COSMO5.0 simulation: Too cold in winter – in #9: CCLM

in #9: CCLM

<p> Dear all, </p> <p> I set up a new simulation with the new version of <span class="caps"> COSMO </span> - <span class="caps"> CLM </span> (5.0.9). The driven data is <span class="caps"> ERA </span> -Interim and simulation region is <span class="caps"> CORDEX </span> -EU. All the settings are following the <span class="caps"> CORDEX </span> -EU suggestions. However, I found that the <span class="caps"> CCLM </span> simulates a too cold winter (with a large <span class="caps"> BIAS </span> : &lt;-2 degree Celsius) in the <span class="caps"> CORDEX </span> -EU region. The temperature in summer seems comparable with the observation ( <span class="caps"> CRU </span> ). The <span class="caps"> BIAS </span> of precipitation in summer ( <span class="caps"> JJA </span> ) and winter ( <span class="caps"> DJF </span> ) are acceptable. Following some statistic metrics, the <span class="caps"> BIAS </span> , <span class="caps"> ACC </span> (anomaly correlation coefficient) and <span class="caps"> PCC </span> (pattern correlation coefficient). Then, I attach the <span class="caps"> YUSPECIF </span> and some plots about this simulation. Do you have any idea about the cold <span class="caps"> BIAS </span> ? How to fix it? Thank you very much. </p> <p> annual JJA DJF BIAS -1.23 0.45 -2.64 <br/> tas ACC 0.91 0.92 0.9 PCC 0.99 0.99 0.95 </p> <p> BIAS -0.04 -0.3 0.08 <br/> pr ACC 0.8 0.73 0.89 PCC 0.98 0.96 0.96 </p> <p> Best regards, <br/> Bo </p>

  @bohuang in #570fad1

<p> Dear all, </p> <p> I set up a new simulation with the new version of <span class="caps"> COSMO </span> - <span class="caps"> CLM </span> (5.0.9). The driven data is <span class="caps"> ERA </span> -Interim and simulation region is <span class="caps"> CORDEX </span> -EU. All the settings are following the <span class="caps"> CORDEX </span> -EU suggestions. However, I found that the <span class="caps"> CCLM </span> simulates a too cold winter (with a large <span class="caps"> BIAS </span> : &lt;-2 degree Celsius) in the <span class="caps"> CORDEX </span> -EU region. The temperature in summer seems comparable with the observation ( <span class="caps"> CRU </span> ). The <span class="caps"> BIAS </span> of precipitation in summer ( <span class="caps"> JJA </span> ) and winter ( <span class="caps"> DJF </span> ) are acceptable. Following some statistic metrics, the <span class="caps"> BIAS </span> , <span class="caps"> ACC </span> (anomaly correlation coefficient) and <span class="caps"> PCC </span> (pattern correlation coefficient). Then, I attach the <span class="caps"> YUSPECIF </span> and some plots about this simulation. Do you have any idea about the cold <span class="caps"> BIAS </span> ? How to fix it? Thank you very much. </p> <p> annual JJA DJF BIAS -1.23 0.45 -2.64 <br/> tas ACC 0.91 0.92 0.9 PCC 0.99 0.99 0.95 </p> <p> BIAS -0.04 -0.3 0.08 <br/> pr ACC 0.8 0.73 0.89 PCC 0.98 0.96 0.96 </p> <p> Best regards, <br/> Bo </p>

COSMO5.0 simulation: Too cold in winter

BH wrote

Dear all,

I set up a new simulation with the new version of COSMO - CLM (5.0.9). The driven data is ERA -Interim and simulation region is CORDEX -EU. All the settings are following the CORDEX -EU suggestions. However, I found that the CCLM simulates a too cold winter (with a large BIAS : <-2 degree Celsius) in the CORDEX -EU region. The temperature in summer seems comparable with the observation ( CRU ). The BIAS of precipitation in summer ( JJA ) and winter ( DJF ) are acceptable. Following some statistic metrics, the BIAS , ACC (anomaly correlation coefficient) and PCC (pattern correlation coefficient). Then, I attach the YUSPECIF and some plots about this simulation. Do you have any idea about the cold BIAS ? How to fix it? Thank you very much.

