1
00:00:01,380 --> 00:00:09,610
Good afternoon, everybody, and welcome to the Oxford Climate Research Network's annual lecture, co-hosted by the Oxford Martin School.

2
00:00:09,610 --> 00:00:16,600
The Oxford Climate Research Network aims to pull together all those across Oxford whose research touches on climate,

3
00:00:16,600 --> 00:00:22,230
and the goal broadly is to help us be more than the sum of our parts. So over the last year,

4
00:00:22,230 --> 00:00:30,360
we've brought in several million pounds worth of grants focussed on understanding the climate system on impacts policy mitigation.

5
00:00:30,360 --> 00:00:38,070
Several members of the network contributed to the IPCC 1.5 degree report.

6
00:00:38,070 --> 00:00:42,870
Others provided data for the BBC's Earth from Space programme.

7
00:00:42,870 --> 00:00:52,590
We continue to engage with the Met to co-fund many of our activities and with the United Nations Framework Convention on Climate Change Process,

8
00:00:52,590 --> 00:01:01,950
as well as reaching out to local organisations and schools. So it's really great to see so many people from a wide range of departments here today.

9
00:01:01,950 --> 00:01:07,950
If you're a research student or a member of academic staff at the university and not already a member of a vocal,

10
00:01:07,950 --> 00:01:17,490
and please do go to our website, which is Climate Doacs, Dot Dot UK and sign up for our mailings.

11
00:01:17,490 --> 00:01:24,750
And after today's talk, there'll be a reception just next door and hope they're all able to stay on a network over a drink.

12
00:01:24,750 --> 00:01:33,010
Get to know each other better. It's a huge pleasure for me today to introduce today's speaker, Professor David Battisti.

13
00:01:33,010 --> 00:01:39,000
Is the Tomoaki Endowed Chair of Atmospheric Sciences at the University of Washington in Seattle.

14
00:01:39,000 --> 00:01:43,790
David's research focuses on understanding natural variability of the climate system.

15
00:01:43,790 --> 00:01:49,220
He's especially interested in understanding how interactions between the ocean, the atmosphere,

16
00:01:49,220 --> 00:01:58,230
land and sea ice lead to variability in climates on timescales ranging from months to several decades.

17
00:01:58,230 --> 00:02:04,600
He also studies the impact of climate variability and climate change on global food security.

18
00:02:04,600 --> 00:02:11,200
He served on numerous international scientific committees, co-authored many international science plans and reports,

19
00:02:11,200 --> 00:02:15,850
published well over 100 papers in atmospheric sciences and oceanography.

20
00:02:15,850 --> 00:02:20,320
Davies is also very passionate about training the next generation of climate scientists,

21
00:02:20,320 --> 00:02:25,930
and as such, he's received many awards for both his research and his teaching.

22
00:02:25,930 --> 00:02:28,750
In addition to his responsibilities at the University of Washington,

23
00:02:28,750 --> 00:02:35,830
David is a fellow at the Food Security Institute at Stanford University and an adjunct professor at the University of Bergen in Norway.

24
00:02:35,830 --> 00:02:39,190
And it's a real honour to have him with us today to deliver our annual lecture.

25
00:02:39,190 --> 00:02:44,980
Today, he's going to talk about the relationship between global climate sensitivity and regional warming.

26
00:02:44,980 --> 00:03:00,360
So if you join me in welcoming David to the state. Thanks, everyone, for coming today, and thank you for inviting me to your community yet again.

27
00:03:00,360 --> 00:03:07,510
And one of these days, you'll probably get tired of me, but I'm never going to get tired of you and.

28
00:03:07,510 --> 00:03:14,590
This talk was kind of inspired by last year I was asked to go to him at this meeting at MIT and talk about the value of regional downscaling,

29
00:03:14,590 --> 00:03:21,070
I don't do regional downscaling, I do a little bit of it, but it's not my it's not my forte.

30
00:03:21,070 --> 00:03:23,860
And so I kind of flipped that around and say, Well,

31
00:03:23,860 --> 00:03:33,820
where could you actually use regional downscaling to kind of either improve your predictions or reduce the level of uncertainty in the predictions?

32
00:03:33,820 --> 00:03:40,910
And I was a little bit surprised by this, and it gives a sense insight on what's controlling regional climate variability and help.

33
00:03:40,910 --> 00:03:44,800
The one thing you come away at the end of the day is that most of what controls regional

34
00:03:44,800 --> 00:03:50,610
climate variability everywhere on the planet is one thing which I won't tell you yet.

35
00:03:50,610 --> 00:03:55,390
OK, OK, so there's the outline. You've been staring at it for a while, so this is going to go ahead.

36
00:03:55,390 --> 00:04:00,670
Climate sensitivity it's defined for the purpose of today is the global average temperature

37
00:04:00,670 --> 00:04:05,650
change due to a doubling of atmospheric carbon dioxide and any weight long enough.

38
00:04:05,650 --> 00:04:12,750
So the climate system is equilibrated. And the answer, according to the last IPCC report, was the best guess.

39
00:04:12,750 --> 00:04:19,030
The answer is three point two degrees Celsius. But there's a huge range of uncertainty in that there's climate models that

40
00:04:19,030 --> 00:04:23,980
give you as low as little bit over two and some that give you almost fives.

41
00:04:23,980 --> 00:04:29,390
The range is basically 100 percent three point two plus or minus 50 percent.

42
00:04:29,390 --> 00:04:39,640
And I used to think. Who cares? Nobody lives on the global mean temperature, so what what how useful is that piece of information?

43
00:04:39,640 --> 00:04:47,500
And let's go back to this at the end of the talk. So if you're a climate specialist, do not answer this question.

44
00:04:47,500 --> 00:04:50,920
What's the major uncertainty in the global average climate?

45
00:04:50,920 --> 00:04:55,330
If you look at the global average temperature change you do, what's the main reason for uncertainty?

46
00:04:55,330 --> 00:05:02,440
There were no, not a climate specialist. There's one thing I mean, obviously, we're seeing the CO2 is doubled,

47
00:05:02,440 --> 00:05:07,000
so it's not a human problem, they're not worried about how much emissions is going to have. Let's say it did double.

48
00:05:07,000 --> 00:05:13,500
There's this factor of two uncertainty and how much warming you're get the global average with clouds, clouds, right?

49
00:05:13,500 --> 00:05:17,560
And in particular, tropical clouds. Right? OK, good.

50
00:05:17,560 --> 00:05:22,990
So it's tropical clouds. Here's a figure from the last IPCC, and he's one of these dots is for climate modelling.

51
00:05:22,990 --> 00:05:27,160
Just tells you like how much feedback you get? Yes. Oh no.

52
00:05:27,160 --> 00:05:32,260
How much feedback do you get under different things? Planck is just the black body radiation from the surface.

53
00:05:32,260 --> 00:05:39,880
There's very old models do that the same. Then there's the long wave and water vapour feedback combined give you very little uncertainty in the end.

54
00:05:39,880 --> 00:05:46,030
But the biggest uncertainty here, the biggest spread is in what happens to clouds and in particular, claptrap of clouds.

55
00:05:46,030 --> 00:05:49,000
And there's a little bit of uncertainty due to libido.

56
00:05:49,000 --> 00:05:56,710
OK, so if you say what's the common pattern across all the models, you take all the models responses to an increase in CO2.

57
00:05:56,710 --> 00:06:03,310
What's the common response? Here's a picture from the last IPCC.

58
00:06:03,310 --> 00:06:05,230
I grabbed 18 climate models and just said,

59
00:06:05,230 --> 00:06:09,970
What's the difference in the annual mean temperature at the end of this century compared to the end of last century?

60
00:06:09,970 --> 00:06:18,530
So this is one hundred year difference. OK, and we'll call the global average change this bracketed Delta T here, and that's the answer.

61
00:06:18,530 --> 00:06:23,930
Global average is three point seventy three degrees C for this group of 18 models at

62
00:06:23,930 --> 00:06:28,020
the end of this century due to a kind of a business as usual emissions scenario.

63
00:06:28,020 --> 00:06:36,950
Not every place forms the same. The main thing you see in this thing is the tropics don't warm as much as the Arctic right,

64
00:06:36,950 --> 00:06:42,770
and the land warms more than the ocean, and that's kind of washed out there.

65
00:06:42,770 --> 00:06:51,080
But the typical global average three point seventy three to typical number in mid-latitude land area is more like five degrees C over.

66
00:06:51,080 --> 00:06:59,990
The ocean is more like three degrees C, and the polar regions is 12 degrees C, OK?

67
00:06:59,990 --> 00:07:08,580
Wise or wise or polar amplification, why does it why did the poles warm more than the oceans?

68
00:07:08,580 --> 00:07:13,680
There's a common answer that everybody says, which is actually not the right answer.

69
00:07:13,680 --> 00:07:17,010
The answer is there's the ice albedo feedback, right?

70
00:07:17,010 --> 00:07:23,940
You get warm, you melt ice and and the melting ice allows more sunlight to be absorbed.

71
00:07:23,940 --> 00:07:27,000
But that's not it. OK. There's actually two.

72
00:07:27,000 --> 00:07:32,640
The two most important things here are the fact that when you have this plank feedback, the BlackBerry feedback,

73
00:07:32,640 --> 00:07:39,120
you're starting from the very cold temperature in the Arctic in a very warm temperature, high temperature in the in the tropical regions.

