Prev  1  2  3  4  5  6  7  8  Next

Comments 151 to 175 out of 190:

151. Tom Curtis at 06:25 AM on 4 December, 2012
     Further to my comment @146, in trying to follow up some further points I downloaded the HadSST3 data for the Nino 3.4 region, using a 15% data cutoff. I then calculated the anomaly average for 1912-1944, 1944-1976, and 1976-2006. (2006 is the last year with data available from the KNMI Climate Explorer. The calculated means are: 1912-1944: -0.068 C (compared to Tisdale's claimed 0.18 C); 1944-1976: -0.042 C (compared to Tisdale's claimed -0.05 C); and 1976-2006: 0.303 C (compared to Tisdale's claimed 1976-2011 mean of 0.16 C) Elsewhere Tisdale has claimed:

     "By smoothing NINO3.4 SST anomalies with a 121-month filter, Figure 3, the periods when El Nino and La Nina events dominated become apparent."

     While not agreeing with the supposition, the 121 month running mean of the data certainly shows Tisdale's claim that the mean of the period 1912-1944 is greater than that from 1944-1976, and even then 1976-2011 to be false: Tisdale does not specify which version of the HadSST he uses, so I obtained the graph of the HadSST2 anomaly data for the Nino 3.4 region to ensure the problem was not just the updated data set: You can easily detect the periods of interest. 1912 immediately follows the very large negative excursion just after 1900 (the largest La Nina event on record according to the SOI). 1944 immediately follows the large positive excursion following 1940. 1976 immediately follows the double troughed negative excursion preceding 1980. The data from 1912-1944 is a little noisy, even though I used the 0% cutoff on data to net every observation in the region. Never-the-less it is clear that the peaks (El Ninos) in 1912-1944 are not as strong as those from 1976 forward; and that the troughs (La Ninas) are both deeper and more frequent. Therefore the difference between my and Tisdale's result is not a consequence of the different data sets. It looks to me as though Tisdale has subtracted a linear trend from the anomaly. However, he does not mention doing so. Bob, would you please specify the exact methods used in obtaining your averages? And would you also specify the exact cutoff dates for your periods of interest?
152. Kevin C at 07:33 AM on 4 December, 2012
     So far I’ve constructed a plausible model which the statistics say is more informative concerning the rest-of-the-world SST data. The next steps are as follows: 1. Dismantle the model to find out how it works. That’s helpful in trying to relate the model to mechanism, which is after all what we are after. 2. Validate the model: Test whether the model is robust to the choice of data and whether it has predictive power. I’ve been working on the first of these. So far I’ve highlighted how the use of two lags improves the fit to the data, as well as reducing the impact of the two anomalous features Bob identified. Whether that reduction is justified we’ll come to in the validation step. However once I started breaking down the model, it became obvious that the two-lag effect, while it makes a difference, is not the principle effect in reducing the two anomalies. The most significant causes are much simpler: The volcanic term, the trend and the scale. These are strongly interlinked in ways which are counterintuitive. Initially it seems implausible that the volcanic term should contribute to the anomalies, because they do not overlap in the time series. However, the interaction of the unmodeled volcanic term and the detrending of the SST data have exactly that effect. The reason is shown in the following animation, which shows the removal of a trend and offset from the volcanic term: Because the volcanoes are negative, detrending them shifts the whole series upwards. Because they are offset towards the start of the frame, the beginning is shifted up more. This creates a positive bias for the non-volcano influenced periods. Add a noise signal and negative excursions become ‘normal termperatures’ while positive excursions become ‘anomalies’. Now the SST signal contains the volcanic term, so detrending the SST data without removing the volcanic term introduces exactly this kind of bias. And the bias is significant in the two regions where the two anomalies occur. We would have detected this straight away had we looked at the data on a longer time frame. The interplay of all these factors in the model is moderately complex. In the following animation I’ve rolled the volcanic term into the Nino data and scaled it from zero to full scale during the animation. That takes us from the simplest Nino34-only model to a single-lag+volcano model. The model is regressed against the data with trend and offset. R2 is shown top right. The other point of note is the scale of the model. At the start the fit is very poor owing to the omission of the volcanic term from the model. The scale of the model curve is also very small - it looks too small. In fact this is the expected behaviour. The explanation is non-trivial, but is consistent with determining the ‘best’ (most probable and/or least noisy) model under the assumptions of least squares. The result can be understood from a perspective of likelihood, or least squares, or random walk statistics, but I’ve not heard a good intuitive explanation. Here are my best efforts: 1. When the model is poor, it contains features which agree and features which disagree with the data. Inflating the model increases both, but the parts that disagree quickly begin to hurt you more, so the optimum scale is smaller than the data. (From least squares.) 2. When the model is poor, you have to leave some variance for the terms that are missing from the model, and so the incomplete model must be on a smaller scale than the data. (From random walk.) This is related to the well known result that when calculating a best fit straight line through some data you get different results depending on which is the dependent variable. In other words, the scaling of the model is chosen to minimise artifacts due to the limitations of the model. It's trying to protect you from interpreting noise as signal. You can of course change the appearance of the graph by playing with the offset, scale and trend, or by adopting heuristics for setting each of these. However, there is only one ‘most probable’ model, and that is the one produced by the least squares calculation subject to the OLS assumptions. So to summarise, the two anomalies identified by Bob appear to arise from the confluence of 3 factors: 1. Bias in the trend and offset caused by the omission of the volcanic term. 2. The use of a single rather than double lag. 3. The scaling of the Nino34 data. That’s the analysis step. As I’ve said, the next step is validation. We need to test whether this model is robust - i.e. whether the results are strongly dependent on the choice of data, and whether it has predictive power.
153. Bernard J. at 11:38 AM on 4 December, 2012
     Kevin C. Kudos for taking the time to tease apart some of the artifactual conclusions in Bob Tisdale's approach. Perhaps if he had been careful to consider and to address the fundamental questions repeatedly put to him, he might have been able to rise above his biases and see the bigger picture, and the factors that impinge on it... assuming that he does not already do so in private. I'm looking forward with interest to the validation post. Between you and Tom Curtis' patient work, perhaps Tisdale will realise that he's holding the wrong end of the stick (again, assuming that he does not already do so in private). And if he cares to refute your work and Tom's perhaps finally he'll do so with some recognition and explicit incorporation and description of the breadth of underlying physical influences.
154. vrooomie at 06:21 AM on 5 December, 2012
     It will be interesting to see if Bob returns to answer the questions he's yet to answer: however, in the meantime, he's back at W**T, galloping along with the same tired, rebutted points that he wouldn't answer here. I was looking forward to his answers.
155. Kevin C at 22:10 PM on 6 December, 2012
     The final step in testing the the validity of Bob’s and my models of the relationship between Nino34 and Rest-of-the-world SST (and what conclusions may be drawn from them) is to validate the models. I thought this was going to take multiple posts, but the results are so clear cut that we can do it in one. The most fundamental test which can be applied to any scientific hypothesis is whether it has predictive power - i.e. whether it can predict observations which were not used in making the hypothesis. In Popper’s model of the scientific method, this constitutes a ‘severe test’, which can falsify a model if the prediction is wrong, or strengthen our confidence in it if the prediction is right. So let’s apply that kind of test to both models. We need some data which the models have not seen. Fortuntely, ERSST SST and Nino34 data are available for over a century, although on monthly rather than weekly resolution. The change in resolution means that we will have to ensure we can reproduce Bob’s results when we switch to the longer data run. The ERSST monthly anomaly and Nino data (in various versions) is available on the KNMI data explorer, and you can also select latitude and longitude limits to reproduce Bob’s rest-of-the-world SST anomaly. I’ve made two relevant claims here:  Firstly that Bob’s detrending of the data on 1982-2012 without including a volcanic term produces the wrong result, and secondly that matching the scales of the Nino and temperature data produces artifacts through upscaling the artifacts caused by inadequacies of the model. To test those, we’ll apply Bob’s scaling method - i.e. detrend only on the 1982- data, and match the variances of the data and model. The result is shown below: First we need to check that the post 1982 features are still present - they are, and show the same features as Bob identified in the weekly data. However, when we look at the pre 1982 agreement, the fit is nonexistent. As I predicted, the trend is well off. So, what if we were to fix the trend? We’ll let Bob’s model cheat and see the trend for the whole data by detrending the SSTs on the entire period. What you got is this: The fit is still very poor. If we follow Bob’s approach and interpret the deviations, we do see the Mt Agung eruption in 1962, but we also predict major ‘warming events’ in the 50’s and early 60’s. No such steps are present in the SST data - in fact the 40's-60's show very little trend at all. That’s it. The hypothesis is falsified. For completeness, we can also apply the same tests to my model. Since I fit the Nino term, the volcano term and the trend simultaneously, it makes no difference to the model if or how how we detrend the data beforehand. Here is the model fit using the temperature data detrended as in the first figure above: And here is the same result using the trend from the second figure: The fit is fairly good, and well within the uncertainty bounds on the trend term. The 2-lag + volcano model seems to have predictive power, even to the extent of predicting a trend over a period equal to the model period. Finally, we can check what my model gives when it sees all the data: Apart from a slight correction to the trend, the results are almost identical. So that’s it. Not only do the post-1982 statistics overwhelmingly favour my model over Bob’s, Bob’s method is falsified by looking at the pre-1982 data, whereas mine performs pretty well. In retrospect we could have saved a lot of time by going straight to the longer data, but then we probably wouldn’t have worked out all the details of what was going on. It's been suggested that I try and write an accessible summary of this. I'll try, but I've put in a lot of time already, so don't expect it immediately.
156. Bernard J. at 22:34 PM on 6 December, 2012
     Kevin, your compelling posts as they stand are quite accessible anyway, so I wouldn't rush to a summary version. And it would be extremely interesting to first wait for Tisdale's response to your approach. He might also consider properly answering the hanging questions directed at him, so that the underlying physical assumptions of your model and his can be appropriately referenced and grounded in real-world conditions. Without a proper summary from him, beginning at first principles, his implicit-bootstrap model of global warming is laying rather bloodied in the middle of the road.
157. KR at 03:12 AM on 7 December, 2012
     Kevin C - Extremely clear, thank you so much for working through the details. It certainly looks like Tisdale has been working from a bad model, one that simply doesn't fit the data. This seems to be a theme in some of the more scientific 'skeptic' approaches - build a model (which doesn't match reality), then make claims contrary to the consensus only supported by that (poor) model. We see it here with Tisdale - other examples include both Spencer and Lindzen on climate sensitivity and feedbacks, and many skeptic temperature predictions. Bad models lead to incorrect conclusions - Kevin C's two-lag model including volcanic forcing is demonstrably a better match to the data (and quite frankly to the physics) than anything so far presented by Tisdale.
158. Bob Tisdale at 19:31 PM on 7 December, 2012