annual JJA DJF BIAS -1.23 0.45 -2.64
tas ACC 0.91 0.92 0.9 PCC 0.99 0.99 0.95

BIAS -0.04 -0.3 0.08
pr ACC 0.8 0.73 0.89 PCC 0.98 0.96 0.96

Best regards,
Bo

View in channel
<p> Please see the temperature differences for <span class="caps"> CCLM </span> 5.0_clm9 to <span class="caps"> EOBS </span> in the attachment and the <span class="caps"> YUSPECIF </span> for a five year simulation 1980-1984 driven by <span class="caps"> ERAI </span> nterim. <br/> There is no such negativ bias as you have in your simulation. <br/> In the following are the main differences between your and my <span class="caps"> YUSPECIF </span> . <br/> When you find which of the parameters is the bad guy, please let us know. </p> <p> <pre> You Me<br/> mu_rain 0.0000 0.5000<br/> entr_sc 0.0002 0.0003<br/> fac_rootdp2 0.9000 1.0000<br/> rlwidth 586666.0000 500000.0000<br/> itype_conv 3 0<br/> hincrad 0.0000 1.0000<br/> nincconv 1 2<br/> ico2_rad 8 2</pre> </p> <p> </p>

  @burkhardtrockel in #ebf20bb

<p> Please see the temperature differences for <span class="caps"> CCLM </span> 5.0_clm9 to <span class="caps"> EOBS </span> in the attachment and the <span class="caps"> YUSPECIF </span> for a five year simulation 1980-1984 driven by <span class="caps"> ERAI </span> nterim. <br/> There is no such negativ bias as you have in your simulation. <br/> In the following are the main differences between your and my <span class="caps"> YUSPECIF </span> . <br/> When you find which of the parameters is the bad guy, please let us know. </p> <p> <pre> You Me<br/> mu_rain 0.0000 0.5000<br/> entr_sc 0.0002 0.0003<br/> fac_rootdp2 0.9000 1.0000<br/> rlwidth 586666.0000 500000.0000<br/> itype_conv 3 0<br/> hincrad 0.0000 1.0000<br/> nincconv 1 2<br/> ico2_rad 8 2</pre> </p> <p> </p>
BR replied

Please see the temperature differences for CCLM 5.0_clm9 to EOBS in the attachment and the YUSPECIF for a five year simulation 1980-1984 driven by ERAI nterim.
There is no such negativ bias as you have in your simulation.
In the following are the main differences between your and my YUSPECIF .
When you find which of the parameters is the bad guy, please let us know. 

               You          Me

mu_rain       0.0000       0.5000

entr_sc       0.0002       0.0003

fac_rootdp2   0.9000       1.0000

rlwidth       586666.0000  500000.0000

itype_conv    3            0

hincrad       0.0000       1.0000

nincconv      1            2

ico2_rad      8            2

View in channel
<p> I have performed a simulation with the same <span class="caps"> YUSPECIF </span> as yours (except the period: 1980-1984). <br/> The comparison to <span class="caps"> EOBS </span> data shows no such large errors as your results (see attachment). <br/> Maybe there is an error in the pre- or postprocessing of your data? </p>

  @burkhardtrockel in #5a4fa54

<p> I have performed a simulation with the same <span class="caps"> YUSPECIF </span> as yours (except the period: 1980-1984). <br/> The comparison to <span class="caps"> EOBS </span> data shows no such large errors as your results (see attachment). <br/> Maybe there is an error in the pre- or postprocessing of your data? </p>
BR replied

I have performed a simulation with the same YUSPECIF as yours (except the period: 1980-1984).
The comparison to EOBS data shows no such large errors as your results (see attachment).
Maybe there is an error in the pre- or postprocessing of your data?

View in channel
<p> Hi Burkhardt, </p> <p> thank you very much. I also found that there is no big changes in the simulation with your <span class="caps"> YUSPECIF </span> . I try to find the reason for such large biases from my post-processing. I will let you know when I find it. </p> <p> Greetings, <br/> Bo </p>

  @bohuang in #2476dbc

<p> Hi Burkhardt, </p> <p> thank you very much. I also found that there is no big changes in the simulation with your <span class="caps"> YUSPECIF </span> . I try to find the reason for such large biases from my post-processing. I will let you know when I find it. </p> <p> Greetings, <br/> Bo </p>
BH replied

Hi Burkhardt,

thank you very much. I also found that there is no big changes in the simulation with your YUSPECIF . I try to find the reason for such large biases from my post-processing. I will let you know when I find it.

Greetings,
Bo

View in channel
<p> Hi Bo, </p> <p> do you have the <span class="caps"> CRU </span> and model data at vilje or norstore? If you can send me the directory, I will have a look as well. <br/> (also compare versus <span class="caps"> EOBS </span> ) </p> <p> Best regards <br/> Andreas </p>

  @andreasdobler in #896bf85

<p> Hi Bo, </p> <p> do you have the <span class="caps"> CRU </span> and model data at vilje or norstore? If you can send me the directory, I will have a look as well. <br/> (also compare versus <span class="caps"> EOBS </span> ) </p> <p> Best regards <br/> Andreas </p>
AD replied

Hi Bo,

do you have the CRU and model data at vilje or norstore? If you can send me the directory, I will have a look as well.
(also compare versus EOBS )

Best regards
Andreas

View in channel