WEBVTT 1 00:00:02.930 --> 00:00:04.680 Fabien Maussion: I should be recording, no. 2 00:00:07.240 --> 00:00:16.190 Gerard H Roe: So, Fabien, so the IPCC attribution methodology is to take the NAT spread. 3 00:00:16.670 --> 00:00:20.890 Gerard H Roe: And say, that's the range of industrial era trends that are possible. 4 00:00:21.320 --> 00:00:28.949 Gerard H Roe: And then you compare that with the observed, or your best estimate of the forced change. 5 00:00:28.970 --> 00:00:45.420 Gerard H Roe: And it's the change divided by the spread that is the measure of, attribution. So that's what gets the temperature spread of 100% plus or minus 20% likely range. And, 6 00:00:46.620 --> 00:01:02.429 Gerard H Roe: so that I understand. I understand that the foundation for how that's done. This is a bit different, and I think it… I don't think you should be using observations, for one, because your observations are not within a model sample. 7 00:01:04.989 --> 00:01:07.619 Gerard H Roe: So, the natural spread isn't… isn't… 8 00:01:07.740 --> 00:01:17.260 Gerard H Roe: ever capturing the observed data point in this case. So, it's not that the observations was… 9 00:01:17.750 --> 00:01:31.260 Gerard H Roe: an extreme outlier from the existing model framework that you have, so what you're getting is… I think what you're getting in your blue lines is just… if you use the ensemble, you're just seeing, 10 00:01:32.060 --> 00:01:46.469 Gerard H Roe: there's an exponential, recovery to the mean climate state of the NATO ensemble, whether you start off with a larger set of glaciers or a lower set of glaciers, and that's where you get those green lines bunching up. 11 00:01:46.780 --> 00:01:50.360 Gerard H Roe: Towards the end of the simulation, right? 12 00:01:50.780 --> 00:01:51.320 Fabien Maussion: Yes. 13 00:01:51.910 --> 00:01:58.629 Gerard H Roe: Yeah. So… The, this, this, this, the standard… 14 00:01:59.330 --> 00:02:10.579 Gerard H Roe: attribution methodology. You've got two ways you can do it. You can either run a spread of forced simulations, and you get a… I'll try and do it in my hands… you get a sort of a… 15 00:02:10.900 --> 00:02:24.589 Gerard H Roe: I forget which is the right way for you with time, but you get a cloud of points which are diving towards retreats, and then you take the ensemble mean and the spread in that ensemble as being your basis for attribution. 16 00:02:24.590 --> 00:02:35.019 Gerard H Roe: The other way you can do it, which is what they actually do it for temperature, is to take the fully natural spread, and that sort of continues on over the industrial era, and then you take the observed. 17 00:02:35.020 --> 00:02:39.469 Gerard H Roe: And the attribution is the ensemble mean. 18 00:02:39.480 --> 00:02:47.749 Gerard H Roe: Minus the observed, so that's the delta. And then you're saying, without anthropogenic forcing, this is the range 19 00:02:47.780 --> 00:02:50.749 Gerard H Roe: Of retreats that can have been contemplated. 20 00:02:53.580 --> 00:02:57.360 Gerard H Roe: And so that's the philosophy of what would nature have done. 21 00:02:58.070 --> 00:03:00.080 Gerard H Roe: Without anthropogenic forcing. 22 00:03:01.240 --> 00:03:05.360 Fabien Maussion: I'm not sure to get the diff… I mean, Mattel, if you can send me the, 23 00:03:05.730 --> 00:03:12.869 Fabien Maussion: the way to compute it, because, like, the left-hand side figure really hides the range of outcomes. 24 00:03:13.140 --> 00:03:13.919 Fabien Maussion: Because this is… 25 00:03:13.920 --> 00:03:15.100 Gerard H Roe: I'll show you the… 26 00:03:15.100 --> 00:03:17.519 Fabien Maussion: This is only showing the median, basically. 27 00:03:17.520 --> 00:03:19.080 Gerard H Roe: Can I share your screen? 28 00:03:19.080 --> 00:03:19.640 Fabien Maussion: Nope. 29 00:03:30.440 --> 00:03:33.530 Gerard H Roe: Okay, so that should be seeing a PowerPoint. 30 00:03:38.680 --> 00:03:44.200 Gerard H Roe: So this is the… the basis. This is the NAT spread. 31 00:03:44.760 --> 00:03:58.190 Gerard H Roe: And then you've got the observed, and then… so this is where you say, well, the observed is about 1-ish, and then the observed spread of… sorry, the counterfactual spread of what could have happened is… 32 00:03:58.350 --> 00:04:00.600 Gerard H Roe: Plus or minus 20% of the… 33 00:04:01.000 --> 00:04:09.100 Gerard H Roe: Of the total. So that's the… that's one approach. The other approach would be to use… the spread… 34 00:04:09.610 --> 00:04:12.649 Gerard H Roe: of this, right? And then that would get you… 35 00:04:12.650 --> 00:04:14.939 Fabien Maussion: We don't have a spread of this, unfortunately. 36 00:04:14.940 --> 00:04:18.420 Gerard H Roe: That's right, that's right. So we don't have that, which is, 37 00:04:21.510 --> 00:04:34.720 Gerard H Roe: which is why… nor does… and the favored IPCC method is, this is what we've observed, this is what could have happened, how far out of the envelope of what could have happened is the observed. 38 00:04:38.500 --> 00:04:50.120 Fabien Maussion: Yeah, I mean, it's pretty much what I do. Yeah, just… I can send you the data. Actually, I think it's easier if I send you the data and you do it the way you want, because… 39 00:04:50.740 --> 00:04:54.070 Fabien Maussion: Yeah, for, for, yeah, so, 40 00:04:54.070 --> 00:04:58.340 Gerard H Roe: Oh, sorry, let me… I did… 41 00:05:06.360 --> 00:05:08.799 Gerard H Roe: So here's the equivalent of that. 42 00:05:09.720 --> 00:05:12.740 Gerard H Roe: for our little Argent share, like, glacier. 43 00:05:13.880 --> 00:05:17.459 Gerard H Roe: With a three-stage model, and, 44 00:05:17.650 --> 00:05:25.799 Gerard H Roe: the… basically, I took low-pass filters of… LMR, Berkeley Earth Observations. 45 00:05:26.710 --> 00:05:35.069 Gerard H Roe: and all of the NAT simulations, and I just added on random white noise interannual variability to each, so it's basically a measure of the low pass 46 00:05:35.200 --> 00:05:39.379 Gerard H Roe: differences, if that's… if that's clear. So I can now generate… 47 00:05:39.920 --> 00:05:49.500 Gerard H Roe: a fake ensemble around observations. So these are alternative interannual variabilities, but the same underlying… Low-pass changes. 