EWGLAM Meeting 2021 - Day 4 szeptember 30, 2021 (10:16) martina tudor: i see blue screen (10:17) Roger Randriamampianina: I don't see the presentation (10:17) Susanna Hagelin (SMHI): same here (10:30) Ekaterina Kurzeneva, FMI: @Marion: did you look at the lake surface temperature? (10:32) Dmitrii Mironov: Wim Thierry (10:33) Dmitrii Mironov: Sorry, he is Wim Thiery (10:34) Ekaterina Kurzeneva, FMI: Thank you Dmitrii! (10:36) Balazs Szintai - C-SRNWP: resume at 10:50 CEST (10:40) Ekaterina Kurzeneva, FMI: @Marion, a publication about thunderstorms over Lake Victoria: https://www.nature.com/articles/ncomms12786 (10:45) Marion Mittermaier: @Ekaterina, thank you! (10:52) Kristian Pagh Nielsen: Looks good (11:09) Florian Meier: Do you face stability issues with 1m lowest model level? What is the timestep in this case? (11:40) Dmitrii Mironov: Bogdan, did you see the effect of wind-farm parameterization on the scalar fields? (11:42) Anke Finnenkoetter: Do you need quite detailed data about existing wind farms to get a good result from the windfarm parametrisation and how difficult have you been finding it to get hold of the relevant input data? (11:42) Florian Meier: There is not much impact on other variables except slight improvment of ff10m and precipitation, but the impact is not really significant. (12:00) seity: @Anke: How to you initialize your 2-moments microphysics in terms or aerosols ? (12:02) Dmitrii Mironov: Anke, can you say what makes RAL3 better, improved treatment of clouds, or the aerosols, or both? (12:04) seity: do you have compared your microphysics in terms of 'object scores': number of cells, size, max reflectivities... ? (12:05) Linda Schlemmer (DWD): @Anke: does the bi-modal cloud scheme include a treatment of the ice phase, or is it liquid only? (12:06) Dmitrii Mironov: Thanks, Anke! (12:07) seity: Thanks for the answers. (12:07) Linda Schlemmer (DWD): thanks! (12:31) Balazs Szintai - C-SRNWP: we resume at 13:30 CEST for the poster session (13:01) Juan Jesus Gonzalez Aleman (AEMET): Balazs (13:21) martina tudor: Breakout Session on Upper-air Physics (chair: Mike Bush) https://bluejeans.com/532243017/7147 (13:43) Xiaohua Yang: @Balazs: on your question about physics in an on-deman 150 m model: our experience is that for orographically enforced extreme wind forecast, physical parameterisation maybe of secondary importance. But of course this is very unlikely the case in ordinary situations. (13:56) Jean-François MAHFOUF: is it the data assimilation breakout session ? (13:56) Jean-François MAHFOUF: already started ? (13:57) Maria Derkova: @JFM: in 5 minutes (13:57) Jean-François MAHFOUF: Thanks (14:03) Balazs Szintai - C-SRNWP: Thanks a lot, Xiaohua! (14:10) Siebren de Haan: @roger: i am present... (14:11) Roger Randriamampianina: Good @siebren (14:11) Roger Randriamampianina: @JFM: did you do specific tuning to go to 50km thinning of ascat? (14:12) Florian Meier: Is the new thinning similar to what is done for aircraft and radar (shifted thinning boxes)? (14:13) Isabel Monteiro: For ASCAT, are you doing across-swath and along swath thinning, in the same way? (14:17) Tsyrulnikov Michael: Maybe, it is low Horizontal resolution of GNSS-RO that caused their little impact in your system? (14:18) Florian Meier: Do you use for GNSS-RO 1D or 2D operator and if 2D which size of model region? (02:19 PM) Klaus Stephan (DWD): Do you have any idea, why precipitation becomes better, with GNSS-RO assimilated? (02:24 PM) Benedikt Strajnar: Was it possible to see any impact/feedback of GNSS-RO on bias correction of radiances in your experiments? (02:24 PM) Isabel Monteiro: Thank you very much! (02:26 PM) Jean-François MAHFOUF: Improvement of precipitation likely to come from improved of UTLS temperature analysis. (02:27 PM) Jean-François MAHFOUF: We mostly take varBC from ARPEGE but we have not been looking at SEVIRI behaviour. It should help (02:28 PM) Jean-François MAHFOUF: We use 1D GNSS-RO obs operator. It would be of interest to examine a 2D operator, but it has a cosr :( (02:28 PM) CP: @Florian: in complement to JFM answer, the shifted boxes are also applied for scatt data, but only as backup, the main thinning being done by the selection on the (regular) observation grid as described by JFM. (02:29 PM) Florian Meier: Thank you, good progress. (02:30 