File talk:Carbon Dioxide 400kyr.png
It is a violation of basic scientific principals to combine, on the same graph, data collected from Hawaii on CO2 levels with levels collected from Ice Cores on the other side of the planet. Not only do these different measurements likely measure different data, but they are likely incompatible as measures of Global CO2. It is especially deceptive to show the high recent CO2 level in Hawaii compared to a prior Warming Period Measurement, based upon ice core samples, because we have no way of knowing if, during a prior warming period, atmospheric levels in Hawaii reached current levels. Hence, presenting the data as a combined chart is dishonest and fails to show the obvious incompatibility of Atmospheric Co2 measurements in Hawaii compared to Ice Cores. I also present the following question for discussion: Does the presence of volcanic activity in Hawaii increase atmospheric Co2 levels?
- Does the presence of volcanic activity in Hawaii increase atmospheric Co2 levels?
- Absolutely. In 1958 when they chose Moana Loa for the site of recording station to monitor background CO2 levels for the globe, they deliberately chose a volcanic island so that CO2 measurements would be inaccurate and greater than global levels. They knew that 50 years later there would be a political conspiracy aimed at generating billions of climate reasearch dollars and this was their contribution towards the conspiracy. That's just how science works.
- .
- Alternatively, if you're not so naive as to think that scientists don't know how to make accurate measurements, and you would like to know some details, have a look at http://http://www.esrl.noaa.gov/gmd/ccgg/trends/co2_data_mlo.html Earth System Researh Laboratory, Atmospheric Carbon Dioxide - Mauna Loa] and be sure to follow the link to the description of how we make measurements at Mauna Loa".
- .
- It is a violation of basic scientific principals to combine, on the same graph, data collected from Hawaii on CO2 levels with levels collected from Ice Cores on the other side of the planet.
- It is? When both are accepted as providing reasonable information on global CO2, does the distance between the two, in either time or space, have so much bearing as to cause a violation?
- .
- Not only do these different measurements likely measure different data
- Different, yes, but incompatible? ".. data from bubbles trapped in polar ice cores. Though not quite as precise as direct atmospheric sampling, these data correlate well with direct measurements during the periods when the two data sets overlap, providing us with confidence that the ice core records are indeed accurate".
- .
- Hence, presenting the data as a combined chart is dishonest and fails to show the obvious incompatibility of Atmospheric Co2 measurements in Hawaii compared to Ice Cores
- On the contrary, I put it to you that the dishonesty is more in your critique than in the science.
- 92.24.113.142 19:17, 6 December 2009 (UTC)
Question about the validity of Hawaiian data and ice cores
[編集]
Well, perhaps violation of scientific principle is rather strong, but it would seem very questionable at the least to compare atmospheric CO2 levels to CO2 levels in ice core samples in this instance. Localized effects both in time and space cannot be ignored. CO2 levels near an active volcanic source should be higher than in ice cores. Now, if you had CO2 levels from Hawaii going back a few thousand years perhaps your argument about increasing CO2 levels would carry more weight.
Down-sampling of the present routinely sampled CO2 level to make a true comparison with the ice-core data possible
[編集]
The removal of the following text in the file description: Please note: the graph shows the historical level of carbon dioxide which is averaged by diffusion of gasses in the ice core and sampled once per approx. 1,000 years, compared to the current level of carbon dioxide which is sampled yearly. The rise due to the industrial revolution compares therefore to less than one sample on the historical scale. For that reason, the shape of the graphs cannot be compared. with the following statements:
- Remove non-sense comment. The shapes are comparable and sampling frequency is often much less than 1000 years for the ice core data
- Saying that it is always 1000 years (which it certainly isn't) and that they can't be compared, isn't in keeping with the scientific understanding of ice core records and paleoclimate
has led to the following discussion on Dragons flight's talkpage. Since he apparently decided to sit mum to end the discussion, I copied the discussion to over here.
I am not talking about the scientific understanding of ice core records and paleoclimate. I am talking about the scientific understanding of sampling and aliasing. I will take my time to improve your current comment, I feel it does not address the (im)possibility of comparing those two incompatible graphs. Jan Arkesteijn (talk) 01:24, 3 March 2012 (UTC)
- Aliasing only matters if you assume that the signal changes rapidly, for paleoclimate and CO2 we have good reasons to believe that such rapid changes do not occur in general. Secondly, even if one is worried about aliasing, it is still relevant to compare the range of variations during the ice ages to the range of variation since the industrial revolution. Specifically, even if one admits the possibility that you are missing some fluctuations, it is nonetheless useful to think about the fluctuations that have been measured. The comparison is still useful and widely made in the paleoclimate literature. Dragons flight (talk) 02:56, 3 March 2012 (UTC)
- Quote: "we have good reasons to believe that such rapid changes do not occur in general" Good reasons? A clincher to me! Good reasons are not enough, Proof should support such a statement. In general? So rapid changes do occur but not to often? Is that what you believe? The point is, there are hardly reasons to believe that (and I believe, believe is for the faithful, not for scientists). The fact that rapid changes are not measured in the ice core does not mean they did not happen. The last century gives evidence for the contrary, unless you regard this by default as an anomaly. (But that is what we are trying to prove, so you should not assume that beforehand.) The point I am making is that it is an error to compare a year by year graph to a graph that is made of samples hundreds of years apart of over hundreds of years diffused and evaporated CO2. There is no validation for that. Measurements of the CO2-level 100 to 400 thousand years ago only gives you an idea of the slow trend (still containing aliasing errors), not about the year by year divergences. Compare the variation in the peak of ca. 130 thousand years ago with the one ca 330 thousand years ago. The variation is much more rapid, indicating that age averages. As for aliasing errors, this occurs when a proces is sampled without in advance being filtered for to high frequency content of the process. Filtering is not possible afterwards. Aliasing errors are therefore a property, an inseperable part of these measurements. Jan Arkesteijn (talk) 15:32, 3 March 2012 (UTC)