74
00:07:39,120 --> 00:07:41,020
So if I just need to warm a little bit,

75
00:07:41,020 --> 00:07:46,200
the tropics to get the same amount of energy back out is the amount of warming I need to do in the polar regions.

76
00:07:46,200 --> 00:07:49,590
So that's that's about a quarter of the answer.

77
00:07:49,590 --> 00:07:56,970
The other quarter of the answer is I put the same amount of energy down everywhere on the planet due to increase in CO2 and in the tropics,

78
00:07:56,970 --> 00:08:03,930
it's really warm. I get some of the energy to evaporate, which means I don't warm as much because I use some of that energy to evaporate.

79
00:08:03,930 --> 00:08:08,280
That vapour goes to the atmosphere and the atmosphere moves that vapour to the pole, and that's where it condenses.

80
00:08:08,280 --> 00:08:15,870
So I have this transfer of energy. It's been absorbed by Joseph, you're on the microphone is cracking, can I swap?

81
00:08:15,870 --> 00:08:21,180
Oh yes, sir. It's a transfer of energy that's been absorbed in the tropics,

82
00:08:21,180 --> 00:08:25,350
that's been moved to the poles by atmospheric circulation since another quarter of the answer.

83
00:08:25,350 --> 00:08:32,720
And I want to tell you what the other half is. Yeah, what I could tell you, but it's not obvious.

84
00:08:32,720 --> 00:08:38,330
OK, so that's that's the main reason for polar amplification. And there's a there's a clue in this.

85
00:08:38,330 --> 00:08:44,560
I'll show you a winter versus summer in a second. Why does land warm more than ocean?

86
00:08:44,560 --> 00:08:52,330
It's a little bit more complicated, but the basic answer is, is a similar answer is I put the amount of energy down over the ocean energy of the land.

87
00:08:52,330 --> 00:08:56,650
And I need to use more of it or use more of it to evaporate over the ocean, over the land.

88
00:08:56,650 --> 00:09:00,970
It's a little bit more complicated that Mike is here, right? Mike, Mike, Mike.

89
00:09:00,970 --> 00:09:07,630
Bring some nice work on this. A little more complicated now, but basically it's pretty well understood why the land warms more than the

90
00:09:07,630 --> 00:09:14,870
ocean and effectively the winds will kind of distribute the stuff around now.

91
00:09:14,870 --> 00:09:17,750
OK, so I won't say so far, yes. OK, good.

92
00:09:17,750 --> 00:09:23,630
So if I look at winter versus summer, there's a winter picture northern hemisphere, you can see now the scales are different here.

93
00:09:23,630 --> 00:09:28,430
So the amount of warming you have in the polar regions about twenty four degrees warmer.

94
00:09:28,430 --> 00:09:31,760
And if you look at the summer here, the amount of warming in as you go over here,

95
00:09:31,760 --> 00:09:35,090
the amount of warming you have here in the Arctic is only about three degrees.

96
00:09:35,090 --> 00:09:39,410
So if it's ice albedo feedback, you're kind of stuck with this strange situation.

97
00:09:39,410 --> 00:09:44,500
But ICE will be the feedback has to be working in the summertime, yet the polar ramifications in wintertime.

98
00:09:44,500 --> 00:09:50,770
So, you know, fundamentally, it's nice a bit of feedback. OK.

99
00:09:50,770 --> 00:09:58,630
And in both winter and in summer, the land warms more than the ocean, if you just go along a latitude line, you just say ocean land, ocean land.

100
00:09:58,630 --> 00:10:02,830
The land is always warmer than the ocean, so that's a basic pattern there.

101
00:10:02,830 --> 00:10:08,050
And it comes about because of this absorption of energy that's used to evaporate the tropics.

102
00:10:08,050 --> 00:10:13,810
That the atmosphere diffuses by winds just mixes that energy to the polar regions where it's colder and it condenses, right?

103
00:10:13,810 --> 00:10:22,600
So that's that's a good part of the answer. As long as you have those winds that are doing this mixing, you're going to have polar amplification.

104
00:10:22,600 --> 00:10:29,760
OK? OK, now let's look at departures from this multi-model mean this is average across all models.

105
00:10:29,760 --> 00:10:35,190
There's the picture of you with that before, and this is just a measure one sigma right?

106
00:10:35,190 --> 00:10:40,750
One standard deviation difference. So every model take a model ensemble right away.

107
00:10:40,750 --> 00:10:46,710
You have the differences and just calculate the standard deviation is a measure of the uncertainty.

108
00:10:46,710 --> 00:10:53,760
And you see the uncertainty is greatest in the Arctic, and it's also great over land and over ocean.

109
00:10:53,760 --> 00:10:56,790
And if I take the ratio of those two things called the coefficient variation,

110
00:10:56,790 --> 00:11:01,530
it gives me kind of the relative uncertainty compared to the ensemble mean change.

111
00:11:01,530 --> 00:11:06,750
And if you look at that ratio here, I don't know if you can see this, but the answer is about twenty five percent.

112
00:11:06,750 --> 00:11:11,130
So one sigma standard deviation twenty five percent and this is consistent with the global mean.

113
00:11:11,130 --> 00:11:16,050
It means you have a plus or minus 50 percent is kind of the extremes plus the models

114
00:11:16,050 --> 00:11:21,480
or the difference from the lowest model on the highest model is one hundred percent. Factor of two.

115
00:11:21,480 --> 00:11:30,390
OK, so regional uncertainty is typically plus or minus twenty five percent of the ensemble average average overall across all models.

116
00:11:30,390 --> 00:11:36,540
By the way, ask me a question any time or if I say something that's really offensive, I call me out on it.

117
00:11:36,540 --> 00:11:44,070
OK. All right, good. So if you do the same thing for wintertime, look at this coefficient of variation.

118
00:11:44,070 --> 00:11:48,630
The the spread. The models divided by the model mean you find that the same answer.

119
00:11:48,630 --> 00:11:53,310
It's about twenty five percent uncertainty plus or minus. And the same thing is true in the summer.

120
00:11:53,310 --> 00:12:01,620
The places with the biggest uncertainty are places here around Antarctica, where you have sea ice edges in in the sea ice.

121
00:12:01,620 --> 00:12:05,160
Projections say what's going to happen in the future is pretty different across models.

122
00:12:05,160 --> 00:12:13,320
Do you have a pretty different? The big uncertainty there and in the North Atlantic, here in the ocean, just south of Greenland,

123
00:12:13,320 --> 00:12:18,360
there's a lot of uncertainty across the models, but the actually the absolute change is very, very small.

124
00:12:18,360 --> 00:12:26,160
So models just differ a lot. But the the the difference is actually tiny because the absolute change is small.

125
00:12:26,160 --> 00:12:34,390
OK. Now, this is every model has these differences subtracted from the the ensemble mean?

126
00:12:34,390 --> 00:12:40,960
And what I did was just said, let's find the pattern that explains most of the variance in this.

127
00:12:40,960 --> 00:12:43,420
If you look at the differences, the residuals,

128
00:12:43,420 --> 00:12:50,200
let's find the patterns explains most of the differences between the models and if if you're if you're into it,

129
00:12:50,200 --> 00:12:54,340
this is just called a NeoGAF analysis of the residuals.

130
00:12:54,340 --> 00:12:59,740
So I just took the ensemble means every model you had the residual you have 18 of these fields,

131
00:12:59,740 --> 00:13:04,240
you calculate the empirical orthogonal functions and you pick out the leading pattern.

132
00:13:04,240 --> 00:13:08,200
And I did this for monthly mean seasonal means annual means.

133
00:13:08,200 --> 00:13:12,280
And the answer is pretty simple I what to you, sir?

134
00:13:12,280 --> 00:13:16,120
And if I lose you, the answer is the results.

135
00:13:16,120 --> 00:13:19,840
If you interpret the most of the regional uncertainty and temperature projections

136
00:13:19,840 --> 00:13:25,480
is related to one generic pattern of non-local uncertainty and feedbacks.

137
00:13:25,480 --> 00:13:33,620
That is, the most of the uncertainty in your place in space is not due to uncertainty in your place, in space, in the physics.

138
00:13:33,620 --> 00:13:36,680
So I'd say that's I immediately got implications for regional downscaling.

139
00:13:36,680 --> 00:13:43,910
There's only so much you can do in your place in space if you want to do a regional downscaling to improve temperature projections.

140
00:13:43,910 --> 00:13:48,710
In fact, pretty much nothing you can do on improvement. OK, let me show you that.

141
00:13:48,710 --> 00:13:53,600
So this is the pattern that explains most of the differences across the models.

142
00:13:53,600 --> 00:13:57,800
And if you look at this pattern, it looks like a pattern we've seen before.

143
00:13:57,800 --> 00:14:01,550
Colour scales a little difference. It's it's all the same sign.

144
00:14:01,550 --> 00:14:06,380
It's warm everywhere. It's more warm in the tropics or in the polar regions and in the tropics.

145
00:14:06,380 --> 00:14:09,980
It's more over the land that is over the ocean. It's the same exact pattern.

146
00:14:09,980 --> 00:14:14,770
It's the ensemble mean pattern and the physics is the same.