48 00:05:49.970 --> 00:05:55.090 Gerard H Roe: So this is my effort to sort of reproduce the IPCC-like, 49 00:05:55.090 --> 00:06:01.400 Fabien Maussion: But that's, that's completely new, so that's… this we can do, so I understand that part, is you're saying, okay, 50 00:06:02.240 --> 00:06:12.939 Fabien Maussion: because of it on arrivability, things happen differently, and so if you want to spread, on the… on the east part, we can, we can do it, but I think the key difference 51 00:06:13.370 --> 00:06:14.520 Fabien Maussion: Between… 52 00:06:14.770 --> 00:06:26.209 Fabien Maussion: my results now and these here is that your grey lines, they stay more or less at zero, and they're centered at zero. And in my simulation, regardless on how you put it, is that if you start from a large glacier state. 53 00:06:26.750 --> 00:06:30.830 Fabien Maussion: In 1850, then your gray distribution in the lower part 54 00:06:31.380 --> 00:06:48.819 Fabien Maussion: is, is not the mass balance or the length, actually, in the length, is the… my distribution is going to be still not zero, basically, so that's the main difference to the two plots you showed. And this… and this makes, this makes, yeah, the most sense to me here, is that, 55 00:06:49.070 --> 00:06:54.729 Fabien Maussion: If you follow this dark blue line here, even if you plot the ensemble of outcomes. 56 00:06:55.240 --> 00:07:03.130 Fabien Maussion: They are different, they… they are, on average, like, the most of the, like, the distribution will look something like that, is that more… 57 00:07:03.310 --> 00:07:08.639 Fabien Maussion: glaciers would melt. Glaciers or glacier volume, like, we would lose glacier volume. 58 00:07:08.790 --> 00:07:24.859 Fabien Maussion: on average in the NAT ensemble. And I'm going to go back to what needle, actually, what is the needle here to actually change this result, but this is what, per default, the simulations show, and this is irrelevant to the how I calibrate OGGM. It's really a pure… 59 00:07:25.240 --> 00:07:32.889 Fabien Maussion: You can, yeah, you could have reached the same conclusion. The numbers vary a little bit, but what I show here is that the anthropogenic share 60 00:07:33.400 --> 00:07:34.400 Fabien Maussion: has… 61 00:07:34.540 --> 00:07:41.699 Fabien Maussion: to depend on the glacier states, and I think… do you disagree with that statement, or… I mean, regardless of the observations here. 62 00:07:48.980 --> 00:07:55.060 Fabien Maussion: Because this is the property of the glacier, not of the model, huh? Is that, we feed the model with temperature data. 63 00:07:55.300 --> 00:07:58.010 Fabien Maussion: And, and, 64 00:07:58.530 --> 00:07:59.860 Gerard H Roe: So, yeah. 65 00:08:00.290 --> 00:08:06.410 Gerard H Roe: So, so, let me… So if, 66 00:08:09.450 --> 00:08:17.730 Gerard H Roe: Can I… Shah… We need two screens, clearly. 67 00:08:18.890 --> 00:08:27.449 Gerard H Roe: So, that's called 2… So, if you look… 68 00:08:27.720 --> 00:08:31.680 Gerard H Roe: Within my spread of blue lines here, in the middle panel. 69 00:08:32.850 --> 00:08:37.480 Gerard H Roe: That's saying there is a spread of… 70 00:08:38.200 --> 00:08:43.139 Gerard H Roe: Possible trends that you could fit within that cloud of blue, right? 71 00:08:43.659 --> 00:08:47.540 Gerard H Roe: So I think that's what you're… that's what you're saying. If I sample the state up here. 72 00:08:48.420 --> 00:08:54.359 Gerard H Roe: It would tend towards head to the mean, but it would be a… a state up here is tending to gonna have a larger… 73 00:08:54.500 --> 00:09:04.489 Gerard H Roe: predicted anthropogenic contribution. This is going to tend to have a smaller one. One of the philosophical challenges is that if you feed this… 74 00:09:04.490 --> 00:09:08.940 Fabien Maussion: No, no, it's the other way around. If you start at the top. 75 00:09:10.740 --> 00:09:20.729 Fabien Maussion: And, because here, again, this is really… it relates… actually, it's funny, because it really relates to, to, to the assumption we started with, at the meeting. 76 00:09:21.020 --> 00:09:33.440 Fabien Maussion: is that in this experiment, and the ones I actually discarded, is that you don't test the full ensemble of outcomes, because you don't test what happened to a gray glacier at the top. 77 00:09:33.670 --> 00:09:49.610 Fabien Maussion: with another of the gray lines in the climate. You're not testing this case in your plot here. And this is what I discarded with my new permutation run, is that the question is not on the blue ensemble, it's on the green or on the gray ensemble. What happens 78 00:09:49.890 --> 00:09:57.340 Fabien Maussion: With the glacier, with the counterfactual, if you stay to the upper grey line, and then you feed it with the same climate afterwards. 79 00:09:58.140 --> 00:10:03.900 Gerard H Roe: So I… the gray line is just natural variability, right? So it should be well-bounded. 80 00:10:03.900 --> 00:10:05.010 Fabien Maussion: Yeah, better, better, better. 81 00:10:05.010 --> 00:10:05.640 Gerard H Roe: the envelope. 82 00:10:05.640 --> 00:10:10.670 Fabien Maussion: You're, you're, you're… but what I'm putting… and what I'm saying is that glaciers… 83 00:10:10.730 --> 00:10:25.889 Fabien Maussion: And this is really the important bit, is that here, in your experiment, you're… you're really in the old… you're in this state where you have the same number of grain line before and after 1850, and I'm… I'm questioning that choice, basically. 84 00:10:29.070 --> 00:10:36.529 Gerard H Roe: this is just what you would expect from a population of glaciers subject to random white noise, right? So, there are… 85 00:10:36.890 --> 00:10:40.579 Gerard H Roe: Let's say there's… let's take the upper 10 glaciers. 86 00:10:40.840 --> 00:10:52.500 Gerard H Roe: in 1850, they will experience a range of different climates, right? And then some of them will remain high, because that's just how the random numbers work out. Some of them will revert to the mean, because. 