value (at each location) for the month. Then you do the same for the next month, and so on.
t2m 和 NOAA 的比较一下。
4.ERA-Interim，Synoptic Monthly means
有total precipitation，等会比较一下。
Synoptic Monthly Means(mnth) from analyses are the monthly averages produced for each of the four main synoptic hours (00, 06, 12, and 18 UTC). For example with temperature: you take the temperature every day at 00:00, and then you average them over every month. So you get the average temperature at 00:00 in January, in February, etc. Then you do the same for the 06:00 temperature, then for the 12:00 temperature, and then for the 18:00 temperature. The result is the "Synoptic Monthly Mean" at 00:00, at 06:00, at 12:00 and at 18:00 for every month. "Synoptic Monthly Means" from forecasts are similar, except that they are for particular forecast start times and forecast steps.
SMM的数据 pass