     Sorry for the delay. I was responding to comments at other blogs and writing new blog posts.

     skywatcher at 138 says: Bob, in all your focus on one region of the Earth, you have apparently neatly dodged important questions with relation to global warming and your unusual conjectures…”

     Apparently, skywatcher, you’ve missed the fact that the discussion of sea surface temperatures was broken down into two parts of the global oceans, the East Pacific (90S-90N, 180W-80W) and the Rest-of-the-World (90S-90N, 80W-180). How is that “one region of the Earth”, skywalker? Sure looks like it covers the global oceans to me.

     Also, are you aware that the warming of land surface air temperatures is primarily in response to the warming of the oceans, skywatcher? Refer to Compo and Sardeshmukh 2009: Oceanic influences on recent continental warming. The abstract reads:

     Evidence is presented that the recent worldwide land warming has occurred largely in response to a worldwide warming of the oceans rather than as a direct response to increasing greenhouse gases (GHGs) over land. Atmospheric model simulations of the last half-century with prescribed observed ocean temperature changes, but without prescribed GHG changes, account for most of the land warming. The oceanic influence has occurred through hydrodynamic-radiative teleconnections, primarily by moistening and warming the air over land and increasing the downward longwave radiation at the surface. The oceans may themselves have warmed from a combination of natural and anthropogenic influences.

     That study was based on climate models by the way and you can get an idea of the order of magnitude using the ModelE Climate Simulations – Climate Simulations for 1880-2003 webpage, specifically Table 3.