147
00:14:14,770 --> 00:14:19,240
And it explains 60 percent of the variance across all the models, so once you take this pattern away,

148
00:14:19,240 --> 00:14:25,690
you're just left with, you've got the uncertainty down by more than a factor of 50 percent.

149
00:14:25,690 --> 00:14:32,110
OK, now and let's see what I want to do this.

150
00:14:32,110 --> 00:14:34,630
OK, so this pattern has its place.

151
00:14:34,630 --> 00:14:45,300
60 percent of the variance is only homogeneous over the ocean land, with more warming over ocean, over land than ocean and its polar amplified.

152
00:14:45,300 --> 00:14:50,620
And so you can think of the physics as being the same physics. It's tropical clouds.

153
00:14:50,620 --> 00:14:56,670
They're explaining the uncertainty and the atmosphere as it's distributed in this heat differently.

154
00:14:56,670 --> 00:15:01,860
Now, if that's the case, let me try to explain this before I show you the answer.

155
00:15:01,860 --> 00:15:02,610
That's the case.

156
00:15:02,610 --> 00:15:11,210
It turns out also that the way the atmosphere equilibrate, this is mostly to bleed heat to space in the tropics, not in the high latitudes.

157
00:15:11,210 --> 00:15:16,280
So let's take the extreme case, here's my tropics. Here's the.

158
00:15:16,280 --> 00:15:21,230
Arctic, here's the Antarctic, and I'm just going to put a lot of CO2 in the atmosphere in the tropics are going

159
00:15:21,230 --> 00:15:25,370
to heat up and it's going to just pretend it is basically come to an equilibrium.

160
00:15:25,370 --> 00:15:29,840
If I just if I just stop the circulation so I can't move energy to the poles,

161
00:15:29,840 --> 00:15:33,710
I'm going to one the tropics by a certain amount amount so that the radiation to space

162
00:15:33,710 --> 00:15:39,640
is equal to what's what's coming back down the surface because of increased CO2.

163
00:15:39,640 --> 00:15:42,100
Now and the rest of the planet hasn't done anything.

164
00:15:42,100 --> 00:15:48,750
If I now take my atmosphere of speculation and move it to the polls now I can warm every place on the planet a lot, right?

165
00:15:48,750 --> 00:15:55,510
And polls more than the tropics. I still have to run the tropics up enough to bleed the same amount of energy to space to come into an equilibrium.

166
00:15:55,510 --> 00:16:04,260
So the global average temperature will be higher in the case where I'm be able to distribute that heat to the poles faster.

167
00:16:04,260 --> 00:16:10,510
OK. And that turns out to be the case. So, first of all, point out that the global average is one degrees Celsius.

168
00:16:10,510 --> 00:16:19,540
If you then say, if I project this pattern to each one of my models, residuals just and plotted versus a climate sensitivity, you see that here.

169
00:16:19,540 --> 00:16:25,840
This is the model of weights. So if a model, for example, has the global average plus twice this pattern, you're out here.

170
00:16:25,840 --> 00:16:31,780
It's got a climate sensitivity of five degrees that increases the temperature of five degrees to the CO2 doubling.

171
00:16:31,780 --> 00:16:41,030
And if the answer is my model gives me my ensemble average minus, say, one stick of this pattern, basically every climate signal sensitivity of two.

172
00:16:41,030 --> 00:16:49,840
So in other words, this pattern is explaining climate sensitivity. The uncertainty in why is there global climate uncertainty?

173
00:16:49,840 --> 00:16:59,090
Again, someone said it over there. The main reason for uncertainty in time models, global average clouds, tropical clouds,

174
00:16:59,090 --> 00:17:06,170
and what this is telling you is that also explain the global uncertainty everywhere, not just in the tropics, but everywhere.

175
00:17:06,170 --> 00:17:14,600
OK, so now that being the case, you have the correlation between climate sensitivity and how much your model projects

176
00:17:14,600 --> 00:17:19,650
under this pattern is 0.9 to four equilibrium climate sensitivity for transient,

177
00:17:19,650 --> 00:17:21,920
as 0.6 of that actually makes some sense.

178
00:17:21,920 --> 00:17:28,330
I won't go into it now, but you expect that result that would be more correlated with equilibrium than transient.

179
00:17:28,330 --> 00:17:32,770
And here's the other thing that based on that argument, I just told you,

180
00:17:32,770 --> 00:17:37,990
the more efficiently your model moves hit the polls, the stronger the warming you should have.

181
00:17:37,990 --> 00:17:42,460
So it turns out that not all climate models move heat to the polls at the same rate.

182
00:17:42,460 --> 00:17:46,480
And this plot here shows how much the model,

183
00:17:46,480 --> 00:17:50,620
how much your model projects onto that pattern and how much energy your model

184
00:17:50,620 --> 00:17:54,460
moves from the equator to the pole in the climatology that you start with.

185
00:17:54,460 --> 00:18:01,480
And there's a positive correlation is that is the more efficient you're the your model in today's climate moves heat from the tropics to the poles,

186
00:18:01,480 --> 00:18:10,170
the stronger your climate sensitivity is. And that's the physics behind this basically more diffusion to the poles, more warming.

187
00:18:10,170 --> 00:18:13,490
A view of clouds that are less bright, you get more warming,

188
00:18:13,490 --> 00:18:19,820
but this diffusion actually moves the stuff from the equator to the poles and gives you this pattern of Arctic amplification.

189
00:18:19,820 --> 00:18:22,520
This called pattern scaling, it's been known for a long time,

190
00:18:22,520 --> 00:18:29,240
but I guess a new part is coming up here is what's left over when you actually subtract out this climate.

191
00:18:29,240 --> 00:18:35,240
If we could solve the tropical cloud problem, what's left to explain your place in space?

192
00:18:35,240 --> 00:18:42,950
And I'll try to show you that here before I go there, though, if you do the same thing and you say, what's the common pattern in the annual mean?

193
00:18:42,950 --> 00:18:49,220
That's the top panel which we just looked at. If the common pattern in the wintertime looks like the wintertime ensemble average,

194
00:18:49,220 --> 00:18:55,070
the common pattern of differences in the summertime looks like the ensemble average in the summertime.

195
00:18:55,070 --> 00:18:59,310
So basically just telling you the same physics is going on and all of these models

196
00:18:59,310 --> 00:19:03,980
and the climate sensitivity is just set by uncertainty in tropical clouds,

197
00:19:03,980 --> 00:19:08,930
but explaining this pattern of variability across the whole planet.

198
00:19:08,930 --> 00:19:12,680
OK, so conclusion is most of the local uncertainty and temperature pressure is

199
00:19:12,680 --> 00:19:17,870
not due to differences in local feedbacks or to uncertainty inside to prices.

200
00:19:17,870 --> 00:19:22,670
It's due to tropical clouds. Now.

201
00:19:22,670 --> 00:19:31,520
Let's take that at the results and say, let's say we solve the problem and I'm not going to remove, so I have a residual that's from each model.

202
00:19:31,520 --> 00:19:36,680
And I mean, remove the global climate sensitivity model and say, what's left over?

203
00:19:36,680 --> 00:19:42,230
So what's left to explain after the leading difference pattern that is climate sensitivity is is solved.

204
00:19:42,230 --> 00:19:48,080
And that's this picture here. And so what this is is for each climate model.

205
00:19:48,080 --> 00:19:54,200
I've subtracted ensemble mean, I've subtracted out the pattern that goes with the climate sensitivity.

206
00:19:54,200 --> 00:19:54,890
And then I said,

207
00:19:54,890 --> 00:20:02,210
I have a residual I don't explain and I have that for each model and I take the standard deviation of those residuals and this is the plot.

208
00:20:02,210 --> 00:20:09,200
And you can see there's still most uncertainty left over to explain in the sea ice region of Antarctica,

209
00:20:09,200 --> 00:20:21,710
actually in the sea ice region now in the Beaufort Sea, which is a place where we've got really stunning changes in sea ice today and temperature.

210
00:20:21,710 --> 00:20:25,700
And if you take that and you and you divide it by the ensemble, mean change again,

211
00:20:25,700 --> 00:20:33,060
so you can you're getting a coefficient of variation after you remove climate sensitivity to this picture.

212
00:20:33,060 --> 00:20:38,360
And let me just put up the original coefficient values that one. Can you see the difference in colours?

213
00:20:38,360 --> 00:20:42,470
You've reduced the uncertainty from twenty five percent down to about 10 percent.

214
00:20:42,470 --> 00:20:47,780
So in any one place in space, once you remove global climate, once you fix the tropical cloud palms,

215
00:20:47,780 --> 00:20:54,350
now you're down to a local uncertainty of plus or minus 10 percent, which is pretty good.

216
00:20:54,350 --> 00:21:01,340
OK. So after accounting for climate says the regional strategy is cut in half to around 10 15 percent,

217
00:21:01,340 --> 00:21:06,190
with the exceptions being the Atlantic, North Atlantic and the Southern Ocean.

218
00:21:06,190 --> 00:21:12,850
Because of sea ice differences across the models. OK, now we can put this back in the original.

219
00:21:12,850 --> 00:21:19,300
Here's the original plot here, which is the ensemble average estimates from the climate models.

220
00:21:19,300 --> 00:21:24,580
And I'm going to put this residual pattern over here on the same scale as this,

221
00:21:24,580 --> 00:21:28,750
so you can get a sense of just how different they are, and that's the answer right there.