87 00:10:52.500 --> 00:11:02.039 Fabien Maussion: You don't test what happens to these 10 states, glacier states, if they were out forced with another of the gray lines climate. 88 00:11:03.170 --> 00:11:10.259 Gerard H Roe: But why would you… I mean, it's all… it's all random, so… at 1850, the glacier doesn't know what climate it's gonna get. 89 00:11:10.450 --> 00:11:11.580 Gerard H Roe: Following it. 90 00:11:12.200 --> 00:11:21.549 Fabien Maussion: Yeah, yeah, and in that case, it's even more weird that you say it's only… so the one gray line, it's only going to get one climate in your simulation afterwards, I'm just saying there should be more. 91 00:11:22.910 --> 00:11:28.530 Gerard H Roe: I guess I… so… this is just the natural distribution of glacier length fluctuations, right? 92 00:11:29.450 --> 00:11:39.139 Gerard H Roe: It's saying, absolute change, this envelope is just gonna exist. It's a… it's a statistical property of the, pre-industrial climate. 93 00:11:40.440 --> 00:11:41.020 Fabien Maussion: Yep. 94 00:11:43.300 --> 00:11:47.170 Gerard H Roe: And so what I want to compare with is how far out of that envelope am I? 95 00:11:51.990 --> 00:11:54.970 Gerard H Roe: I understand that story very clearly in my mind. 96 00:12:00.510 --> 00:12:03.209 Gerard H Roe: So I'm… that's… so I'm not… so, 97 00:12:05.720 --> 00:12:14.529 Gerard H Roe: Yeah, that's how I… that's my understanding of how the IPC… that's my definition of attribution, is, like, how far outside… what… what fraction… 98 00:12:14.930 --> 00:12:20.960 Gerard H Roe: Might be due to natural variability, and then the answer is what could be the natural 150-year trend? 99 00:12:22.560 --> 00:12:27.419 Gerard H Roe: just like we would do for warming. We'd do the same thing for warming. We'd say, 100 00:12:28.320 --> 00:12:35.320 Gerard H Roe: What's the range of natural industrial-era warings, and therefore how much might be accountable by natural? 101 00:12:35.550 --> 00:12:37.040 Gerard H Roe: And then we get a number. 102 00:12:37.830 --> 00:12:40.100 Gerard H Roe: Which give us that plus or minus 20%. 103 00:12:51.770 --> 00:12:52.450 Fabien Maussion: Yeah. 104 00:12:53.400 --> 00:13:00.650 Gerard H Roe: So I think there's different… there's different ways of… Framing the question, 105 00:13:07.740 --> 00:13:11.309 Gerard H Roe: I don't think we should, but we could generate 106 00:13:12.810 --> 00:13:19.519 Gerard H Roe: an ensemble based around observations, and we sort of know what we'd get, but I think that's going down Frankenstein territory too much. 107 00:13:20.710 --> 00:13:27.309 Fabien Maussion: Yeah, I'm not… I'm not getting what you're… what you're getting into, or what you're suggesting to do, basically, so perhaps, 108 00:13:29.160 --> 00:13:38.650 Fabien Maussion: Because this, this story, so with the exception of the extrapolation to observations, which is a bit shaky, basically, this story is the central, is the central… 109 00:13:40.540 --> 00:13:50.480 Fabien Maussion: points… of everything I've done so far, basically, is that You… 110 00:13:51.390 --> 00:14:04.970 Fabien Maussion: You have a range of natural climates from 1850 onwards, right? And this range doesn't change in what I do as simulation. It's what Give Mira gave me, and this is a range of climates which are plausible for the Alps in a natural climate. 111 00:14:07.180 --> 00:14:07.630 Gerard H Roe: Yeah. 112 00:14:07.630 --> 00:14:10.219 Fabien Maussion: And my argument is, you can't… 113 00:14:10.840 --> 00:14:28.770 Fabien Maussion: you can't tell me that the state in 1850 doesn't matter for what happens to the glacier in this range of climates. Like, just pick one climate. So we don't care even about the range, we pick one natural climate, which has happened. Let's say it would be, like, the actual perfect counterfactual. 114 00:14:30.790 --> 00:14:35.280 Fabien Maussion: How much this glacier is going to melt in the early part of the century. 115 00:14:35.560 --> 00:14:39.429 Fabien Maussion: In this one climate, depends on where it was before. 116 00:14:39.840 --> 00:14:47.550 Fabien Maussion: And this is the central key in this study, so regardless of how we compute it, this kind of thing has to happen. 117 00:14:48.790 --> 00:14:52.259 Fabien Maussion: Does it make sense to join in Mira? Like, you're quiet or something, 118 00:14:53.860 --> 00:14:56.999 Mira Berdahl: No, I'm trying to wrap my hand around it. I'm… 119 00:14:57.770 --> 00:15:00.460 Mira Berdahl: I think I understand what you're saying, Fabien. 120 00:15:05.100 --> 00:15:06.630 John Christian: So these all have the same… 121 00:15:06.630 --> 00:15:11.990 Fabien Maussion: this is… we are back to basically what we're having for, like, for, like, years now with Ben and with everyone, is that… 122 00:15:11.990 --> 00:15:27.709 Fabien Maussion: The central question… we can argue, like, on how we compute this anthropogenic part, but I really don't care about the number here, because it's very… I'm going to show it's completely uncertain. I don't… my point is that I don't think we can compute it. We can't compute it if we don't have a very clear… 123 00:15:28.500 --> 00:15:34.569 Fabien Maussion: And I can show you what other needle you can move, basically, but if you don't have a very clear 124 00:15:35.220 --> 00:15:36.970 Fabien Maussion: vision on… 125 00:15:36.970 --> 00:15:56.440 Fabien Maussion: And not vision, but actually a very clear, like, indication on how the temperature was in the 50 years before 1850, and what you do, and what you do with the contrafactual afterwards. You can't say anything meaningful about the anthropogenic fraction. Because, like, in OGGM, you have, like, a lot of plausible simulations 126 00:15:58.090 --> 00:16:03.399 Fabien Maussion: Which gives you something between 100 and SFI, for example. 