     And as I’ve shown, the satellite-era sea surface temperature and ocean heat content records do not confirm the existence of an anthropogenic warming component. Maybe you’ve missed this: the divergences during the La Niña events of 1988/89 and 1998-01 are when the Rest of the World data acquires its long-term trend. If the Rest-of-the-World sea surface temperature data cooled proportionally during the La Niña events of 1988/89 and 1998-2001, it would have no warming trend like the East Pacific!

159. Bob Tisdale at 19:34 PM on 7 December, 2012

     skywatcher at 138 says: “1: Where's the heat coming from? The oceans, globally, are warming, the atmosphere is warming, and yet the Sun is not getting any brighter. What's your energy source?”

     The sun is the primary energy source, but sea surface and ocean heat content warming can and do also take place without the exchange of heat, which is the result of teleconnections. The process through which the sun creates the warm water for El Niño events was described in my comment 139, where I replied to composer99:

     El Niño and La Niña events are part of a coupled ocean-atmosphere process. Sea surface temperatures, trade winds, cloud cover, downward shortwave radiation (aka visible sunlight), ocean heat content, and subsurface ocean processes (upwelling, subsurface currents, thermocline depth, downwelling and upwelling Kelvin waves, etc.) all interact. They’re dependent on one another. During a La Nina, trade winds are stronger than normal. The stronger trade winds reduce cloud cover, which, in turn, allows more downward shortwave radiation to enter and warm the tropical Pacific.

     If you’re having trouble with my explanation because it’s so simple, refer to Pavlakis et al (2008) paper “ENSO Surface Shortwave Radiation Forcing over the Tropical Pacific.” Note the inverse relationship between downward shortwave radiation and the sea surface temperature anomalies of the NINO3.4 region in their Figure 6. During El Niño events, warm water from the surface and below the surface of the West Pacific Warm Pool slosh east, so the sea surface temperatures of the NINO3.4 region warm, causing more evaporation and more clouds, which reduce downward shortwave radiation. During La Niña events, stronger trade winds cause more upwelling of cool water from below the surface of the eastern equatorial Pacific, so sea surface temperature to drop in the NINO3.4 region, in turn causing less evaporation. The stronger trade winds also push cloud cover farther to the west than normal. As a result of the reduced cloud cover, more downward shortwave radiation is allowed to enter and warm the tropical Pacific during La Niña events.

     To complement that, here’s a graph to show the interrelationship between the sea surface temperature anomalies of the NINO3.4 region and cloud cover for the regions presented by Pavlakis et al.

     That discussion explains why the long-term warming of the Ocean Heat Content for the tropical Pacific was caused by the 3-year La Nina events and the unusual 1995/96 La Niña. First, here’s a graph of tropical Pacific Ocean Heat Content. It’s color coded to isolate the data between and after the 3-year La Niña events of 1954-57, 1973-76 and 1998-2001. Those La Niña events are shown in red. Note how the ocean heat content there cools between the 3-year La Niña events. Anyone who understands ENSO would easily comprehend how and why that happens. It’s tough to claim that greenhouse gases have caused the warming of the tropical Pacific when the tropical Pacific cools for multidecadal periods between the 3-year La Niñas, Composer99.

     As you can see, the warming that took place during the 1995/96 La Niña was freakish. Refer to McPhaden 1999 “Genesis and Evolution of the 1997-98 El Niño”.

     McPhaden writes:

     For at least a year before the onset of the 1997–98 El Niño, there was a buildup of heat content in the western equatorial Pacific due to stronger than normal trade winds associated with a weak La Niña in 1995–96.

     Based on the earlier description, that “build up of heat content” resulted from the interdependence of trade winds, cloud cover, downward shortwave radiation and ocean heat content. Simple. As you can see in the above graph, the upward spike caused by the 1995/96 La Niña skews the trend of the mid-cooling period, and if we eliminate the data associated with it and the 1997/98 El Niño, then the trend line for the mid-period falls into line with the others.

     HHH

     skywatcher, so La Niña events provide the naturally created fuel for El Niño events. More ENSO basics: An El Niño releases that heat, which is stored as warm water, from below the surface of the West Pacific Warm Pool. The warm water travels east and spreads across the surface of the eastern tropical Pacific. The El Niño releases heat through evaporation into the atmosphere, which is why tropical Pacific ocean heat content drops during an El Niño. Additionally, the convection, cloud cover and precipitation accompany that warm water east, sometimes almost halfway around the globe. These changes in location of the primary source of tropical Pacific convection and the resulting changes in atmospheric circulation, not a direct transfer of heat, are what cause surface temperatures to warm in areas remote to the eastern tropical Pacific.

     Further to this, as I replied to doug_bostrom in my comment 103:

     On the other hand, are you aware of teleconnections? Are you aware that there’s no heat transfer with teleconnections? Example: Why do the tropical North Atlantic sea surface temperature anomalies warm during an El Nino, doug? Do you know? There’s no direct exchange of heat yet the tropical North Atlantic warms during an El Niño. Why, doug? Could it have something to do with the slowing of the trade winds in the tropical North Atlantic in response to the El Niño? That would result in less evaporation, which is the primary way the oceans release heat. If there’s less evaporation, sea surface temperatures warm, do they not? Also, when the trade winds slow in the tropical North Atlantic in response to an El Niño, there’s less upwelling of cool waters from below the surface and less entrainment of that cool subsurface water. That would cause the seas surface temperatures to warm too.

     HHH

     If you don’t like my explanation, skywatcher, refer to Wang (2005) ENSO, Atlantic Climate Variability, And The Walker And Hadley Circulation for a more detailed discussion. And if you’d like a discussion of teleconnections for the rest of the world, refer to Trenberth et al (2002) Evolution of El Niño–Southern Oscillation and global atmospheric surface temperatures.

     A quick note about Trenberth et al (2002). They qualify the results of their global warming attribution with:

     Although it is possible to use regression to eliminate the linear portion of the global mean temperature signal associated with ENSO, the processes that contribute regionally to the global mean differ considerably, and the linear approach likely leaves an ENSO residual.

     The divergences between the Rest of the World data and the scaled NINO3.4 data during the 1988/89 and 1998-01 La Niña events are those ENSO residuals.