PM) Klaus Stephan (DWD): @JFM: Sorry, was does UTLS means? (02:30 PM) CP: @Isabel: I am confirming, the scatt thinning is done in the both directions of swath in a similar way (02:31 PM) Jean-François MAHFOUF: Upper Troposphere Lower Stratosphere = UTLS (02:31 PM) Klaus Stephan (DWD): 👍 (02:38 PM) Isabel Monteiro: @CP merci beaucoup😊 (02:43 PM) Martin Leutbecher: Michael, very interesting talk. Could you say something about the inter-variable and vertical correlations of the random field in AMPT? (02:46 PM) Klaus Stephan (DWD): Spread at beginning of forecast seems very low. It that because of COSMO-LEPS conditions? (02:47 PM) Tsyrulnikov Michael: Thank you, Martin, we ended up with the vetical length scale of some 2.5 km length scale. (02:47 PM) Tsyrulnikov Michael: Yes, Klaus, that was the reason. (02:48 PM) Klaus Stephan (DWD): Thank you (03:01 PM) Martin Leutbecher: @Michael, I would also be interested to learn about the standard deviation of the AMPT random fields (I assume they are dimensionless like in SPPT). (03:02 PM) Jean-François MAHFOUF: Thanks Siebren. ANy plan for E-GVAP to distribute STD data ? (03:02 PM) Klaus Stephan (DWD): Hi Siebrien, thanks for the nice presentation. Do you see any impact on precipitation? (03:03 PM) Jana Sanchez: Thanks @Siebren, nice results!! (03:04 PM) Claude F: The loss of information of the STD obs after 12h, if due to the impact of LBCs, could then also be visible in other sorts of sensitivity experiments (+/- obs types or changing ICs). Is there any such other evidence ? (03:05 PM) Claude F: thanks for nice talk & the work with Martin. (03:06 PM) Klaus Stephan (DWD): Thanks Siebrien (03:06 PM) Tsyrulnikov Michael: Martin, the random field has exactly unit variance. It's multiplied by 0.75 for T,u,v and 0.5 for qv,qc,qi. Then the result is multiplied by the area-averaged physical tendency (global LAM averaging for T,u,v,T_soil and moving-window (radius 50 km) averaging for W_soil, qv,qc,qi. (03:07 PM) Martin Leutbecher: thank you (03:18 PM) Magnus Lindskog: Nice talk Siebren! How did you assign sigmao? (03:28 PM) Florian Meier: Do you expect the difference to be bigger if no LHN is applied in both experiments? (03:29 PM) Florian Meier: Thank you nice talk! (03:33 PM) Klaus Stephan (DWD): @Florian, LHN depends strongly at microphysical settings. Differences between 2Mom and 1Mom are very intense. LHN needs to be optimize. However, as Kobra already mentioned, results are not changing too much. Without LHN 2Mom will be a bit more comfortable. 😉 (03:42 PM) Claude F: @Kobra: in the hourly RUC , do you cycle all humidity fields (eg all prognostic variables from the microphysics) ? (03:44 PM) Kobra.Khosravian_dwd: @Thank you Claude for your question. Actually, I don't do this experiment since I am not sure about it but as I know all the first guess comes from 1-mom microphysics. (03:51 PM) Claude F: what is the typical time range of impact of LHN in the forecast ? (I kind of remember this used to be 2-3h from talks of the last few years). What are your opinions ? (03:52 PM) Jean-François MAHFOUF: @Kobra : linked to Claude's question, what are the variables that you initialise regarding microphysics ? (03:54 PM) Klaus Stephan (DWD): For ICON it nowadays depends strongly at the time when forecast is starting. At night, we still have 3 hour, followed by some hour of detrimental effects. But 12 UTC forecast, we find even 10 to 12 hour improvement in precipitation (03:54 PM) Florian Meier: We find something like +3h, but on very small domain. (03:56 PM) Bruce Macpherson: Claude, a recent Met Office study showed 4-5hr impact from latent heat nudging. (03:57 PM) Kobra.Khosravian_dwd: @Jean-François: some variables which I could mention is: T, qv, qc, v, w, qi, qr, qg, qs, tke, ... (03:59 PM) Jean-François MAHFOUF: @Kobra: Thanks ! no number concentration quantities that are needed by the 2 moment scheme ? (04:01 PM) Claude F: (about LHN still ...): that suggests that "glueing" LHN (in an appropriate design) with a VAR or EnVar assimilation could provide a sensible seamless DA for NWC & NWP ranges. What are others' opinion on this ? (04:02 PM) Claude F: thank you Roger for the chairing ! (04:03 PM) Kobra.Khosravian_dwd: @Jean-François: actually I am not sure about the number concentration! prefer not to say a solid answer :)