127 00:16:04.800 --> 00:16:08.989 Fabien Maussion: And I'm going to make… 128 00:16:09.550 --> 00:16:17.940 Fabien Maussion: on the actual needle. In OGG, I mean, how we do our simulations right now. 129 00:16:18.130 --> 00:16:22.510 Fabien Maussion: is that I made a… I made a new simulation, so the blue… 130 00:16:22.800 --> 00:16:25.409 Fabien Maussion: Here is what Mira gave me. 131 00:16:25.600 --> 00:16:32.159 Fabien Maussion: And so this is LMR plus… and this is really temperature, so no glacial state and everything. And so… 132 00:16:32.230 --> 00:16:50.019 Fabien Maussion: here in this plot, you see a few of the things I was mentioning earlier, is that there is a little bit this issue of, in the 30-year mean, is that during the blending period, like, the viability changes a little bit, and then afterwards, we had this huge burst of viability. The only thing I did afterwards is that I said, okay, let's, 133 00:16:50.320 --> 00:16:59.649 Fabien Maussion: Let's take… the… the ensemble that was given to me, and I'm just going to shift it by 0.2K. 134 00:17:00.210 --> 00:17:01.929 Fabien Maussion: In a sense, I'm… I'm… 135 00:17:03.080 --> 00:17:13.390 Fabien Maussion: I'm questioning the blend, or I'm doing the blend differently, basically. I'm just saying, you could find another way to do the blend, and I would say, okay, my natural forcing, but no, it sits here. 136 00:17:14.030 --> 00:17:17.619 Fabien Maussion: And because, like, these tenths of a degree matter so much. 137 00:17:18.550 --> 00:17:28.369 Fabien Maussion: in the final numbers, if I do exactly what I've done so far, afterwards, I go from 52% to 75% of heterogeney fraction, again, with the huge differences. 138 00:17:28.590 --> 00:17:32.419 Fabien Maussion: And this also, in my mind, makes a lot of sense, basically, so… 139 00:17:32.940 --> 00:17:42.509 Fabien Maussion: what really matters is, for the anthropogenic fraction, is what happens to these blue lines afterwards. If you make the blue line go up like this instead. 140 00:17:42.850 --> 00:17:46.020 Fabien Maussion: Then your anthropogenic share increases as a result. 141 00:17:47.670 --> 00:17:49.150 Fabien Maussion: And, 142 00:17:49.780 --> 00:17:58.990 Fabien Maussion: And that's it. And so, the short answer is that, for this paper, at least, like, I can't, like, work for months and months on it, is that… 143 00:17:59.740 --> 00:18:10.439 Fabien Maussion: you can explore the uncertainty of this anropogenic share, and I think it's going to be huge no matter what. And I'm pretty sure it's not 100, I'm pretty sure it's somewhere… 144 00:18:10.490 --> 00:18:27.629 Fabien Maussion: It could be more, it could be less, it depends on how you define your… your… your… your signal, your signal afterwards, basically. So the anthropogenic share is extremely, extremely, is extremely, at least in how we do it right now, like, extremely sensitive to this kind of needles. 145 00:18:28.000 --> 00:18:29.260 Fabien Maussion: If that makes sense. 146 00:18:34.220 --> 00:18:41.330 Fabien Maussion: And so, what I suggest, like, moving forward, is, to show all this. 147 00:18:43.600 --> 00:18:49.740 Fabien Maussion: I acknowledge that there is a huge, a huge, a huge uncertainty in the share, or a large one. 148 00:18:49.930 --> 00:18:56.389 Fabien Maussion: And, and see what we can, we can do with this, which is a bit, like, more… 149 00:18:57.480 --> 00:19:14.429 Fabien Maussion: positive, which is, I think, like, really related to these image pairs, but we don't have time now, and I don't like to, like, go over time. So I think I'll write that now a little bit, like, a plan for the paper, and see if we agree on the main structures and the figures and everything. 150 00:19:14.600 --> 00:19:19.600 Fabien Maussion: And in the meantime, I'd be really keen on knowing how… 151 00:19:19.600 --> 00:19:20.200 Gerard H Roe: Hmm. 152 00:19:20.200 --> 00:19:26.570 Fabien Maussion: you guys want me to compute the anthropogenic share. I'm going to send you the script about how we do it, actually. 153 00:19:26.590 --> 00:19:40.930 Fabien Maussion: Or I do it, it's very simple, and the entrepreneur share is just, how the red line differs from the blue line, in, overtime, basically, so it's very easy. I'm happy to hear your thoughts about how to compute it properly. 154 00:19:40.970 --> 00:19:51.589 Fabien Maussion: And, and then from there, we can, we can… we can try to… we can try to conclude. Because, like, in my OGGM simulation, I can give you whatever, yeah, the share… 155 00:19:51.690 --> 00:19:55.690 Fabien Maussion: The share depends, depends a lot on, on where we start. 156 00:19:55.850 --> 00:19:57.650 Fabien Maussion: And it makes sense in my mind. 157 00:20:01.570 --> 00:20:02.250 Fabien Maussion: Does that sound? 158 00:20:02.250 --> 00:20:07.329 Gerard H Roe: The, left-hand figure is, 159 00:20:08.570 --> 00:20:16.700 Gerard H Roe: I think it encapsulates both… so you could put one of my little gray probability distributions on the right axis, right? 160 00:20:18.110 --> 00:20:24.110 Gerard H Roe: And that would be… that would be… that would be the band of the natural variability 161 00:20:24.840 --> 00:20:27.979 Gerard H Roe: Of industrial era changes. 162 00:20:28.670 --> 00:20:29.940 Gerard H Roe: we would… 163 00:20:33.430 --> 00:20:46.290 Gerard H Roe: There are different ways of, calculating that. So, to say length change is not the same thing as to say rate of retreat, which is another way of characterizing this. 164 00:20:46.830 --> 00:20:50.580 Gerard H Roe: But I, I think… the… 165 00:20:52.300 --> 00:21:07.770 Gerard H Roe: I think what it comes down to is that you are… broadening… The range of, starting points for… 166 00:21:07.900 --> 00:21:12.110 Gerard H Roe: the, Industrial era retreat. 167 00:21:13.610 --> 00:21:16.069 Gerard H Roe: I'm saying I could start up here, I could start down here. 168 00:21:17.930 --> 00:21:21.909 Gerard H Roe: The central estimate would remain the median of the distribution. 169 00:21:22.250 --> 00:21:28.609 Fabien Maussion: Not only the retreat, but the Nat Ensemble also differs, so if you take on… if you give… 170 00:21:31.620 --> 00:21:45.879 Fabien Maussion: just to keep it as it is, is, and if this is a small glacier in a climate which is on average the range that Mira gave me, basically, if the glacier is too small for that climate, it has to adjust, and on average. 171 00:21:46.510 --> 00:21:50.340 Fabien Maussion: across the ensemble is going to grow. And so in this scenario here on the left. 172 00:21:50.640 --> 00:21:54.910 Fabien Maussion: Manarthropogenic fraction is 120%, because Glacier would have grown. 