160. Bob Tisdale at 19:36 PM on 7 December, 2012

     skywatcher at 138 says: “2: What's the physical mechanism involved, if it's not the rise in greenhouse gases? Sloshing water about the oceans does not appear adequate if the oceans, as a whole, are warming.”

     Your representation of ENSO as “Sloshing water about the oceans” clearly indicates you lack a basic understanding of the topics being discussed. Please refer to my preceding reply to you for the answer to this question.

161. Bob Tisdale at 19:39 PM on 7 December, 2012

     skywatcher at 138 says: “3: Why is this mechanism unidirectional, when ENSO is oscillatory?”

     The impacts of ENSO are clearly not oscillatory. One look at the tropical Pacific OHC record clearly dismisses that idea. They would be better described as a discharge-recharge oscillator, where the recharge takes place during La Niña phases and the discharge takes place during El Niño. Also, the recharge varies, as can be seen in the response of the tropical Pacific Ocean Heat Content to the 1995/96 La Niña. I described above the recharge and discharge phases and the resulting teleconnections and confirmed it with data and with peer-reviewed papers. To further contradict your assertion that ENSO is oscillatory, with respect to sea surface temperatures, an El Niño releases warm water from below the surface of the west Pacific Warm Pool and the warm water spreads across the eastern tropical Pacific. That was the discharge phase. An El Niño does not consume the warm water. At the end of the El Niño, the warm water on the surface that’s left over from the El Niño is carried to the West Pacific and tropical East Indian Oceans, raising the surface temperatures there. The reverse does not occur during a La Niña. Further, there’s leftover warm water below the surface after the El Niño.

     This leftover warm water was discussed in my reply to KevinC at 135:

     The reasons for the divergences in the Rest-Of-the-World data during the 1988/89 and 1998-2001 La Niñas are physical, KevinC. You can try to eliminate or minimize them using models, but they exist. East Pacific El Niños like the 1986/87/88 and 1997/98 El Niños release vast amounts of warm water from below the surface of the west Pacific Warm Pool. Much of that warm water spreads across the surface of the central and eastern tropical Pacific. For the East Pacific El Niño events, like those in 1986/87/88 and 1997/98, that warm water impacts the surface all the way to the coast of the Americas (while with Central Pacific El Niño events it does not). The El Niños do not “consume” all of the warm water. At the conclusion of an El Niño, the trade winds push the leftover warm surface water back to the West Pacific. Additionally, there is left over warm water below the surface that’s returned to the west Pacific and into the East Indian Ocean via a Rossby wave or Rossby waves. This animation captures a Rossby wave returning warm water to the West Pacific and East Indian Oceans after the 1997/98 El Niño. Watch what happens when it hits Indonesia. It’s like there’s a secondary El Niño taking place in the Western Tropical Pacific and it’s happening during the La Niña. It’s difficult to miss it. (The full JPL animation is here.) Gravity causes that warm water to rise to the surface with time. The leftover warm water exists and it cannot be accounted for with a statistical model based on an ENSO index. You can see—you can watch it happen—the impacts of that warm water in this animation. There are no ENSO indices that account for the leftover warm water.

162. Bob Tisdale at 19:41 PM on 7 December, 2012

     skywatcher at 138 says: “4: Assuming that a new unidirectional process must be a recent or temporary occurrence, otherwise we would have boiled or frozen awfuly quickly ... Why is your proposed physical mechanism a recent occurrence, when ENSO has been around for millennia, perhaps hundreds of thousands of years?”

     The multiple assumptions are yours not mine and your assumptions are monumental. For example, we’ve got reasonable sea surface temperature-based ENSO records dating back to the opening of the Panama Canal in 1914. Can you prove ENSO did not have the same effect and cause the warming from about 1917 to 1944? I’ve limited my discussion to the satellite era of sea surface temperatures because the data is globally complete and it’s considered to be “the truth”. As I replied to Michael Sweet in my comment 133, which was a cut and paste from my book:

     Refer to Smith and Reynolds (2004) Improved Extended Reconstruction of SST (1854-1997). It is about the Reynolds OI.v2 data we’ll be using as the primary source of data for this book:

     Although the NOAA OI analysis contains some noise due to its use of different data types and bias corrections for satellite data, it is dominated by satellite data and gives a good estimate of the truth.

     The truth is a good thing, don’cha think?

     skywatcher, as I’ve shown repeatedly throughout this exchange, the satellite-era sea surface temperature records and the ocean heat content data do not confirm the presence of an anthropogenic warming signal. You can believe what you want, but I refer to data, not climate models, for my understandings

163. Bob Tisdale at 19:42 PM on 7 December, 2012

     skywatcher at 138 says: “5: Why is the well-understood mechanism of an enhanced greenhouse effect from long-lived GHGs not operating according to their physics, despite this physics neatly explaining both present climate and palaeoclimate changes?”

     I’ll repeat my basic statement once more: the satellite-era sea surface temperature records and the ocean heat content records do not confirm the existence of an anthropogenic global warming component. And as I replied to KR at 134:

     Downward longwave radiation appears to do nothing more cause a little more evaporation from the ocean surface, which makes perfect sense since it only penetrates the top few millimeters.

164. Bob Tisdale at 19:44 PM on 7 December, 2012

     John Hartz says at 141: “You have not responded to the question I posed to you in #121. For everyone's convenience, I will repost it here… Do you believe the following graph to be a valid representation of Jan-Oct global land & and surface temperature anomalies with respect to the 1961-1990 base period for calendar years 1950 through 2012?”

     The data in that graph appears to be an honest representation of the data, but would you like to explain what this has to do with a discussion about the long-term effects of ENSO?

     To save you the time needed to scroll up, the following is what I wrote in an earlier comment to skywatcher:

     Are you aware that the warming of land surface air temperatures are primarily in response to the warming of the oceans, skywatcher? Refer to Compo and Sardeshmukh 2009: Oceanic influences on recent continental warming. The abstract reads:

     Evidence is presented that the recent worldwide land warming has occurred largely in response to a worldwide warming of the oceans rather than as a direct response to increasing greenhouse gases (GHGs) over land. Atmospheric model simulations of the last half-century with prescribed observed ocean temperature changes, but without prescribed GHG changes, account for most of the land warming. The oceanic influence has occurred through hydrodynamic-radiative teleconnections, primarily by moistening and warming the air over land and increasing the downward longwave radiation at the surface. The oceans may themselves have warmed from a combination of natural and anthropogenic influences.