222
00:21:28,750 --> 00:21:36,170
So basically, there's really nothing left to explain locally. There's no point in regional downscaling for temperature.

223
00:21:36,170 --> 00:21:43,220
And in fact, I think it's probably going to make things much worse, if you look at the uncertainty even in regional models,

224
00:21:43,220 --> 00:21:46,760
when you drive them with the observed boundary conditions, you look at the differences across the models.

225
00:21:46,760 --> 00:21:51,470
The errors are actually much bigger than this in the climate, in the regional models.

226
00:21:51,470 --> 00:21:58,850
So in terms of regional downscaling of temperature, you can only do worse by doing that.

227
00:21:58,850 --> 00:22:03,650
I mean, the best thing to do is to leave the climate models alone and not touch it. OK.

228
00:22:03,650 --> 00:22:08,870
Why does this work? This is a little technical for a minute or two.

229
00:22:08,870 --> 00:22:11,960
It works because of the atmosphere is basically diffusive.

230
00:22:11,960 --> 00:22:17,480
The circulation has storms in the mid-latitudes and must move heat from the from the tropics to the poles.

231
00:22:17,480 --> 00:22:23,210
Basically, an illustration of that is, Oh, this got washed out and.

232
00:22:23,210 --> 00:22:34,010
And both of them. So this is the South Pole, the North Pole, the black line here is the ensemble average feedback from all of the IPCC climate models.

233
00:22:34,010 --> 00:22:43,760
How much maybe local response in temperature or radiation you get due to an increase in CO2 and all the squiggly lines you can barely see?

234
00:22:43,760 --> 00:22:49,990
Here are the feedbacks from each particular model. And so you can see every model does something pretty different.

235
00:22:49,990 --> 00:22:53,300
All right. There's a lot of uncertainty in the individual model,

236
00:22:53,300 --> 00:23:02,780
but that the ensemble average and you can define a feedback is the amount of force in at that latitude divided by the temperature response.

237
00:23:02,780 --> 00:23:11,860
OK, now. If you just take a simple diffusion model and I say about two parts per million

238
00:23:11,860 --> 00:23:15,190
squared or whatever it is to a WCO to two that I put it in my atmosphere everywhere,

239
00:23:15,190 --> 00:23:19,930
and I have this distribution of feedbacks I can solve for the temperature change.

240
00:23:19,930 --> 00:23:31,240
And there's nothing fancy and this is just pure diffusion. And then I can say, OK, that's my that's my ensemble mean response to the increase in two,

241
00:23:31,240 --> 00:23:34,840
and then I can put her uncertainty in the local feedback and say, OK, in the tropics.

242
00:23:34,840 --> 00:23:39,370
I don't know exactly what the answer is. I so I put it a little uncertainty in and I can say, Well, what?

243
00:23:39,370 --> 00:23:43,450
What does that do to the global average temperature or due to the temperature distribution?

244
00:23:43,450 --> 00:23:46,240
So there's this didn't come out of it. That's the feedback.

245
00:23:46,240 --> 00:23:53,650
So we put in little delta functions are close to delta functions at each latitude of uncertainty and a feedback, and they're all positive feedback.

246
00:23:53,650 --> 00:23:59,920
So every one of these simulations is going to be warmer than the ensemble mean simulation.

247
00:23:59,920 --> 00:24:07,480
And so here's an example of that, so that the black line here are the Black Red Line, is the ensemble mean from the climate models?

248
00:24:07,480 --> 00:24:14,380
And that little red dash, the coming up is just a perturbation we're going to add on at that latitude band and that's the only place to get it.

249
00:24:14,380 --> 00:24:21,640
So we just make it more positive there. So for the same amount of force in locally, expect warm more in that region in that particular model.

250
00:24:21,640 --> 00:24:26,470
But the atmosphere situation is going to move that heat away and distribute it in other places.

251
00:24:26,470 --> 00:24:33,130
They can take that equation there, and you can add a perturbation and uncertainty in the feedback.

252
00:24:33,130 --> 00:24:34,630
A delta lambda.

253
00:24:34,630 --> 00:24:44,050
And you can subtract out the ensemble average change and come up with an equation for the responsive temperature that perturbation temperature.

254
00:24:44,050 --> 00:24:50,170
That is the uncertainty in one model due to that perturbation at that latitude.

255
00:24:50,170 --> 00:24:56,800
And if life is really good for you and you have constant diffusivity, which you don't and constant feedbacks,

256
00:24:56,800 --> 00:25:02,890
you just end up with base exponential decay around where you put in that uncertainty so locally have a warm,

257
00:25:02,890 --> 00:25:07,960
but the circulation will take this stuff and distribute it away. And this is extremely powerful.

258
00:25:07,960 --> 00:25:12,040
So let me show you what the answer looks like. That black line is the equator.

259
00:25:12,040 --> 00:25:16,480
The equator to pole to pole south is poles on the left.

260
00:25:16,480 --> 00:25:24,040
Sorry, this didn't come out very well. North poles on the right equator is in the middle of the black line is the ensemble average change you

261
00:25:24,040 --> 00:25:30,400
get from using the ensemble average feedbacks from the climate models from the scene at five miles.

262
00:25:30,400 --> 00:25:36,340
And each one of those other lines is if I put a perturbation in place, for example, here, let me just show this.

263
00:25:36,340 --> 00:25:43,580
Here's a perturbation that we put in in 20 degrees here. When we put it on the equator is 120 degrees north is 30 degrees north.

264
00:25:43,580 --> 00:25:47,080
Here's 50 degrees north. So we put perturbations in these.

265
00:25:47,080 --> 00:25:53,740
One of these lines is the solution for what the temperature is with a model that has a perturbation at that latitude.

266
00:25:53,740 --> 00:25:58,870
And you can see the all the lines are all above it. It's a positive feedback perturbation, so they all have to be warmer.

267
00:25:58,870 --> 00:26:03,540
But the distribution the shape is exactly the same as the ensemble mean shape.

268
00:26:03,540 --> 00:26:06,000
Does this is the the equivalent of pattern scaling?

269
00:26:06,000 --> 00:26:15,270
It's just there because of diffusivity, the atmosphere is just distributing this heat by storms and moving it from the tropics to the poles.

270
00:26:15,270 --> 00:26:20,010
OK, so these are really hard to resist. Let me just show you this part, an honest man.

271
00:26:20,010 --> 00:26:23,160
So if you just subtract, if you just like normalise each one of these curves,

272
00:26:23,160 --> 00:26:30,480
you take out the ensemble mean so that you adjust the mean temperature of the global mean temperature.

273
00:26:30,480 --> 00:26:35,070
And these kids all have the same thing. They all kind of collapse under the same curve.

274
00:26:35,070 --> 00:26:39,480
So it says no matter where my uncertainty is, if it's if it's at 20 degrees south in one model,

275
00:26:39,480 --> 00:26:46,680
20 degrees north, one model and it's like a more positive in Model A and less positive in Model B.

276
00:26:46,680 --> 00:26:59,120
The result is going to give me the same shape in the end of the temperature. Profile change, it's going to be polar amplified, right and right.

277
00:26:59,120 --> 00:27:04,570
Yeah, and that's all because of diffusion. It's because the fusion is roughly the same in every model for circulation.

278
00:27:04,570 --> 00:27:12,430
Every model has one strong track in the mid-latitudes moving heat and water vapour from the tropics to the poles.

279
00:27:12,430 --> 00:27:18,820
OK. Anything else I want to say about that? So I'm going to skip the next one because it's going to be impossible to read.

280
00:27:18,820 --> 00:27:23,600
And this is true for any uncertainty between 50 degrees south and north.

281
00:27:23,600 --> 00:27:28,600
So it's really powerful weapons, daily works, because the atmosphere is basically Diffusive,

282
00:27:28,600 --> 00:27:35,800
and you have this need to move vapour from the tropics to the polar regions, which give you that polar amplification.

283
00:27:35,800 --> 00:27:39,280
And that's that's a two dimensional, more static energy and small is nothing fancy.

284
00:27:39,280 --> 00:27:44,020
There's no land in it, and it tells you exactly what the climate models are doing.

285
00:27:44,020 --> 00:27:53,740
OK? So certainly in regional feedbacks, give rise to uncertainty in global global average global temperature distribution implications.

286
00:27:53,740 --> 00:27:57,310
If you're interested in, say, temperature change in Oxford in the future,

287
00:27:57,310 --> 00:28:06,850
then you can't gain much by resolving take a regional model and say, downscaling all the GC-MS for for Oxford.

288
00:28:06,850 --> 00:28:13,840
Because basically the uncertainty Oxford is determined by by tropical clouds, not by the stuff going on around here.

289
00:28:13,840 --> 00:28:21,770
That's true of every place in the space, pretty much, except for the sea ice ages where sea ice actually plays a pretty big role.

290
00:28:21,770 --> 00:28:31,300
OK. So whatever whatever gives you large and certainly a climate sensitivity will give you large distal regional uncertainty.

291
00:28:31,300 --> 00:28:33,370
It's the largest uncertainties and tropical feedbacks.

292
00:28:33,370 --> 00:28:38,950
Hence the largest uncertainty in list of reasonable locations is due to uncertainty in the tropics.

