Jinshi-Ben-meeting 2020-06-30

[bpbond]

For this meeting, let's both read Will's 2013 paper and discuss?

[jinshijian]

Sounds good, thank

[jinshijian]

• Will mentioned extending Rh outputs of landmodels to 1980, or 1960?

• Annual Rh data cover 1961-2017, there are totally ~10,000 observations, and ~2,000 observations are from 1961-1997 (~20%).

• So we could also show 1961-2016?

• Dalei suggests combining figure 1 and figure 2, what do you think?

• Simplify Figure 3? show only the density in one panel, but not show the scatter plot?

• Global annual RH were summed up based on RH rate in each cell and the area of the cell, all data products were under the same projection to avoid the projection related bias.

• I used a random forest approach trying to figure it out under which condition landmodels tend to overestimate or underestimate Rh

• The results showed that absolute latitude and MAP are the most important factors

• Scatter plot of Mean bias error (MBE) vs MAP

• Scatter plot of Mean residual (MBE) vs absolute latitude

• Scatter plot of Mean bias error (MBE) vs MAT

[jinshijian]

Collar and time

Isn't this only a problem in cases of hysteresis, e.g. when T-responses differ between day and night and this is propagated with scaling? Worth mentioning in a sentence?