     That study was based on climate models by the way and you can confirm the results at the ModelE Climate Simulations – Climate Simulations for 1880-2003 webpage, specifically Table 3.

     HHH

     And as I’ve shown, John, the satellite-era sea surface temperature and ocean heat content records do not confirm the existence of an anthropogenic component.

     Could you have missed the importance of the divergences during the 1988/89 and 1998-01 La Niñas, John? The global oceans did not respond proportionally to the La Niña events of 1988/89 and 1998-01. That failure to cool proportionally during those La Ninas was why the sea surface temperatures for the Atlantic-Indian-West Pacific oceans (referred to as the Rest-of-the-World in earlier comments) acquired its long-term warming trend.

165. Bob Tisdale at 19:48 PM on 7 December, 2012

     doug_bostrom says at 142: “Bob, where's the heat pump? How does it function?”

     I don’t know how you’re defining heat pump, doug.

     But I’ll cut and paste my earlier reply to skywatcher, as an explanation of ENSO to see if it agrees with how you’re defining heat pump. I wrote:

     The process through which the sun creates the warm water for El Niño events was described in my comment 139, where I replied to composer99:

     El Niño and La Niña events are part of a coupled ocean-atmosphere process. Sea surface temperatures, trade winds, cloud cover, downward shortwave radiation (aka visible sunlight), ocean heat content, and subsurface ocean processes (upwelling, subsurface currents, thermocline depth, downwelling and upwelling Kelvin waves, etc.) all interact. They’re dependent on one another. During a La Nina, trade winds are stronger than normal. The stronger trade winds reduce cloud cover, which, in turn, allows more downward shortwave radiation to enter and warm the tropical Pacific.

     If you’re having trouble with my explanation because it’s so simple, refer to Pavlakis et al (2008) paper “ENSO Surface Shortwave Radiation Forcing over the Tropical Pacific.” Note the inverse relationship between downward shortwave radiation and the sea surface temperature anomalies of the NINO3.4 region in their Figure 6. During El Niño events, warm water from the surface and below the surface of the West Pacific Warm Pool slosh east, so the sea surface temperatures of the NINO3.4 region warm, causing more evaporation and more clouds, which reduce downward shortwave radiation. During La Niña events, stronger trade winds cause more upwelling of cool water from below the surface of the eastern equatorial Pacific, so sea surface temperature to drop in the NINO3.4 region, in turn causing less evaporation. The stronger trade winds also push cloud cover farther to the west than normal. As a result of the reduced cloud cover, more downward shortwave radiation is allowed to enter and warm the tropical Pacific during La Niña events.

     To complement that, here’s a graph to show the interrelationship between the sea surface temperature anomalies of the NINO3.4 region and cloud cover for the regions presented by Pavlakis et al.

     That discussion explains why the long-term warming of the Ocean Heat Content for the tropical Pacific was caused by the 3-year La Nina events and the unusual 1995/96 La Niña. First, here’s a graph of tropical Pacific Ocean Heat Content. It’s color coded to isolate the data between and after the 3-year La Niña events of 1954-57, 1973-76 and 1998-2001. Those La Niña events are shown in red. Note how the ocean heat content there cools between the 3-year La Niña events. Anyone who understands ENSO would easily comprehend how and why that happens. It’s tough to claim that greenhouse gases have caused the warming of the tropical Pacific when the tropical Pacific cools for multidecadal periods between the 3-year La Niñas, Composer99.

     As you can see, the warming that took place during the 1995/96 La Niña was freakish. Refer to McPhaden 1999 “Genesis and Evolution of the 1997-98 El Niño”.

     McPhaden writes:

     For at least a year before the onset of the 1997–98 El Niño, there was a buildup of heat content in the western equatorial Pacific due to stronger than normal trade winds associated with a weak La Niña in 1995–96.

     Based on the earlier description, that “build up of heat content” resulted from the interdependence of trade winds, cloud cover, downward shortwave radiation and ocean heat content. Simple. As you can see in the above graph, the upward spike caused by the 1995/96 La Niña skews the trend of the mid-cooling period, and if we eliminate the data associated with it and the 1997/98 El Niño, then the trend line for the mid-period falls into line with the others.

     HHH

     skywatcher, so La Niña events provide the naturally created fuel for El Niño events. More ENSO basics: An El Niño releases that heat, which is stored as warm water, from below the surface of the West Pacific Warm Pool. The warm water travels east and spreads across the surface of the eastern tropical Pacific. The El Niño releases heat through evaporation into the atmosphere, which is why tropical Pacific ocean heat content drops during an El Niño. Additionally, the convection, cloud cover and precipitation accompany that warm water east, sometimes almost halfway around the globe. These changes in location of the primary source of tropical Pacific convection and the resulting changes in atmospheric circulation, not a direct transfer of heat, are what cause surface temperatures to warm in areas remote to the eastern tropical Pacific.

     Further to this, as I replied to doug_bostrom in my comment 103:

     On the other hand, are you aware of teleconnections? Are you aware that there’s no heat transfer with teleconnections? Example: Why do the tropical North Atlantic sea surface temperature anomalies warm during an El Nino, doug? Do you know? There’s no direct exchange of heat yet the tropical North Atlantic warms during an El Niño. Why, doug? Could it have something to do with the slowing of the trade winds in the tropical North Atlantic in response to the El Niño? That would result in less evaporation, which is the primary way the oceans release heat. If there’s less evaporation, sea surface temperatures warm, do they not? Also, when the trade winds slow in the tropical North Atlantic in response to an El Niño, there’s less upwelling of cool waters from below the surface and less entrainment of that cool subsurface water. That would cause the seas surface temperatures to warm too.

     HHH

     If you don’t like my explanation, skywatcher, refer to Wang (2005) ENSO, Atlantic Climate Variability, And The Walker And Hadley Circulation for a more detailed discussion. And if you’d like a discussion of teleconnections for the rest of the world, refer to Trenberth et al (2002) Evolution of El Niño–Southern Oscillation and global atmospheric surface temperatures.