173 00:21:55.020 --> 00:22:07.660 Fabien Maussion: If I've really defined by anthropogenic fraction as the loss, which has changed since 1850, then here it's 120%, because glacier would have grown in, in, in that world. 174 00:22:08.080 --> 00:22:10.440 Fabien Maussion: While, in that world, Glacier would… 175 00:22:11.440 --> 00:22:21.339 Fabien Maussion: would have shrunk. And again, it's really the data you gave me, huh? It's just how we start the model. So I don't change any of the climate data, the counterfactual. 176 00:22:21.530 --> 00:22:24.810 Fabien Maussion: Changes only by how glaciers are in 1850. 177 00:22:25.550 --> 00:22:31.760 Gerard H Roe: Yeah, so everything should get drugged, so the tendency is going to be back towards the median, right? The tendency is going to be to revert back to… 178 00:22:31.760 --> 00:22:48.690 Fabien Maussion: Exactly, and the two… the two end up being exactly exact, and actually, it's funny, because before the new calibration, they ended up exactly the same. It's a bit unfortunate that they don't, and it's because of hysteresis a little bit in the system, is that some small… I never see this before, and I tried to have a look at it a little bit, but… 179 00:22:49.220 --> 00:22:52.170 Fabien Maussion: Ideally, they should be the same. 180 00:22:52.330 --> 00:22:58.139 Fabien Maussion: And it's just because the model has some histories, like, I didn't know it was actually possible to find it so clearly, but some. 181 00:22:58.140 --> 00:23:01.459 Gerard H Roe: Yeah, you may have to kill some glaciers and never recover. 182 00:23:02.970 --> 00:23:22.730 Fabien Maussion: Exactly, exactly, yeah, yeah, yeah, exactly, yeah, yeah, yeah. And this… and you can pin it down, actually, you can fetch those glaciers, and they're interesting to look at, but, yeah, it was a bit easier in the previous world, where they were more sensitive to temperature, and basically they would merge, but it's, it's really like hysteresis in the model. 183 00:23:23.910 --> 00:23:32.329 Fabien Maussion: But the story holds true in both, in both cases, yeah. In all the cases, because you can still compute for each of these lines, yeah. 184 00:23:33.330 --> 00:23:34.680 Fabien Maussion: And, 185 00:23:35.470 --> 00:23:40.990 Fabien Maussion: And so for me, like, it matches my mental picture when I started the study very much. It's, 186 00:23:41.240 --> 00:23:53.579 Fabien Maussion: Of the hypothesis, basically. What, what, what I would do… From there is… 187 00:23:57.210 --> 00:24:03.520 Fabien Maussion: one of what I would do, but let's put it that way, is that I guess I hoped when we would start, that… 188 00:24:04.720 --> 00:24:06.630 Fabien Maussion: We would have… 189 00:24:07.140 --> 00:24:18.069 Fabien Maussion: a simulation which matches observation at the little ISH perfectly then, and then we would pick. And I'm still, like, tempted to do this, is to say, okay, my most likely explanation is this one, and therefore… 190 00:24:19.110 --> 00:24:31.439 Gerard H Roe: Yeah, I think that's actually… I think that's wrong to do it that way, Fabien, because you can't guarantee that the data is part of your model framework, so if you pick an outlier, you're gonna tend to skew 191 00:24:32.630 --> 00:24:36.469 Gerard H Roe: The results, and you want to have things internally consistent to the model. 192 00:24:36.690 --> 00:24:40.779 Fabien Maussion: Yeah, I agree, I agree to it, I agree, and I agree to this. 193 00:24:41.300 --> 00:24:56.779 Fabien Maussion: So the other… the other thing would be to say, so our range of attribution goes from the red line to the… to the… to this line, for example, so this gives, even a more crazy, range of outcomes, basically. 194 00:24:56.820 --> 00:25:12.299 Fabien Maussion: Another thing I'd be keen on hearing your thoughts about is about this part here. I mean, of course, it's very brutal to change the 0.2 here, but I think it boils down, and I'm going to ask Mira to send me, like, really the methods of how 195 00:25:12.300 --> 00:25:20.779 Fabien Maussion: we go from the CMIP ensemble to the LMR ensemble, but the key here, and this is… this relates also to what I know from glacier modeling, is that 196 00:25:21.230 --> 00:25:39.840 Fabien Maussion: what I call bias correction, because the net ensemble… the CMIP ensemble has surely been bias-corrected to some… to LMR, or to some… or to the 1980, or something, and this is key, basically, in changing the share, is whether you use the purple, which are plausible visually. 197 00:25:39.910 --> 00:25:48.379 Fabien Maussion: Or you use the blue, which are also plausible visually, then it's very… it's a very different outcome, basically, for glaciers afterwards. 198 00:25:48.940 --> 00:25:54.189 Mira Berdahl: Yeah, I think my problem with this that I can't quite wrap my head around is that, like Gerard's saying, I think 199 00:25:54.320 --> 00:26:11.979 Mira Berdahl: I understand why you're taking from each state at 1850 the high state or low state or whatever, and then branching your 200 from there. That kind of makes sense to me, but then at the end of it, you're… at 2024 or whatever, you're getting this really wide range, which I don't think is actually representative of what the HISNA 200 00:26:12.480 --> 00:26:17.739 Mira Berdahl: simulation distribution should be, like, that seems artificially broadened, I think, maybe, if you're taking it. 201 00:26:18.740 --> 00:26:23.150 Mira Berdahl: Like, we're comparing each one to its own self, that could be… 202 00:26:23.490 --> 00:26:27.190 Mira Berdahl: that could… I can understand that, but then when you sort of… 203 00:26:27.350 --> 00:26:32.040 Mira Berdahl: combine them all, you're sort of stacking these in a way that 204 00:26:32.210 --> 00:26:38.320 Mira Berdahl: doesn't track with what the HisNot… the original HisNot range should be. Does that make sense? 