293
00:28:38,950 --> 00:28:43,090
OK. This doesn't work for precipitation, not a surprise,

294
00:28:43,090 --> 00:28:49,060
because precipitation is controlled by circulation changes, so let me just show you how bad this is for.

295
00:28:49,060 --> 00:28:54,050
This is again, this is not great colour scheme, but anywhere it's blue.

296
00:28:54,050 --> 00:28:59,890
This is the ratio of the precipitation at the end of the century compared to the precipitation at the end of last century.

297
00:28:59,890 --> 00:29:02,110
In the end, it's blue in the high latitudes.

298
00:29:02,110 --> 00:29:09,160
These are regions where it's going to be wetter in terms of more rainfall at the end of the century and then the regions where it's wet,

299
00:29:09,160 --> 00:29:16,300
which you can barely see. I'm sorry, drier, which is it's hard to see, but there is subtropical regions here, so well,

300
00:29:16,300 --> 00:29:21,010
for example, the regions where it will be drier in the future compared to just today.

301
00:29:21,010 --> 00:29:25,000
And if you do this same thing and you look at the standard deviation of differences across the models,

302
00:29:25,000 --> 00:29:35,440
you find out the numbers about the same models differ by about 20 to 30 30 percent in terms of what they say about precipitation.

303
00:29:35,440 --> 00:29:40,870
That's the leading pattern of differences across the models is pretty messy.

304
00:29:40,870 --> 00:29:45,850
It's tropics centric, but it only spend 17 percent of the variance.

305
00:29:45,850 --> 00:29:52,420
So there's not one pattern that says, OK, climate sensitivity is can is is controlling the amount of precipitation here or there.

306
00:29:52,420 --> 00:30:01,300
Every model does it differently. So that's one problem. It's not explained by simple diffusion and it's not related to climate sensitivity.

307
00:30:01,300 --> 00:30:07,660
So that's the raw uncertainty we looked at before the standard deviation of this precipitation ratio.

308
00:30:07,660 --> 00:30:13,990
And that's what you see after you remove the common pattern of differences in the answer is you've made no improvement whatsoever.

309
00:30:13,990 --> 00:30:20,860
So precipitation is one of these things that if there's going to be improvement and it's going to be, I don't know what it's going to take.

310
00:30:20,860 --> 00:30:24,990
It's not going to be good, though. I mean, it's a hard problem.

311
00:30:24,990 --> 00:30:31,780
And and if you did have to downscale, then there's an issue of do I trust a regional model or just I empirically down scale?

312
00:30:31,780 --> 00:30:35,410
It's it's it's a it's a it's a tricky trade-off.

313
00:30:35,410 --> 00:30:43,840
OK, so so where we at, we're at that most of the uncertainty in the climate system is due to tropical clouds.

314
00:30:43,840 --> 00:30:51,050
Most of the uncertainty at your place in space is due to tropical clouds wherever you live.

315
00:30:51,050 --> 00:30:56,170
And the physics of this is is pretty simple, just the fusion.

316
00:30:56,170 --> 00:31:05,940
OK, now this actually makes life easier if you're actually asking what's the impact of the sort of temperature changes in the future?

317
00:31:05,940 --> 00:31:13,050
You don't have to downscale or you don't even have to look at all the different climate models, you just have to look at two, right?

318
00:31:13,050 --> 00:31:17,050
You just have to look at a model with high sensitivity and model with low sensitivity,

319
00:31:17,050 --> 00:31:21,430
and that should bracket pretty much the whole uncertainty in space.

320
00:31:21,430 --> 00:31:24,960
Now let me show you an example that works pretty well, OK?

321
00:31:24,960 --> 00:31:30,360
And this is examples in heat, and there's a lot of ways you can measure heat stress.

322
00:31:30,360 --> 00:31:35,640
No measures if by some combination of temperature and relative humidity.

323
00:31:35,640 --> 00:31:40,590
If it means anything to you and it's always out to be wet bulb temperature for Noah,

324
00:31:40,590 --> 00:31:49,810
but there's other indices called humidity humid, humid x, and there's various things that I'll tell you roughly the same thing.

325
00:31:49,810 --> 00:31:55,130
The in the Noah language, basically, when you see it, a wet bulb temperature of twenty seven degrees,

326
00:31:55,130 --> 00:31:58,790
you're in this danger zone where you have heat cramps and heat exhaustion, the like.

327
00:31:58,790 --> 00:32:04,370
The heat stroke is probable and with continued activity.

328
00:32:04,370 --> 00:32:08,330
And the list of symptoms are really horrendous experiences.

329
00:32:08,330 --> 00:32:11,030
I've experienced this. It's not fun.

330
00:32:11,030 --> 00:32:19,370
You get fatigue, nausea, headache, excessive thirst, muscle muscle aches, confusion weakness is slowed heartbeat, dizziness, fainting.

331
00:32:19,370 --> 00:32:23,720
And it's pretty hard to work under these conditions. OK.

332
00:32:23,720 --> 00:32:32,510
And then you can go to extreme danger where heatstroke is imminent. This is you take all the symptoms that I just read and you add to that vertigo,

333
00:32:32,510 --> 00:32:37,070
shortness of breath, vomiting, blood in the urine, delirium, loss of consciousness.

334
00:32:37,070 --> 00:32:42,710
I mean, I don't know anyone who experiences convulsions. It's bad.

335
00:32:42,710 --> 00:32:52,070
OK, so let's look at the last thirty five years from era and say under these two definitions of danger and extreme danger,

336
00:32:52,070 --> 00:32:57,440
how often we see these conditions. So this is actually taking the the era data,

337
00:32:57,440 --> 00:33:13,130
the last thirty five years of six hourly data and saying How often do we exceed the danger category today and guys who didn't come out very well?

338
00:33:13,130 --> 00:33:19,110
But there's you can see some three days per year. So on average is off places like Oman and Bangladesh.

339
00:33:19,110 --> 00:33:21,200
We we're like three weeks per year.

340
00:33:21,200 --> 00:33:30,290
You exceed the danger zone where you we really have to be really careful about being outside and being active in any way.

341
00:33:30,290 --> 00:33:35,450
And I want to see that right. There's a few other places like the town here where you were green.

342
00:33:35,450 --> 00:33:39,530
So like a couple of days per year, there's there's Eastern China,

343
00:33:39,530 --> 00:33:47,390
where you're up to maybe two to five days per year and then you can say, well, how often the extreme extreme danger in this?

344
00:33:47,390 --> 00:33:50,570
Actually, there's only two points on this one here.

345
00:33:50,570 --> 00:33:56,700
And whatever this place is, you know, it's it's this little thing that sticks out of Central America.

346
00:33:56,700 --> 00:34:01,260
You know, I mean. I've never been there, so I don't know.

347
00:34:01,260 --> 00:34:05,070
But yeah, and then there's one place here, I guess you like near Oman,

348
00:34:05,070 --> 00:34:12,780
where basically one day in the last thirty five years hit the extreme danger category where you basically if you are out, you deck your dead.

349
00:34:12,780 --> 00:34:21,900
OK, so that's today. And then what I did is I said, OK, let's look at the end of this century using the business as usual mission style ensemble mean.

350
00:34:21,900 --> 00:34:26,700
So we're just averaging all the time miles. We're getting a new annual cycle in temperature.

351
00:34:26,700 --> 00:34:32,490
We're assuming the same weather. So we're just adding that change in the annual cycle temperature to the same weather.

352
00:34:32,490 --> 00:34:34,710
From that, we've experienced the last thirty five years,

353
00:34:34,710 --> 00:34:44,760
which is probably a pretty conservative estimate of of how often you extreme exceed this extreme danger because there's reasonable evidence,

354
00:34:44,760 --> 00:34:49,500
which is just as you warm up, the very variance of temperature should increase,

355
00:34:49,500 --> 00:34:54,290
which means you're going to exceed extremes that warm extremes more often than not.

356
00:34:54,290 --> 00:35:01,880
OK. So same weather, plus this change in the annual cycle of temperature and is going to look again how often

357
00:35:01,880 --> 00:35:06,380
you see these things and now remember this is the same picture I showed you before.

358
00:35:06,380 --> 00:35:11,840
So the global average temperature change is three point seven degrees, but you have an annual cycle associated with that.

359
00:35:11,840 --> 00:35:16,960
And this is what you get with end of the century and the numbers here have changed.

360
00:35:16,960 --> 00:35:21,590
Then the scale goes up to 320 days per year. So basically, it's our year.

361
00:35:21,590 --> 00:35:29,840
So if you look at the Amazon, it's all a year when you're in the danger zone. If you look in the southeast US here, you're in yellow and orange.

362
00:35:29,840 --> 00:35:39,320
So you may be like a month a little bit more than a month a year that you basically have to be really careful about being outside northern Australia.

363
00:35:39,320 --> 00:35:44,090
You're looking at almost there's those of the coastal rim here is about a half a year.

364
00:35:44,090 --> 00:35:53,480
You know, you have to be careful. India, obviously, you're looking at one or two months per year, sometimes one or two months per year.

365
00:35:53,480 --> 00:35:59,300
And then how often the extreme, the extreme danger. We're basically heat kills you very quickly.

366
00:35:59,300 --> 00:36:09,440
It's maybe a week per year in the southeast US, a few like a month per year in the Amazon, maybe a month per year in northern India.