- Yes, aliasing is a real process. No, scientists don't believe we are missing large CO2 changes because of it due to the indirect evidence from other sources (e.g. temperature responsive isotopes of O and H, carbon cycle responsive isotopes of C, models of ocean / atmosphere carbon cycle response, etc.) You seem to want to introduce criticism of this plot that goes much further than what the scientific community currently believes. That sort of original research is not really appropriate here. Dragons flight (talk) 19:21, 3 March 2012 (UTC)
- No, there is no original research in my comments. Sampling is a well understood mathematical concept. Anyone knows that the process in-between samples can not be reconstructed in higher detail. It annoys me, though, that you ignore the limitations of comparing those two graphs, with only a very general "we believe"-remark as an explanation, do not address my arguments, and now try to disqualify me with an original research accusation. Jan Arkesteijn (talk) 20:30, 3 March 2012 (UTC)
- Sampling is well understood. The implications of sampling biases depend on the characteristics of the signal. A 500-year sampling is very bad if the underlying process has high spectral power at centennial scales, but it is generally adequate if the underlying process is dominated by changes over tens of thousands of years. One has consider additional information to draw conclusions about the underlying process, and this is well understood in the existing literature. You are making inferences based on sampling alone that ignores the additional information well-known to the climate science community that lead most scientists to conclude there isn't a significant missing signal here. Your broad statements such as "the shape of the graphs cannot be compared" are original research, and ignore the literature that routinely compares such graphs. Knowing that such comparisons are useful requires more information than is represented by these time series alone, but science doesn't occur in a vacuum. Writing about such series requires a knowledge of the larger context in which they occur. Dragons flight (talk) 23:17, 3 March 2012 (UTC)
- Df, the Nyquist frequency for a 500 years sampling rate is 1 per 1000 years (not centennial but millennial scale). Again referring to the peak of the CO2 level at ca. 130 thousand years ago, variations with that
velocityfrequency are not unthinkable. But I’d rather talk about your accusation of original research. To compare different graphs, they should be normalized. That means that the industrial revolution-graph should be down-sampled to one sample per 500 years (let’s stick with your number, if you think that represents the icecore-graph well). To prevent aliasing the data should be averaged first, to filter out frequency components on or above the Nyquist frequency. This is the only way to compare those graphs. It’s not original research, it’s mathematical common sense. Jan Arkesteijn (talk) 23:08, 4 March 2012 (UTC)
- Df, the Nyquist frequency for a 500 years sampling rate is 1 per 1000 years (not centennial but millennial scale). Again referring to the peak of the CO2 level at ca. 130 thousand years ago, variations with that
- Sampling is well understood. The implications of sampling biases depend on the characteristics of the signal. A 500-year sampling is very bad if the underlying process has high spectral power at centennial scales, but it is generally adequate if the underlying process is dominated by changes over tens of thousands of years. One has consider additional information to draw conclusions about the underlying process, and this is well understood in the existing literature. You are making inferences based on sampling alone that ignores the additional information well-known to the climate science community that lead most scientists to conclude there isn't a significant missing signal here. Your broad statements such as "the shape of the graphs cannot be compared" are original research, and ignore the literature that routinely compares such graphs. Knowing that such comparisons are useful requires more information than is represented by these time series alone, but science doesn't occur in a vacuum. Writing about such series requires a knowledge of the larger context in which they occur. Dragons flight (talk) 23:17, 3 March 2012 (UTC)
- No, there is no original research in my comments. Sampling is a well understood mathematical concept. Anyone knows that the process in-between samples can not be reconstructed in higher detail. It annoys me, though, that you ignore the limitations of comparing those two graphs, with only a very general "we believe"-remark as an explanation, do not address my arguments, and now try to disqualify me with an original research accusation. Jan Arkesteijn (talk) 20:30, 3 March 2012 (UTC)
- Yes, aliasing is a real process. No, scientists don't believe we are missing large CO2 changes because of it due to the indirect evidence from other sources (e.g. temperature responsive isotopes of O and H, carbon cycle responsive isotopes of C, models of ocean / atmosphere carbon cycle response, etc.) You seem to want to introduce criticism of this plot that goes much further than what the scientific community currently believes. That sort of original research is not really appropriate here. Dragons flight (talk) 19:21, 3 March 2012 (UTC)
- Quote: "we have good reasons to believe that such rapid changes do not occur in general" Good reasons? A clincher to me! Good reasons are not enough, Proof should support such a statement. In general? So rapid changes do occur but not to often? Is that what you believe? The point is, there are hardly reasons to believe that (and I believe, believe is for the faithful, not for scientists). The fact that rapid changes are not measured in the ice core does not mean they did not happen. The last century gives evidence for the contrary, unless you regard this by default as an anomaly. (But that is what we are trying to prove, so you should not assume that beforehand.) The point I am making is that it is an error to compare a year by year graph to a graph that is made of samples hundreds of years apart of over hundreds of years diffused and evaporated CO2. There is no validation for that. Measurements of the CO2-level 100 to 400 thousand years ago only gives you an idea of the slow trend (still containing aliasing errors), not about the year by year divergences. Compare the variation in the peak of ca. 130 thousand years ago with the one ca 330 thousand years ago. The variation is much more rapid, indicating that age averages. As for aliasing errors, this occurs when a proces is sampled without in advance being filtered for to high frequency content of the process. Filtering is not possible afterwards. Aliasing errors are therefore a property, an inseperable part of these measurements. Jan Arkesteijn (talk) 15:32, 3 March 2012 (UTC)