205 00:26:40.100 --> 00:26:41.770 Fabien Maussion: No, I'm not getting it. 206 00:26:41.930 --> 00:26:47.479 John Christian: Is it like you're shifting the… you're making that 1850 state the new mean of… 207 00:26:47.800 --> 00:26:50.230 John Christian: Something that starts spreading out from there. 208 00:26:50.700 --> 00:26:53.300 John Christian: Is that kind of what you're… 209 00:26:53.960 --> 00:26:54.689 Fabien Maussion: Ole, let's go back. 210 00:26:54.690 --> 00:26:56.719 John Christian: Sort of dragging them all. 211 00:26:56.720 --> 00:27:10.719 Fabien Maussion: Because this plot was before… this plot was before the new, the new permutation here, and the ensemble is the same, huh? This is really your data fed into a GGM. Is the… are you questioning the spread here, or how I interpret. 212 00:27:10.720 --> 00:27:12.750 Mira Berdahl: No, I'm not questioning that spread. 213 00:27:14.340 --> 00:27:19.390 Mira Berdahl: Which one? I'm… I'm… I'm… like, the new… the new method you have 214 00:27:20.710 --> 00:27:25.620 Mira Berdahl: I don't know what slide it was, it was another one, but you're… you had… it was the last one we were talking about. 215 00:27:25.750 --> 00:27:27.600 Mira Berdahl: You had a couple different spreads. 216 00:27:27.970 --> 00:27:29.180 Mira Berdahl: This one? 217 00:27:29.350 --> 00:27:32.739 Mira Berdahl: No… it was a time series, it was just the… 218 00:27:33.910 --> 00:27:34.610 Fabien Maussion: See ya. 219 00:27:35.520 --> 00:27:36.230 Fabien Maussion: This one… 220 00:27:36.230 --> 00:27:38.480 Mira Berdahl: The one above. Yeah, this. 221 00:27:39.730 --> 00:27:48.080 Mira Berdahl: So, like, the very last time step, Now you're getting… A range that's, like… 222 00:27:48.180 --> 00:27:50.999 Mira Berdahl: From the purple at the bottom to the blue at the top. 223 00:27:51.000 --> 00:27:57.539 Fabien Maussion: I don't think that's the new range. No, no, sorry, I haven't made myself clear, is that I changed the full ensemble 224 00:27:57.820 --> 00:28:08.160 Fabien Maussion: So I'm just doing everything as I did it before, but I shifted the nut ensemble by 2KVIN, so I'm not comparing this to this. 225 00:28:08.540 --> 00:28:17.059 Fabien Maussion: I'm comparing what I get as a result if I… if I did my attribution analysis on the blue ensemble, or if I do it on the purple ensemble. 226 00:28:17.060 --> 00:28:19.220 Mira Berdahl: Oh, okay. Okay. 227 00:28:20.040 --> 00:28:25.370 Fabien Maussion: And… and then something… and this happens then. So this is… This is… 228 00:28:25.740 --> 00:28:37.449 Fabien Maussion: with the highest glacier state, again, in 1850, and this is basically, what happened before. So we… glacier would melt a little bit in the natural case, in the natural state. 229 00:28:37.560 --> 00:28:55.330 Fabien Maussion: Because they were too big for the median… the average climate, I feed into them. And then this, in this, in this other case here, they are, they are growing, and I'm actually not even sure why my anthropogenic here, I would have to go back to my computation, basically, but my anthropogenic share should be closer to 100% here. 230 00:28:55.810 --> 00:29:05.839 Fabien Maussion: I guess… I don't know what happened… I would have to go back to my computation, because here, in my book here, this should be 100%, basically, and not 65. 231 00:29:07.690 --> 00:29:10.250 Gerard H Roe: That's because you're doing it to the Reintar, it looks like. 232 00:29:10.620 --> 00:29:11.220 Fabien Maussion: Yep. 233 00:29:11.340 --> 00:29:23.810 Fabien Maussion: Yes, exactly, right, yeah. Good memory, Richard, thanks, yeah. Indeed, yeah. That's because of the Reintale extrapolation, but, at least this dot here, this one. 234 00:29:23.970 --> 00:29:27.349 Fabien Maussion: goes exactly. This one would go from 75… 235 00:29:28.290 --> 00:29:28.930 Gerard H Roe: Hmm. 236 00:29:31.510 --> 00:29:38.260 Fabien Maussion: the percent, if I shift the AI is wrong or right. I'm just saying that whatever you do. 237 00:29:39.070 --> 00:29:45.889 Fabien Maussion: And this is the same for our projections, is that the bias correction is extremely determinant to, of the. 238 00:29:51.600 --> 00:29:53.650 Gerard H Roe: You are breaking up, Fabien. 239 00:29:55.480 --> 00:29:57.259 Fabien Maussion: Oh, so good now. 240 00:30:00.030 --> 00:30:02.530 Gerard H Roe: You're… 241 00:30:03.420 --> 00:30:04.110 Fabien Maussion: Yeah, I'm sorry. 242 00:30:04.110 --> 00:30:05.089 Gerard H Roe: The audio is choppy. 243 00:30:09.020 --> 00:30:09.530 Fabien Maussion: Yep. 244 00:30:10.520 --> 00:30:28.000 Fabien Maussion: Maybe it's mine? Anyways, it doesn't matter, because I think it's 6 now, and I really need to get back to my family. It's been a lot of information today. I'm sending my… I'm going to send the slides from the talk with the written conclusions, which don't change much with the new, 245 00:30:28.710 --> 00:30:37.219 Fabien Maussion: with the new simulations anyway, and then the slides are presented today, and we should meet again soon. I'd like you to… 246 00:30:37.220 --> 00:30:48.880 Fabien Maussion: think… I'll add some methodological parts to a written document, so you can actually challenge what I've done, because I'm really interested in, in you guys, challenging all this. 247 00:30:49.000 --> 00:30:56.519 Fabien Maussion: And we should meet again in a week or so, so that you have time to digest everything. 248 00:30:56.670 --> 00:31:00.350 Fabien Maussion: And I can react as well to whatever thoughts you have on the methods. 249 00:31:02.850 --> 00:31:07.310 Gerard H Roe: Do we have access to the… Data. 250 00:31:08.510 --> 00:31:15.499 Fabien Maussion: Yeah, I can send it, you do… you do… you don't right now, but it's, it's on the link online, and I can send you the data, yeah, for sure. 251 00:31:19.090 --> 00:31:20.920 Fabien Maussion: Mmm… 252 00:31:20.920 --> 00:31:23.380 Mira Berdahl: Do you want to choose a time to meet now? 253 00:31:23.540 --> 00:31:25.040 Fabien Maussion: Yeah, if that works. 254 00:31:26.520 --> 00:31:42.400 Fabien Maussion: Is in a week's time okay? I'm not… I don't think I'll have much time to do much more work on it, but I think, like, it is a… I can… I can at least write down the methods, like, the core of the methods, I can write it down, and and then you can have a think by then, or should we meet later? 