     A quick note about Trenberth et al (2002). They qualify the results of their global warming attribution with:

     Although it is possible to use regression to eliminate the linear portion of the global mean temperature signal associated with ENSO, the processes that contribute regionally to the global mean differ considerably, and the linear approach likely leaves an ENSO residual.

     The divergences between the Rest of the World data and the scaled NINO3.4 data during the 1988/89 and 1998-01 La Niña events are those ENSO residuals.

     HHH

     Would you consider ENSO to be a heat pump, doug?

166. Bob Tisdale at 19:51 PM on 7 December, 2012

     IanC says at 143: “So in short, you are very certain that you are correct, and see no possible reason for your analysis to be flawed? By the way, I'm mainly referring to PDO and decadal temperature trends.”

     Is this a repeated question, IanC? If so, why are you asking it when I’ve already answered it? If not, then I’ll ask, which decadal temperature trends? Global? East Pacific? North Pacific? Pacific? If it’s the latter 3, I’ll repeat my earlier answer:

     From my comment 103: As I noted earlier, the PDO does not represent the sea surface temperature of the North Pacific, the Pacific basin as a whole, or the East Pacific. Here’s a graph that compares the decadal variability of the PDO and the detrended and standardized sea surface temperature anomalies of the East Pacific, North Pacific (north of 20N, same as the PDO), and the Pacific as a whole:
     http://i49.tinypic.com/slhb8y.jpg

167. Bob Tisdale at 19:52 PM on 7 December, 2012

     Kevin C at 144: “You implicit model seems to be something like this: Ignoring volcanoes, the rest of the world temperatures track Nino34 with a single lag, except for some significant deviations which require another explanation.”

     I’ve presented those explanations to you. You can watch them happen. Here’s my earlier reply:

     The reasons for the divergences in the Rest-Of-the-World data during the 1988/89 and 1998-2001 La Niñas are physical, KevinC. You can try to eliminate or minimize them using models, but they exist. East Pacific El Niños like the 1986/87/88 and 1997/98 El Niños release vast amounts of warm water from below the surface of the west Pacific Warm Pool. Much of that warm water spreads across the surface of the central and eastern tropical Pacific. For the East Pacific El Niño events, like those in 1986/87/88 and 1997/98, that warm water impacts the surface all the way to the coast of the Americas (while with Central Pacific El Niño events it does not). The El Niños do not “consume” all of the warm water. At the conclusion of an El Niño, the trade winds push the leftover warm surface water back to the West Pacific. Additionally, there is left over warm water below the surface that’s returned to the west Pacific and into the East Indian Ocean via a Rossby wave or Rossby waves. This animation captures a Rossby wave returning warm water to the West Pacific and East Indian Oceans after the 1997/98 El Niño. Watch what happens when it hits Indonesia. It’s like there’s a secondary El Niño taking place in the Western Tropical Pacific and it’s happening during the La Niña. It’s difficult to miss it. (The full JPL animation is here.) Gravity causes that warm water to rise to the surface with time. The leftover warm water exists and it cannot be accounted for with a statistical model based on an ENSO index. You can see—you can watch it happen—the impacts of that warm water in this animation. There are no ENSO indices that account for the leftover warm water.

     HHH

     How does your model account for the leftover warm water, KevinC?

     Same replies to KevinC at 149 and 152.

168. Bob Tisdale at 19:54 PM on 7 December, 2012

     Bernard J. says at 145: “Specifically, I would like to know by exactly how much Tisdale believes that "[d]ownward longwave radiation" does or does not contribute to warming of the planet in terms of a particular DLR flux, and by what mechanisms that warming does - or indeed does not - does not occur.”

     I believe this has been the underlying topic of discussion, Bernard J. The satellite-era sea surface temperature records and the ocean heat content records do not indicate that downward shortwave radiation has made any contribution to the warming.

     Bernard J. says at 145: “Numbers and primary references would assist to make an objective and nuanced case.”

     I’ll present the four datasets once again. For sea surface temperatures there’s the East Pacific Ocean and the Rest of the World data. In case you’ve missed this, the divergences during the La Niña events of 1988/89 and 1998-01 are when the Rest of the World data acquires its long-term trend. If the Rest-of-the-World sea surface temperature data cooled proportionally during the La Niña events of 1988/89 and 1998-2001, it would have no warming trend like the East Pacific! For ocean heat content, I presented as examples the Tropical Pacific and the North Pacific north of the tropics, both of which contradict the hypothesis of manmade greenhouse gas-driven global warming.

169. Bob Tisdale at 19:58 PM on 7 December, 2012

     Tom Curtis at 146 says: “Why did Tisdale include a La Nina dominated period in his final period, when he must know that doing so will distort the comparison with the earlier period? And if La Nina dominated periods cause cooling, why has there been no cooling in the La Nina dominated period from Jan 1999 to the present?”

     Tom, you’re referring to this graph. If you had been following the conversations on this thread you would have discovered that the global oceans did not respond proportionally to the La Niña events of 1988/89 and 1998-01 and that that failure to cool proportionally during those La Ninas was why the sea surface temperatures for the Atlantic-Indian-West Pacific oceans (referred to as the Rest-of-the-World in earlier comments) acquired its long-term warming trend. So why then are breaking the latest warming period at 1999 during the 1998-2001 La Nina when the Atlantic-Indian-West Pacific data acquires part of its long-term trend? If the Atlantic-Indian-West Pacific data cooled proportionally during the La Niña events of 1988/89 and 1998-2001, it would have no warming trend like the East Pacific!

     Tom Curtis at 146 says: “Interestingly, in his comment @137 Tisdale quotes Compo and Sardeshmuk…”

     You then go on, Tom, to attempt (and fail at your attempt) to fault me for using a single index in the above graph, but then immediately compromise your position by referring to your preferred single index, the Southern Oscillation Index.

     Second, you’ve apparently misunderstood Compo and Sardeshmuk. Their discussion of a using a single index had to do with regression analysis. They wrote:

     In particular, defining ENSO in terms of a single index and ENSO-related variations in terms of regressions on that index, as done in many previous studies, can lead to wrong conclusions. This paper argues that ENSO is best viewed not as a number but as an evolving dynamical process for this purpose.