367
00:36:09,440 --> 00:36:18,420
It's not good. OK, now now I'm going to take this pattern of change and just add to it the thought that

368
00:36:18,420 --> 00:36:23,480
the common pattern of difference and just multiply it by the most extreme climates.

369
00:36:23,480 --> 00:36:27,320
So I'm just saying, let's just assume this pattern exists.

370
00:36:27,320 --> 00:36:32,780
I have the ensemble mean, I just add that change to it, and I just do that calculation again.

371
00:36:32,780 --> 00:36:41,900
Yeah. OK, good. OK, so that's what you get here, and you can see the numbers go up pretty significantly here in the southeast us,

372
00:36:41,900 --> 00:36:45,980
you're now into orange, so you're looking at maybe three months per year, basically all summer.

373
00:36:45,980 --> 00:36:51,170
You have to be really careful about the amount of time you spend outdoors.

374
00:36:51,170 --> 00:36:58,610
There's a lot us to [INAUDIBLE], basically here in Europe over six months per year, there's there's Southeast Asia,

375
00:36:58,610 --> 00:37:06,500
India, where you're like five months per year, where you're exceeding the danger category and then eastern China.

376
00:37:06,500 --> 00:37:11,930
Here we're in the yellow, so maybe six weeks per year and then extreme danger here this now it's getting

377
00:37:11,930 --> 00:37:15,770
pretty significant where you're maybe three weeks per year in the southeast US,

378
00:37:15,770 --> 00:37:20,510
you cannot go outside for more than an hour, basically, and you want to be active.

379
00:37:20,510 --> 00:37:29,870
So these are pretty extreme conditions. Now to illustrate how this how this is useful, this pattern is useful.

380
00:37:29,870 --> 00:37:36,740
What I did is I went through every single climate model and I did this calculation how many days per year you exceed this danger zone?

381
00:37:36,740 --> 00:37:40,190
And I said for any one, places places big Oxford where you never exceed it.

382
00:37:40,190 --> 00:37:45,810
Nice, you right? I mean, you're on an island, for God's sakes. OK, but let's talk.

383
00:37:45,810 --> 00:37:52,310
I think someplace where no one want to live, like Atlanta may be like, OK, like down here.

384
00:37:52,310 --> 00:37:57,260
OK. So you picked that place. You say, like, OK, I take my ensemble.

385
00:37:57,260 --> 00:38:03,980
Mean, I've added just like that extreme climate sensitivity, and I've got a number this whatever it is,

386
00:38:03,980 --> 00:38:07,910
it looks like it's maybe orange, maybe one hundred and seven days a year.

387
00:38:07,910 --> 00:38:12,740
So three and a half months per year, you can't be you have to be really careful about being outside.

388
00:38:12,740 --> 00:38:16,370
And what I did is I went to every single climate. I'll forget about these pattern things.

389
00:38:16,370 --> 00:38:19,820
I just said, here's the climate change this model predicted at the end of the year.

390
00:38:19,820 --> 00:38:23,240
I'm going to do this calculation and figure out how many days per year exceeds it.

391
00:38:23,240 --> 00:38:28,370
And I'm going to take the model with the most at that point in space. And compared to that number,

392
00:38:28,370 --> 00:38:36,380
some of the words if I just take this shortcut is one one pattern downscaling one model fogerty's by the heat application.

393
00:38:36,380 --> 00:38:44,960
How far off I am by. Instead of using all 18 models and probing the model and say, just like what's what's the worst case scenario?

394
00:38:44,960 --> 00:38:51,530
And that's this, this is the additional days you get if you actually probe every climate models say,

395
00:38:51,530 --> 00:38:57,350
is there ever a climate model that gives me more than the ones on the right hand side that one estimate the right hand side?

396
00:38:57,350 --> 00:39:04,070
And the thing to point out here is the scales change. So basically in the southeast U.S., you get an extra month.

397
00:39:04,070 --> 00:39:07,490
OK, OK. So there was a climate model that gave you an X.

398
00:39:07,490 --> 00:39:14,900
Yeah. I'm just wondering just looking at this. So I'm just looking at a lot of like desert areas.

399
00:39:14,900 --> 00:39:20,420
Yeah, Western states, Western Sahara.

400
00:39:20,420 --> 00:39:25,550
Yeah, this is this because it's exactly they're dry.

401
00:39:25,550 --> 00:39:29,340
They're really dry. So it's very hot, but it's extremely dry. Good point.

402
00:39:29,340 --> 00:39:36,350
Yeah. So like western us, a lot of Central Asia is is white, and that's because they're really dry areas.

403
00:39:36,350 --> 00:39:41,810
So the heat stress is not as great, even though in temperatures higher. Yep.

404
00:39:41,810 --> 00:39:45,760
OK, good one. OK. But basically, you know what?

405
00:39:45,760 --> 00:39:52,540
What more do you get like if you're in Europe? You can see maybe there's like a light green there, which we see an additional week.

406
00:39:52,540 --> 00:40:02,800
You know, there's some model in that in that close to 80 models that gave you one more week of extreme danger compared to to see a guy can go.

407
00:40:02,800 --> 00:40:06,860
I can like, Oh, look at this. Yeah.

408
00:40:06,860 --> 00:40:14,120
So a [INAUDIBLE], Africa, for example, Orange. One hundred and fifty days and on the right hand side, I had maybe another 40 days.

409
00:40:14,120 --> 00:40:22,070
The point here is that, OK, yeah, it's worse. But I mean, it's already when you're talking about three months of camp uninhabitable.

410
00:40:22,070 --> 00:40:32,610
What's another month, right? It's it's like you've got a vision here, which is, you know, using all 18 months doesn't give you much more information.

411
00:40:32,610 --> 00:40:36,140
OK, and then here the extreme maximum temperature that kills people.

412
00:40:36,140 --> 00:40:39,590
I don't know if you can see it, but there's basically there's less than five extra days,

413
00:40:39,590 --> 00:40:43,790
so there might be one Mali can find that gives you like five more days of extreme dust.

414
00:40:43,790 --> 00:40:48,350
That's five days on top of a month. Right?

415
00:40:48,350 --> 00:40:51,290
The bad number is it's a month.

416
00:40:51,290 --> 00:41:01,130
OK, now, since the 1.5 degree report just came out, let me show you the answer for a world is one point eight degrees warmer and that's here.

417
00:41:01,130 --> 00:41:06,590
All right. So I think that maybe this was in the report, but I haven't read the report yet.

418
00:41:06,590 --> 00:41:09,870
Sorry, Miles. OK, I read that. I read some of it.

419
00:41:09,870 --> 00:41:16,350
I am totally OK. But basically, you know, here's the difference between eight point five high climate sensitivity.

420
00:41:16,350 --> 00:41:19,550
So this is pretty extreme. And here's that one point eight degrees.

421
00:41:19,550 --> 00:41:28,470
This is the RCP four point five scenario, and you still end up with a lot of yellow here, which is on the order of six weeks per year of danger.

422
00:41:28,470 --> 00:41:36,950
Right. So one point eight degree warmer world is still a real problem when it comes to heat stress for people.

423
00:41:36,950 --> 00:41:44,030
OK. All right, so the summary here, there's a common pattern in the uncertainty and temperature change due to anthropogenic

424
00:41:44,030 --> 00:41:49,970
forcing more warming over land than over ocean global and extend polar amplified.

425
00:41:49,970 --> 00:41:58,930
And that pattern is really basic physics. It's strongly correlated to climate sensitivity, which means it's related to tropical clouds.

426
00:41:58,930 --> 00:42:09,940
Oh, let's see. Go here for change, it indicates that 60 percent of the global regional differences are uncertainty is due to one generic

427
00:42:09,940 --> 00:42:17,500
uncertainty due to tropical clouds in pop tropical feedbacks sea because the atmosphere is really high.

428
00:42:17,500 --> 00:42:24,670
Diffusive After accounting for this comparative uncertainty, the regional scale uncertainties is greatly reduced.

429
00:42:24,670 --> 00:42:28,120
Local prices account for less than I've got to say in an absolute sense.

430
00:42:28,120 --> 00:42:32,440
The uncertainty in your place in space is less than a degree at the end of the century,

431
00:42:32,440 --> 00:42:42,190
a lesser degree compared to the ensemble mean of typically five or six degrees over land by 12 degrees over the pole.

432
00:42:42,190 --> 00:42:44,260
The indications are that in most cases, numerical,

433
00:42:44,260 --> 00:42:49,480
downscaled and sea five models can't reduce regional uncertainty in mostly most likely is make it worse.

434
00:42:49,480 --> 00:42:52,490
I have a couple of slides that just have this list of.

435
00:42:52,490 --> 00:42:59,960
Problems that come with regional can think there's a point here in the rare instance in numerical downscaling is required and efficient

436
00:42:59,960 --> 00:43:09,530
approaches to just about any regional sun is just they use the common pattern and just scale up the ensemble mean change with that pattern.

437
00:43:09,530 --> 00:43:12,830
And so I guess going back to that first statement is like, why would you care about the global?

438
00:43:12,830 --> 00:43:19,050
Meanwhile, turns out that all of the uncertainty in your place in space is related to that global mean answer.

439
00:43:19,050 --> 00:43:22,970
Because because of the atmospheric dynamics. All right.