255 00:31:44.000 --> 00:31:47.089 Fabien Maussion: Thursday would also be okay, the 12th, for example. 256 00:31:48.930 --> 00:31:52.619 Mira Berdahl: Thursday's tricky for me in the morning, I have a doctor appointment. 257 00:31:52.620 --> 00:31:59.329 Fabien Maussion: Okay, no, that's okay. Monday, then? Is it okay? 258 00:32:01.260 --> 00:32:02.280 Mira Berdahl: I could do Monday. 259 00:32:02.530 --> 00:32:04.060 Gerard H Roe: Monday the 9th? 260 00:32:04.310 --> 00:32:04.890 Fabien Maussion: Nope. 261 00:32:08.150 --> 00:32:14.780 Mira Berdahl: I'd have to be at 8.45 again, because my husband's away. We could do that. 262 00:32:15.210 --> 00:32:16.140 Mira Berdahl: Yam. 263 00:32:16.870 --> 00:32:17.760 Fabien Maussion: networks. 264 00:32:21.340 --> 00:32:22.739 Fabien Maussion: Does it work for everyone? 265 00:32:23.510 --> 00:32:24.310 John Christian: Yep. 266 00:32:30.340 --> 00:32:31.490 Fabien Maussion: Perfect. 267 00:32:32.190 --> 00:32:34.890 Gerard H Roe: Okay, so, Fabien, I think in… 268 00:32:35.080 --> 00:32:39.830 Gerard H Roe: reciprocity, so I, I, I'm, I'm, 269 00:32:41.360 --> 00:32:53.330 Gerard H Roe: clear about what the standard IPCC attribution methodology is, and that's what we are anchoring around. That's the, there is a cloud of what-ifs without anthropogenic forcing. 270 00:32:54.690 --> 00:32:59.120 Gerard H Roe: And then there's a comparison of what the industrial era changes have been relative to that cloud. 271 00:33:00.980 --> 00:33:03.249 Gerard H Roe: So that's the standard protocol. 272 00:33:03.550 --> 00:33:04.080 Fabien Maussion: Nope. 273 00:33:04.840 --> 00:33:07.020 Gerard H Roe: I think you're bringing up that 274 00:33:07.800 --> 00:33:12.100 Gerard H Roe: The initial condition might change what the estimate is. 275 00:33:12.100 --> 00:33:12.670 Fabien Maussion: Yep. 276 00:33:12.820 --> 00:33:19.750 Gerard H Roe: The central estimate will remain what's due to observations, so that's how it works in that. 277 00:33:19.870 --> 00:33:29.439 Gerard H Roe: Protocol, so… and that… that's built into the uncertainties, or it should be built into the uncertainties of what the natural simulations would have done anyway. 278 00:33:29.720 --> 00:33:30.380 Gerard H Roe: Yeah. 279 00:33:34.280 --> 00:33:35.350 Fabien Maussion: Exactly. 280 00:33:39.030 --> 00:33:44.580 Gerard H Roe: So I think that one figure is where… the, 281 00:33:45.590 --> 00:33:50.939 Gerard H Roe: differences from the IPCC methods need to be understood. 282 00:33:51.980 --> 00:33:52.760 Fabien Maussion: Yep. 283 00:33:54.030 --> 00:33:56.740 Fabien Maussion: Or well presented, I think, that's fine. 284 00:33:57.450 --> 00:34:07.249 John Christian: And the… the only difference here is between… essentially, in what gives you those different initial states, Fabien, is the… is the synthetic… 285 00:34:07.460 --> 00:34:10.880 John Christian: Inter-annual variability in the decades before, right? 286 00:34:11.040 --> 00:34:12.210 Fabien Maussion: Yeah, yeah. 287 00:34:12.210 --> 00:34:13.679 John Christian: I mean, is… 288 00:34:13.969 --> 00:34:25.120 John Christian: And so the question is, sort of, were we on a bit of a cooler excursion or a bit of a warmer excursion? Is there… is there a way to bring in, like, long… like, older… 289 00:34:25.860 --> 00:34:31.690 John Christian: instrumental data sets? Like, maybe not into the modeling framework, but does that… 290 00:34:32.520 --> 00:34:38.439 Fabien Maussion: That's a very good point, and I'm glad… and so I think that the key… 291 00:34:39.699 --> 00:34:48.839 Fabien Maussion: That's exactly the point. That's exactly the point, is that all these gray lines are possible 292 00:34:49.159 --> 00:34:55.749 Fabien Maussion: pasts, because the LMR is too smooth, and these are basically possible ones, and 293 00:34:56.190 --> 00:35:02.039 Fabien Maussion: And, and all of them are equally plausible in this model framework, but 294 00:35:02.150 --> 00:35:20.679 Fabien Maussion: But, nature has picked something… some trajectory in it, and the question is, are we? Is one of them, like, following the nature's trajectory or not? And this is where observations could fit in. For example, like, a naive, statement would be to say. 295 00:35:21.260 --> 00:35:25.899 Fabien Maussion: If you, if you give me a much wider ensemble. 296 00:35:26.100 --> 00:35:36.220 Fabien Maussion: before, if you did diverge from LMR in much crazier ways, for example, then I can tell you, like, one of these lines is going to fit the observations, for example. 297 00:35:38.560 --> 00:35:47.000 Fabien Maussion: And, and, and we have some evidence from, from, 298 00:35:47.770 --> 00:35:56.100 Fabien Maussion: and then… but still, I agree with what you said, Gerard, is that I'm not necessarily saying that in this model framework. 299 00:35:57.520 --> 00:36:08.560 Fabien Maussion: the question is more how far you are from the median, basically, and how the little ice age was in comparison. If you can't answer that question, and we can't, John, I think. 300 00:36:08.900 --> 00:36:28.280 Fabien Maussion: Is that, is that, is that… if you can't answer that question, then you can answer the… what happens afterwards, in a way, or you can try to answer, because we also don't have the counterfactual, but the key really is how far the glaciers are in the equilibrium, and this is where we started, how far our glaciers are from, 301 00:36:28.670 --> 00:36:32.869 Fabien Maussion: From what they… what was before in 1850, in a way, yeah, yeah. 302 00:36:33.970 --> 00:36:52.400 Gerard H Roe: Well, right, so that's Mira's reference mass balance kind of does that with the different, reconstructions of past millennium. The other thing I just want to raise quickly, though, is that, mass balance is a different measure of attribution, and what you're picking up on is a reversion to the mean. 303 00:36:52.870 --> 00:37:03.119 Gerard H Roe: in length, but there's a natural timescale of that which is causing the differences in the initial conditions. So we found under those circumstances, that if we look at