     In this graph, I did not use the ENSO index in a regression analysis. I simply averaged the NINO3.4 sea surface temperature anomalies. There’s a very basic difference, Tom.

     You also failed to recognize the importance of what I wrote after I quoted Compo and Sardeshmuk. It was:

     Compo and Sardeshmukh have not accounted for the left over warm water associated with major El Niño events, like the 1986/87/88 and 1997/98 El Niños. In time, maybe they will.

     Compo and Sardeshmuk are a step in the right direction, as I noted before I quoted them in my comment at 137.

     Tom Curtis at 146 says: “Therefore simply taking the SST anomaly in the Nino 3.4 region over vastly different time periods cannot plausibly be considered a measure of ENSO activity. It in no way allows for the effects of other factors which we know will cause changes in the SST in the Nino 3.4 region as much as anywhere else.”

     What “other factors” are referring to, Tom? Are you aware the authors of peer-reviewed papers about ENSO consider NINO3.4 sea surface temperature anomalies to be “a measure of ENSO activity”? An example is the Giese et al (2009) paper The 1918/19 El Niño. They argued that the 1918/19 portion of the 1918/19/20 El Niño was underestimated in the NINO3.4 sea surface temperature reconstructions, and that it was likely comparable in strength to the 1982/83 and 1997/98 El Niño events. Giese et al (2009) also suggested that the 1912/13 and 1939/40/41/42 El Niño events were also under-rated. If as you say, the NINO3.4 data “cannot plausibly be considered a measure of ENSO activity”, why would Giese et al (2009) have even written their paper, Tom? Why would NOAA use NINO3.4 data as the basis for its Ocean NINO Index? Why would Kevin Trenberth recommend NINO3.4 data over the SOI? Read on.

     Please also note the dataset used by the ENSO experts who wrote that paper, Tom. It was HADISST, the same one I used. It was not HADSST3 or HADSST2, which you used in your comment 151.

     With respect to your preference to the Southern Oscillation Index data, are you aware of all of the difficulties with that dataset? Refer to Trenberth (1997) The Definition of El Niño. He writes (my boldface):

     Various versions of the SOI exist although, in recent years, most deal only with atmospheric pressures and usually only those at Darwin and Tahiti. In using the SOI based on just two stations, it must be recognized that there are many small scale and high frequency phenomena in the atmosphere, such as the Madden-Julian Oscillation, that can influence the pressures at stations involved in forming the SOI, but which do not reflect the Southern Oscillation itself. Accordingly, the SOI should only be used when monthly means are appropriately smoothed (Trenberth 1984, Trenberth and Hoar 1996a). For many years, Tahiti data were available only after 1935. Ropelewski and Jones (1987) outline an extension of the SOI prior to then using newly discovered Tahiti data, and they also discuss different ways of standardizing the data for use in the SOI. However, there are questions about the integrity of the Tahiti data prior to 1935 (Trenberth and Hoar 1996a), as the Tahiti-Darwin correlation is much lower in spite of strong evidence that the SO was present from other stations, and the noise level and variance in the early Tahiti data is higher than in the more recent period. Although Ropelewski and Jones (1987) state that they believe that months with data contained a full complement of daily values, several monthly means are missing and the monthly means that are present for the earlier period are more consistent with the view that they originate from an incomplete dataset in which values contributing to monthly means are missing. There are also questions about whether the diurnal cycle in surface pressure, which contains a strong semidiurnal tide component, has been adequately dealt with in forming the means. Ropelewski and Jones (1987) use a 5 month running mean to define their indices (as is done in NOAA's Climate Diagnostics Bulletin).

     It took a couple of years after that paper, but in response to those deficiencies in the SOI, Tom, NOAA and many scientists adopted NINO3.4 sea surface temperature anomalies and the Cold Tongue Index (which is based on sea surface temperature anomalies of a similar region) as the better measure of the frequency, magnitude and duration of El Niño and La Niña events.

     Further to your comment 151: One of the reasons Giese et al used HADISST is because HADSST2 or HADSST3 are spatially incomplete. That is, there’s missing data. This can be seen in the output from the KNMI Climate Explorer of NINO3.4 data for the periods of 1912 through 1944 and 1944 through 1976, using a 15% cutoff as you did:

     HADSST2 NINO3.4 from 1912-1944

     HADSST2 NINO3.4 from 1944-1976

     HADSST3 NINO3.4 from 1912-1944

     HADSST3 NINO3.4 from 1944-1976

     Note how much data is missing. I’ve limited the graphs to 3+ decade periods so that you could see all of the missing data, unlike your graphs in comment 151. That lack of data appears to have skewed the results you presented in comment 151, Tom.

     If you had asked me what dataset I used in this graph, Tom, I would have told you HADISST. It’s been infilled so there’s no missing data. Here are the HADISST-based graphs from the KNMI Climate Explorer of NINO3.4 SST anomalies so you can visually compare them to the HADSST2- and HADSST3-based graphs above:

     HADISST NINO3.4 from 1912-1944

     HADISST NINO3.4 from 1944-1976

     The differences between the HADISST-based NINO3.4 data and the others are quite extraordinary. Also if you had asked, I would have explained why I used HADISST—the HADSST2 and HADSST3 datasets are spatially incomplete in the NINO3.4 region. And I would have told you I used the base years of 1950-1979 in agreement with Trenberth (1997) The Definition of El Niño. He stated that 1950 to 1979 was the best base period for NINO3.4 sea surface temperature anomalies. Trenberth writes:

     Figure 1 shows the five month running mean SST time series for the Niño 3 and 3.4 regions relative to a base period climatology of 1950-1979 given in Table 1. The base period can make a difference. This standard 30 year base period is chosen as it is representative of the record this century, whereas the period after 1979 has been biased warm and dominated by El Niño events (Trenberth and Hoar 1996a). Mean temperatures are higher in the Niño 3.4 region than in Niño 3 and its proximity to the Pacific warm pool and main centers of convection is the reason for the physical importance of Niño 3.4.

     If you had asked, Tom, I would have been happy to explain all those things. You could have saved yourself a lot of time, and you would not have jumped to all the wrong conclusions.