440
00:43:22,970 --> 00:43:41,540
And I think I'm going to stop there. Thank you, David, for a great motivators, also think about tropical clouds chess club, yeah.

441
00:43:41,540 --> 00:43:46,620
So David is happy to take questions, but if I can just ask that you do wait for a microphone because this is being filmed,

442
00:43:46,620 --> 00:43:51,980
so just to make sure that the audio is captured by understand, right?

443
00:43:51,980 --> 00:43:56,960
If you do pattern scaling based on one target time, but a lot of takeoff,

444
00:43:56,960 --> 00:44:04,820
Kyle Chalmers were indicates that as the ocean comes into equilibrium, the patterns corresponding to a given warming actually change.

445
00:44:04,820 --> 00:44:08,570
And I wonder to what extent you've explored what happens if you go, you know, say,

446
00:44:08,570 --> 00:44:15,440
five hundred years out and so forth or in terms of the just a change in the pattern scaling.

447
00:44:15,440 --> 00:44:19,370
I don't know if you remember, but there's the scatter plot of transient and equilibrium time sensitivity.

448
00:44:19,370 --> 00:44:24,650
And the closer you get to equilibrium, the better this works. And that's simply because the yeah, the I mean,

449
00:44:24,650 --> 00:44:30,230
the uncertainty on short timescales is the uncertainty and ocean heat uptake or the uncertainty where sea ice is going to go.

450
00:44:30,230 --> 00:44:33,950
But it's just going to completely go and there's no uncertainty anymore.

451
00:44:33,950 --> 00:44:39,530
And basically, when the ocean up to heat up, which is different, every model actually does come to adjustment.

452
00:44:39,530 --> 00:44:41,990
It doesn't matter anymore. So in equilibrium,

453
00:44:41,990 --> 00:44:50,630
you get your the pattern works best right and it doesn't work great in that in the transient correlation was only zero point six two.

454
00:44:50,630 --> 00:45:00,440
So that's that's an issue where but I'd say that so so places where you still I'll have to explain the North Atlantic,

455
00:45:00,440 --> 00:45:05,840
where the changes are small anyways, where you have sea ice that's retreating is very different needs model that

456
00:45:05,840 --> 00:45:09,650
doesn't come back too much to haunt the tropics or the rest of the poles to be.

457
00:45:09,650 --> 00:45:17,180
Because if you think about those two uncertainties are really high latitudes and all the diffusion is kind of more equator.

458
00:45:17,180 --> 00:45:17,720
Where to this?

459
00:45:17,720 --> 00:45:25,250
So in fact, you can you have to really get big uncertainties here before you can believe that that temperature and back into the tropics,

460
00:45:25,250 --> 00:45:27,200
we try this a little bit. It's really difficult.

461
00:45:27,200 --> 00:45:36,820
I mean, you get huge uncertainty in polar temperatures, but it just doesn't communicate to the rest of the world.

462
00:45:36,820 --> 00:45:40,270
Thanks, David. I guess proponents of downscaling,

463
00:45:40,270 --> 00:45:46,570
if they were in the room might suggest that one of the main reasons you want to do it is actually for the hydrology down scale,

464
00:45:46,570 --> 00:45:50,860
you know, better topography and, you know, better weather rainfalls and therefore,

465
00:45:50,860 --> 00:45:55,270
you know, better which catchments got the water in which soil is moist.

466
00:45:55,270 --> 00:46:01,720
So it is kind of a two part question for us is would you disagree that it's useful to do dams getting for that for hydrology?

467
00:46:01,720 --> 00:46:10,180
And the second part is, is there any likelihood that there's actually a possible feedback if you get the hydrological balance better at small scale?

468
00:46:10,180 --> 00:46:18,370
Will that feedback into the large scale climate of the temperature that you've questioned? Flexible, OK, which of the two questions first?

469
00:46:18,370 --> 00:46:23,170
Yes, if you really want hydrology, you have the down scale because it's about precipitation.

470
00:46:23,170 --> 00:46:26,230
And none of this is getting worse than the issue, of course,

471
00:46:26,230 --> 00:46:36,490
is the rainfall is done pretty poorly in climate models and any place where you have really stable precipitation,

472
00:46:36,490 --> 00:46:42,700
mid-latitudes high latitudes, then regional downscaling to get sharper, greatest say due to mountains.

473
00:46:42,700 --> 00:46:48,220
But that is a sensible thing to do. Next question would be do you want to do it with numerical modelling and empirical modelling?

474
00:46:48,220 --> 00:46:55,690
My choice will always be empirical. Just because the regional models can give you big uncertainties, in fact, I could show you slide.

475
00:46:55,690 --> 00:46:59,110
But the regional models for Europe, for example,

476
00:46:59,110 --> 00:47:05,470
they've been mining them for a long time and and actually the uncertainty if you drive them with the modern day boundary conditions.

477
00:47:05,470 --> 00:47:07,450
So we actually know what's coming in.

478
00:47:07,450 --> 00:47:13,570
There's no problem with the GCM having the storm track the wrong places are driving it with the primary analysis and say,

479
00:47:13,570 --> 00:47:18,970
how well do the models reproduce the seasonal average precipitation? Well, the answer is plus or minus 40 percent?

480
00:47:18,970 --> 00:47:22,600
That's actually worse than the GCM, much worse than the GCM.

481
00:47:22,600 --> 00:47:31,120
So, so I'd say like, yeah, if you want to do better for changes in precipitation in places where precipitation is strata form,

482
00:47:31,120 --> 00:47:36,820
precipitation is not dependent on your prioritisation of convection empirically down scale, if you can.

483
00:47:36,820 --> 00:47:42,430
If you got the data, if you don't have the data becomes something else, that would be my response.

484
00:47:42,430 --> 00:47:47,530
Now is there? Is there a feedback? There probably is, but it's probably not fundamental.

485
00:47:47,530 --> 00:47:53,170
I mean, I think it's probably a a second order process. That would be my guess.

486
00:47:53,170 --> 00:47:59,140
Yeah. But if you look at it, I mean, there's these big office of regional models and how well they produce the modern climate.

487
00:47:59,140 --> 00:48:06,220
And they're generally the uncertainties the because of the well, these are not built for climate purposes, right?

488
00:48:06,220 --> 00:48:10,630
If you look at the uncertainties in precipitation like says plus or minus 40 percent in the air,

489
00:48:10,630 --> 00:48:18,510
not even in the that's the ensemble mean, that's the climatology. And then the interannual variability in these models is horrendous.

490
00:48:18,510 --> 00:48:25,740
Temperature uncertainties are the same order as the GCM, so I don't I don't think in general, unless you happen to be in a place like Norway,

491
00:48:25,740 --> 00:48:34,140
maybe or you even do that because every model puts a storm track in a slightly different place and they change it in a different way.

492
00:48:34,140 --> 00:48:43,670
So I mean, I know people want to know regional precipitation changes, but I just don't think it's something we should be promising to deliver.

493
00:48:43,670 --> 00:48:47,810
It's asking me the question. Hello. I live it, thank you, thank you very much indeed.

494
00:48:47,810 --> 00:48:53,220
This is fascinating stuff. I'm trying to reconcile it in my head is Thomas Hoenig goal?

495
00:48:53,220 --> 00:48:58,250
Exactly. It's on behalf of. Yeah, he was pretty one of the people who failed to bribe the doorman.

496
00:48:58,250 --> 00:49:03,920
But some of which I believe there are quite a few. Thomas, well, I'm going to ask you, what would you want?

497
00:49:03,920 --> 00:49:10,190
Ask questions, shall I? I mean, if you say so, Thomas has done a bunch of experiments.

498
00:49:10,190 --> 00:49:14,730
Yeah, it's always done stripey sunshades. Why doesn't? Why doesn't this apply to stripe his conscience?

499
00:49:14,730 --> 00:49:20,180
Do you want to question in what respect does it not apply to them? Wait, wait, apply to what?

500
00:49:20,180 --> 00:49:23,480
OK, so so Thomas is imposed if I get this right,

501
00:49:23,480 --> 00:49:34,540
Thomas stripey geoengineering experiments where he puts his or sunshade across the planet and finds quite a lot of structure in the response,

502
00:49:34,540 --> 00:49:41,240
at least. Yes, yes. But over time, the temperature response that you get is presumably behaving in a similarly diffusive way.

503
00:49:41,240 --> 00:49:50,030
I imagine in the residual as well. Yeah, I mean, it's not just one percent, yes, but it's over the course of the first decade.

504
00:49:50,030 --> 00:49:55,100
It starts to diffuse out a little bit over time. OK. Do you do you have a question?

505
00:49:55,100 --> 00:49:58,550
I apologise. It was. It was going to be important, but apparently it isn't.

506
00:49:58,550 --> 00:50:03,050
OK. But OK, well, I'd say, would this apply?

507
00:50:03,050 --> 00:50:08,740
Would this apply to a geoengineering a regional geoengineering plan?

508
00:50:08,740 --> 00:50:16,030
It should and actually, Malta Stuka has done these prescribed CO2 bans, and it actually works pretty well in those things.

509
00:50:16,030 --> 00:50:22,540
And the idea, of course, if you apply a CO2 warming to the polar regions for the reason we were just talking about the second ago,

510
00:50:22,540 --> 00:50:29,110
it's hard to get that warming out of there. So you really have to warm it up a lot to diffuse that to get the energy diffusion back.