mass loss. 304 00:37:03.580 --> 00:37:16.520 Gerard H Roe: we actually get something centered on 100%, because if the temperature's 100%, then the mass loss is 100%. So it's less sensitive to initial conditions, and that may be a lesson that we need to communicate here. 305 00:37:16.520 --> 00:37:25.210 Fabien Maussion: Yeah, yeah, and I think that's a key lesson. The only, the only thing is that if we go back to what drove, 306 00:37:25.540 --> 00:37:32.439 Fabien Maussion: the… the… I think my personal motivation is what you brought up with the pictures, Gerard, is that… 307 00:37:34.390 --> 00:37:43.100 Fabien Maussion: the key question is, is, it boils down to this plot, is what did climate do after 1850? Like, is it… 308 00:37:44.630 --> 00:37:47.530 Fabien Maussion: And again, the blue lines you gave me. 309 00:37:47.900 --> 00:38:00.520 Fabien Maussion: are actually already part of the… we don't need much, like, we don't even need, like, a glacier model to see that something happened before 1815. Afterwards, there is a warming in the ensemble you gave me, and this is enough. 310 00:38:01.010 --> 00:38:10.010 Fabien Maussion: to already explain a lot of the part, and then the question is, how much do Glacia respond to this warming after a few… just… if we just look at the central part. 311 00:38:10.460 --> 00:38:21.089 Fabien Maussion: we have a bit of a cooling before 1850, and then we have a bit of a warming, and this is enough. If you feed this circulation model, it's maybe not going to be the crazy numbers I had, like 50%. 312 00:38:21.220 --> 00:38:36.889 Gerard H Roe: That's right. So I'm happy with an attribution of 80% on length, because I do see that warming, and so there's… that's going to be carried on. But I think… so, in order for those pictures to be supported, what you need is a scientific statement saying, we are very sure 313 00:38:36.890 --> 00:38:43.319 Gerard H Roe: that most of what we see as the difference between these two landscapes is anthropogenic warming. So… 314 00:38:43.320 --> 00:38:47.870 Fabien Maussion: would be given by that line, for example, like, we are able to stand in the model. 315 00:38:47.870 --> 00:39:03.000 Gerard H Roe: So, it's not necessarily about when you cross 100%, it's about what's the primary cause of the difference between those two landscapes. That, I think, would be the most useful information to deliver. That's what everyone is implicitly talking about. 316 00:39:03.000 --> 00:39:04.320 Fabien Maussion: No, exactly, and I'm not… 317 00:39:04.760 --> 00:39:11.360 Fabien Maussion: Good to look at area more than volume, actually, because, because the pictures are actually area and not volume, and therefore. 318 00:39:11.360 --> 00:39:17.749 Gerard H Roe: Well, I would also look at… I would also look at mass loss, because you have thinning as well as retreat, 319 00:39:17.750 --> 00:39:24.720 Fabien Maussion: Yeah, but this is what mass… this is mass loss, and this is total volume change, so it's, it's mass loss, so it's, 320 00:39:24.860 --> 00:39:32.330 Fabien Maussion: It's, it's not just… it's just not the annual mass balance, it's the… the sum of the mass balances with time, yeah. 321 00:39:32.330 --> 00:39:33.130 Gerard H Roe: Yeah. 322 00:39:34.080 --> 00:39:44.750 Fabien Maussion: This is basically also another interesting result of this, which we didn't talk about, is that, yeah, so put aside, like, the differences in the temperature average and everything. 323 00:39:44.990 --> 00:39:50.340 Fabien Maussion: Is that, the temperature signal gets damped. 324 00:39:50.530 --> 00:40:07.530 Fabien Maussion: by, glaciers, when it becomes extremely… the temperature signal becomes extremely strong in the signal-to-noise ratio. It is followed by mass balance very clearly over time, but then this, they diverge after some time, because glaciers have already reacted, and they're smaller, and the smaller they get. 325 00:40:08.880 --> 00:40:17.610 Fabien Maussion: The less, the less, they are linked to… because the smaller they get, basically. 326 00:40:17.660 --> 00:40:31.140 Fabien Maussion: the more they have already retreated to areas where temperature matters less, basically, and so therefore they diverge a little bit, these two lines, after some time. And at some point, if glaciers have disappeared, then this line becomes… there is no signal anymore, basically, because 327 00:40:31.730 --> 00:40:32.430 Fabien Maussion: Yeah, that's… 328 00:40:32.430 --> 00:40:35.319 Gerard H Roe: There are limits as a function of science, yeah, for sure. 329 00:40:35.320 --> 00:40:36.320 Fabien Maussion: Yeah, yeah. 330 00:40:36.680 --> 00:40:47.989 Fabien Maussion: Okay, cool, yeah, thanks. It's really, really super exciting. I'm going to follow up with written statements and documents so that we can actually all have a think 331 00:40:48.400 --> 00:40:53.789 Fabien Maussion: about what's done correctly and wrongly in what I've showed today, and what can be done differently. 332 00:40:53.960 --> 00:41:07.300 Fabien Maussion: And, in turn, so that's for you, so you… I'll write something for you to read. I won't come to it until… yeah, I… yeah, we have a meeting on Monday, but I'll try to do it by Wednesday, so that you have some time to digest. 333 00:41:07.600 --> 00:41:15.380 Fabien Maussion: And in turn, I'll ask… Mira to write down the methods of… 334 00:41:16.260 --> 00:41:21.750 Fabien Maussion: Of, of the blending here, basically, and so for me to really understand what's going on there as well. 335 00:41:22.440 --> 00:41:24.740 Mira Berdahl: Yeah, and if you could send the link to the… 336 00:41:24.910 --> 00:41:25.460 Fabien Maussion: the data. 337 00:41:25.460 --> 00:41:26.529 Mira Berdahl: That would be great. 338 00:41:27.060 --> 00:41:27.950 Fabien Maussion: I'll wait up. 339 00:41:30.070 --> 00:41:31.010 Fabien Maussion: Excellent. 340 00:41:31.220 --> 00:41:32.110 Fabien Maussion: Good luckily. 341 00:41:33.100 --> 00:41:34.020 Mira Berdahl: Thank you. 342 00:41:34.020 --> 00:41:40.090 Fabien Maussion: Thank you as well, and lots to think about on a Monday morning, and I'm ready to… Bye-bye, everyone!