170. Bob Tisdale at 20:00 PM on 7 December, 2012

     Michael Sweet says at 147: “Your claim here that Gavin Schmidt at Realclimate says you can compare a single realization (what happened) to the model average and not to the model envelope is simply untrue.”

     I don’t believe I claimed there “that Gavin Schmidt at Realclimate says you can compare a single realization (what happened) to the model average and not to the model envelope…”. I quoted Gavin Schmidt’s reply to the RealClimate comment at 30 Sep 2009 at 6:18 AM. It was:

     [Response: Any single realisation can be thought of as being made up of two components - a forced signal and a random realisation of the internal variability ('noise'). By definition the random component will uncorrelated across different realisations and when you average together many examples you get the forced component (i.e. the ensemble mean). Just as in weather forecasting, the forecast with the greatest skill turns out to be this ensemble mean. i.e. you do the least badly by not trying to forecast the 'noise'. This isn't necessarily cast in stone, but it is a good rule of thumb. - gavin]

     Gavin said what he said. He was not talking about model-data comparisons; he was talking about model outputs in general, as was the quote I provided you from NCAR. NCAR wrote:

     Unless you are interested in a particular ensemble member where the initial conditions make a difference in your work, averaging of several ensemble members will give you best representation of a scenario.

     I’m NOT interested in the model noise, Michael; I’m also NOT interested in a “particular ensemble member where the initial conditions make a difference in your work”; I’m interested in the forced component that’s presented by the multi-model ensemble mean of all of the CMIP3 or CMIP5 climate models in my model-data comparisons, because it gives you “the best representation of a scenario”.

     Note: The NCAR webpage I linked earlier seems to be offline, but the Wayback Machine has captured the text.

171. Bob Tisdale at 20:01 PM on 7 December, 2012
     A closing note: Thanks to all of you for your feedback. It does not appear we’re going anywhere with this discussion. That is, it’s quite obvious we are not going to agree about anything. Therefore, it’s likely I won’t be back on this thread. And thank you, moderators. Enjoy your holidays.
172. JasonB at 20:44 PM on 7 December, 2012

     A closing note: Thanks to all of you for your feedback. It does not appear we’re going anywhere with this discussion. That is, it’s quite obvious we are not going to agree about anything. Therefore, it’s likely I won’t be back on this thread

     That's one strategy to studiously avoid the meat of the criticism and instead hope that everyone will be overwhelmed by the spam, I suppose. In that veritable Gish Gallop of posts the only attempt to respond to Kevin C's devastating results was to insist they'd already be addressed by an earlier comment (#167). How disappointing. I suppose this stands as a warning to "skeptics" of the dangers of egging on SkS regulars to stop ignoring fringe theories and instead take a close look at them. You're better off staying below the radar.
173. Doug Bostrom at 02:42 AM on 8 December, 2012
     If nothing else is accomplished Bob at least lends new credence to the concept of "catastrophism."
174. Philippe Chantreau at 03:11 AM on 8 December, 2012
     Bob Tisdale certainly makes a tremendous effort to have the last word, or the last couple of thousand words. I'm hoping that it will be examined closely, although I confess I do not have the leisure of doing so. I did notice this: "If the Rest-of-the-World sea surface temperature data cooled proportionally during the La Niña events of 1988/89 and 1998-2001..." The fact that they did not could certainly point to the existence of a warming forcing...
175. KR at 04:19 AM on 8 December, 2012
     The most recent replies from Tisdale are extremely verbose - and avoid addressing criticisms. Some notes from my point of view (not exhaustive, mind you, but if you find the first 10% of a Gish Gallop to be nonsense, you can reasonably expect the same quality in the rest): * Compo and Sardeshmukh 2009: Oceanic influences on recent continental warming - They found that running the models with oceanic warming (>93% of the energy from the GH effect) but without continental GHG warming (2.1% continental, 2.3% atmospheric) produced nearly identical model results to using full forcings. Tisdale claims this in some fashion invalidates ocean warming - but if you have only subtracted 5% of the forcing, leaving in 95% of the GHG effect, such results are entirely unsurprising. A nonsense claim from Tisdale. * And what of the more than sufficient warming from an enhanced greenhouse effect? And as I replied to KR at 134: Downward longwave radiation appears to do nothing more cause a little more evaporation from the ocean surface, which makes perfect sense since it only penetrates the top few millimeters. This statement alone represents pure denial on Tisdale's part. He has been pointed to the physics and to the observations confirming oceanic warming from the enhanced greenhouse effect - his response here is nothing more than a repeated, armwaving denial of the physics and the data. * Models. Kevin C has taken the time to examine Tisdale's model (single lag, no volcanic term - which, I will point out, Tisdale has never clearly explained), confirmed by replicating Tisdale's results. He has in addition examined the behavior of a more physically realistic two-lag model with volcanic forcing. When tested against data outside the training set, Tisdale's model fails completely. Kevin C's more realistic model replicates the data both within and outside the training set. Conclusions drawn from Tisdale's model are therefore unsupported. Tisdale's reply consists of just repeating his poor model results. That is no reply at all - Tisdale's model is a failure. * Tisdale claims that "global oceans did not respond proportionately" to ENSO events, but argues only from Sea Surface Temperatures (SST's). Not ocean heat content (OHC), but just surface temperatures. The energy involved in the ENSO is considerable in comparison to atmospheric energies, but not compared to the ocean mass itself. SST is by no means a sufficient measure of the global oceans - rather a noisy subset thereof. If you actually examine the global oceanic response you do indeed see a proportional response to forcing over the period of observations, with small variations occurring related to (among other things) the ENSO. Again, a nonsense claim from Tisdale, looking at the wrong data set relative to his hypothesis. --- Finally, after a Gish Gallop of (non)replies, Tisdale exits the thread. Leaving (IMO) questions unanswered, criticisms unaddressed, and multiple examples of denial of the physics and the evidence. At this point, after considering Tisdale's arguments, I would consider the myth of the ENSO driving global warming to be busted.

Prev  1  2  3  4  5  6  7  8  Next

Post a Comment

Political, off-topic or ad hominem comments will be deleted. Comments Policy...

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.