511
00:50:29,110 --> 00:50:37,870
But in other bands like one, when multiplied in bands of CO2, basically it's distributed really nicely by distribution of more static energy.

512
00:50:37,870 --> 00:50:46,870
If you wait long enough, you have to wait long enough, right? Do you have any idea how long you'd have to wait 100 years, 100 years?

513
00:50:46,870 --> 00:50:54,710
OK? Maybe less, but eventually you have a slab model of a GCM with an expectation.

514
00:50:54,710 --> 00:51:01,400
So if and a GCM, if you apply this in a way, there's hope if you applied it right in a place where there's a very,

515
00:51:01,400 --> 00:51:05,300
very deep ocean mixed layer, but you don't have to wait for a long time for this thing to happen.

516
00:51:05,300 --> 00:51:10,100
But once it warms up enough, it'll diffuse just like this, right? So I think that might be the difference.

517
00:51:10,100 --> 00:51:16,490
If you play a strip at 50 degrees north, right, you have the mixed layer that's 4000 metres deep in North Atlantic.

518
00:51:16,490 --> 00:51:26,000
It's going to take a long time to get up to an equilibrium in one spot. But as it's coming up, basically that he's going to be diffused the way.

519
00:51:26,000 --> 00:51:30,680
On the my questions about the the human health effect.

520
00:51:30,680 --> 00:51:40,160
Have you looked at in addition to the absolute stress levels that you showed as the extra stress that people suffer in different areas,

521
00:51:40,160 --> 00:51:45,260
the point being that people who are living on the equator are already used to living it?

522
00:51:45,260 --> 00:51:49,850
Yeah, which is a great and high levels of humidity and the absolute stress that you

523
00:51:49,850 --> 00:51:54,080
show might not be much of an increment over where they are at the moment,

524
00:51:54,080 --> 00:51:58,010
right? The extreme danger one is a physiological thing.

525
00:51:58,010 --> 00:52:02,270
So even if you live in the tropics, you're dead. But the danger zone is is not.

526
00:52:02,270 --> 00:52:08,180
I mean, that's a little bit like where you live. So a lot as you probably know more about this than I do, but a lot locally,

527
00:52:08,180 --> 00:52:14,420
what's considered hot is varies from place to place and depending on how much have you adjusted to it, how much society's adjusted to it.

528
00:52:14,420 --> 00:52:21,980
So this is just a ball. I think the purpose here wasn't to say that's the answer for heat stress in the future, you know, under RCP 8.5,

529
00:52:21,980 --> 00:52:28,070
but but is there a can you can you take shortcuts rather than saying, I need to downscale all my climate models?

530
00:52:28,070 --> 00:52:31,940
Can I just down scale one that has high sensitivity and that's going to give

531
00:52:31,940 --> 00:52:37,940
me like an upper bound in one with a low sense of you give me a lower bound. I think the answer is, yeah, you can get away with that.

532
00:52:37,940 --> 00:52:43,970
But what you what you call critical place. There's lots of ways you can do this and we're starting to look at this.

533
00:52:43,970 --> 00:52:51,560
We're actually working on how heat as stress affects agricultural workers in the US, mostly migrant workers.

534
00:52:51,560 --> 00:52:56,060
You can imagine how popular we are in the US, and that's what we're working on.

535
00:52:56,060 --> 00:53:03,950
And it's it's a tough problem. But that's yeah, it's a question of time over here.

536
00:53:03,950 --> 00:53:13,760
Hi, David. Erm, can you sort of compare the temperature of precipitation effects by saying that the precept was very circulation bound,

537
00:53:13,760 --> 00:53:17,750
whereas temperature is not? I mean, that's only true to some extent.

538
00:53:17,750 --> 00:53:24,590
And for example, here in the UK, heatwaves are clearly linked to circulation anti-psychotics, circulation patterns.

539
00:53:24,590 --> 00:53:27,590
And so any attempt to estimate, you know,

540
00:53:27,590 --> 00:53:37,490
whether the number of of of dangerous or death just deathly heatwaves has got to take into account the propensity for changes in circulation,

541
00:53:37,490 --> 00:53:41,540
for sure. This is something which seem that models do not do a good job.

542
00:53:41,540 --> 00:53:50,870
So I guess my question is you have assumed implicitly that uncertainty is actually represented well by the same models.

543
00:53:50,870 --> 00:53:55,580
But I would suggest that they're at the regional level that could be called into question.

544
00:53:55,580 --> 00:54:02,020
OK, so because I know there's a problem with that in the models, what we did for the future one is we assume the same weather is today.

545
00:54:02,020 --> 00:54:10,350
You know, that's an assumption. But but again, that assumes, you know, that cyclones occur with the same frequency, but that's a crucial question.

546
00:54:10,350 --> 00:54:16,560
Yeah, that's an open question. Yeah. And see, the smart thing is they want to hold on home.

547
00:54:16,560 --> 00:54:22,890
Uh, oh, yeah. The other thing which is we're just counting days like we're not defining heat waves as four days in a row,

548
00:54:22,890 --> 00:54:26,670
so that's another issue that makes this complicated.

549
00:54:26,670 --> 00:54:35,040
But yes, the frequency of blocking how that's going to change the future, those are all really cool science problems.

550
00:54:35,040 --> 00:54:39,000
And you can add that on top of this doesn't change any of the implications of this.

551
00:54:39,000 --> 00:54:40,890
But but because each one different,

552
00:54:40,890 --> 00:54:49,110
whether you want to and it might be interesting to see what's the range of uncertainty and blocking climate models,

553
00:54:49,110 --> 00:54:54,750
and that might be different from the range of uncertainty. Climate sensitivity certainly will be just like that, like in precipitation.

554
00:54:54,750 --> 00:54:58,680
They should say that we didn't try too hard to do patterns of precipitation.

555
00:54:58,680 --> 00:55:04,410
There certainly should be ways like there's some common things like, you know, the rich get richer, poor, get poorer kind of arguments.

556
00:55:04,410 --> 00:55:09,600
People may and every climate, but every time I'll put precipitation in a different place, the tropics.

557
00:55:09,600 --> 00:55:13,830
So just kind of sweeping this under the rug is a little bit unfair because every model,

558
00:55:13,830 --> 00:55:18,120
for example, might do the same thing is an increase in precipitation.

559
00:55:18,120 --> 00:55:25,110
But the ITCZ might be in different place started in each model, which case it's show that there's more uncertainty than there really is.

560
00:55:25,110 --> 00:55:34,050
We didn't try to do that, but there's ways you could maybe eke out more information from from the models in terms of scaling for precipitation.

561
00:55:34,050 --> 00:55:41,500
I've not tried it. It's hard. Yeah, we just take well, this is a question from a layman.

562
00:55:41,500 --> 00:55:48,700
Do your conclusions reinforce the idea that it's less important whether you have relative success in

563
00:55:48,700 --> 00:55:54,280
some regions versus other regions of reducing CO2 emissions and the overall success around the world?

564
00:55:54,280 --> 00:55:59,370
Who was that audience even before the U? OK, wait. Say that again.

565
00:55:59,370 --> 00:56:03,880
How do you can't you come up to the point that it's the global mean change the same?

566
00:56:03,880 --> 00:56:11,530
That's the relevant thing. Some countries might be more successful, or some regions in reducing CO2 emissions are yes than others.

567
00:56:11,530 --> 00:56:16,480
What does that not really matter? Because at the end of the day, overall, what is the total accumulation?

568
00:56:16,480 --> 00:56:21,340
At the end of the day, the atmosphere is really good at distributing. The stuff was that was not known before.

569
00:56:21,340 --> 00:56:29,020
Oh yeah. Yeah, yeah, yeah. Thank you. Way before I was born. Even before me.

570
00:56:29,020 --> 00:56:33,640
Yeah, no. The atmosphere doesn't. I mean, that's what makes us a hard problem.

571
00:56:33,640 --> 00:56:36,550
The atmosphere just accumulates CO2 and distributed.

572
00:56:36,550 --> 00:56:41,590
So the difference in concentration and CO2 around the planet, you know, varies by like eight percent.

573
00:56:41,590 --> 00:56:48,130
And so it doesn't matter who puts it and where it all accumulates, and it comes out thousands of years later.

574
00:56:48,130 --> 00:56:53,970
And if we. If we can wait that long. All right.

575
00:56:53,970 --> 00:56:58,820
I think you have the time before I thank David again. One, one one.

576
00:56:58,820 --> 00:57:01,490
Oh yeah. Oh, I don't know. Can we do it over this?

577
00:57:01,490 --> 00:57:07,280
Do I suggest we do have a drinks before we thank David again for a French talk that has generated lots of discussion,

578
00:57:07,280 --> 00:57:12,850
which will no doubt spill over into into the drinks? I'd like to thank the Oxford Martin School again for hosting us.

579
00:57:12,850 --> 00:57:17,540
And remind everybody and thank you all for coming and remind everyone that there is a drinks reception now just next door.

580
00:57:17,540 --> 00:57:23,210
So please don't dash off. Please come and continue the conversation. Catch David if you haven't had a chance to ask you a question.

581
00:57:23,210 --> 00:57:34,247
But now, as you can just join me again in thanking David for a great.