the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Apr 2024
A global Data Assimilation of Moisture Patterns from 21,000–0 BP (DAMP-21ka) using lake level proxy records
Abstract. Global hydroclimate significantly differed from modern climate during the mid-Holocene (6 ka) and Last Glacial Maximum (21 ka). Consequently, both periods have been described as either a partial or reverse analogue for current climate change. To reconstruct past hydroclimate, an offline paleoclimate data assimilation methodology is applied to a dataset of 130 lake status records which provide relative estimates of water level measured using percentile units. The proxy observations are integrated with the climate dynamics of two transient simulations (TraCE-21ka and HadCM3) using a multivariate proxy system model (PSM) which estimates relative lake status from available climate simulation variables. The resulting DAMP-21ka (Data Assimilation of Moisture Patterns 21,000–0 BP) reanalysis reconstructs annual lake status and precipitation values at 500-year resolution and represents the first application of the methodology to global hydroclimate on timescales spanning the Holocene and longer. Validation using Pearson’s correlation coefficients indicates that the reconstruction (0.33) is more skillful, on average, than model simulations (0.10), particularly in portions of North America and East Africa where data density is high and proxy-model disagreement is prominent during the Holocene. Results of the PSM and assimilation are used to evaluate climatic controls on lake status, spatiotemporal patterns of moisture variability, and proxy-model disagreement. During the mid-Holocene, wetter conditions are reconstructed for North and East Africa, Asia, and southern Australia, but, in contrast to the model prior, negative anomalies are observed in North America resulting in drier than modern conditions throughout the Northern Hemisphere midlatitudes. Proxy-model disagreement in western North America may reflect a bias in model simulations to stronger sea level pressure gradients in the North Pacific during the mid-Holocene. The data assimilation framework is able to reconcile these differences by integrating the constraints of proxy observations with the dynamics of the model prior to produce a more robust estimation of hydroclimate variability during the past 21,000 years.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(2154 KB)
-
Supplement
(2191 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2154 KB) - Metadata XML
-
Supplement
(2191 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-746', Anonymous Referee #1, 10 May 2024
The authors present a novel reconstruction of global hydroclimate over the Holocene by integrating 130 lake records with two transient climate model simulations through a paleoclimate data assimilation technique. This reconstruction offers significant value to the scientific community, particularly for evaluating the growing number of transient simulations. Although I defer to other reviewers for their expertise on lake records, I found the manuscript clear and the analysis robust. The authors examined the sensitivity of their reconstruction by using two different climate models in their data assimilation method and evaluating the selected records in the final reconstructions. I appreciate their approach of incorporating lake records through a somehow process-based proxy system model rather than solely relying on statistical methods such as simple linear regression.
My primary concern is the notable discrepancy between proxy records and transient model simulations in North America and North Africa. As the authors noted, while proxy records suggest divergent trends in these regions, the climate models indicate a common forced response. Since data assimilation depends on modelled covariance to distribute local information from assimilated variables across space and other variables, I question the effectiveness of these models for reconstructing global hydroclimate over the Holocene. The authors attempt to address this issue by applying a localization radius, but this adjustment may limit some of the advantages of data assimilation in understanding hydroclimate variability.
Regarding the mechanisms behind these trends, the authors mainly focus on the Northern Hemisphere, particularly North America. They also mention wetter conditions in southeastern Australia during the mid-Holocene. Many records in mid- and high-latitude regions often reflect changes in wind direction and intensity. The authors could enhance their analysis by exploring the potential drivers of the wetting trend. Specifically, do the authors identify a particular influence of atmospheric circulation variability in this wetting trend?
Citation: https://doi.org/10.5194/egusphere-2024-746-RC1 -
AC1: 'Reply on RC1', Christopher Hancock, 16 Jul 2024
Thank you for your detailed comments. The responses to each comment are provided below.
Comment: My primary concern is the notable discrepancy between proxy records and transient model simulations in North America and North Africa. As the authors noted, while proxy records suggest divergent trends in these regions, the climate models indicate a common forced response. Since data assimilation depends on modelled covariance to distribute local information from assimilated variables across space and other variables, I question the effectiveness of these models for reconstructing global hydroclimate over the Holocene. The authors attempt to address this issue by applying a localization radius, but this adjustment may limit some of the advantages of data assimilation in understanding hydroclimate variability.
Response: As noted in the text, we agree with this concern. However we argue that the effectiveness of models to capture covariance relationships is scale dependent so that global climate relationships are more difficult to accurately simulate than regional ones. This is likely more true of precipitation than temperature. We highlight the advantage of using data assimilation with a multi-model prior which provides more confidence when the two independent simulations agree (e.g., Line 384). 0—Additionally, the use of covariance localization is a common feature in data assimilation (e.g., Osman et al., 2021) to ensure that local proxies are favored over remote proxies when the modeled covariance relationships may be imperfect. Data assimilation also has an advantage over proxy-only reconstructions as it provides a method for considering site specific relationships between lake status and precipitation so that in regions where the two variables covary strongly, the model prior is updated by a greater amount.
Comment: Regarding the mechanisms behind these trends, the authors mainly focus on the Northern Hemisphere, particularly North America. They also mention wetter conditions in southeastern Australia during the mid-Holocene. Many records in mid- and high-latitude regions often reflect changes in wind direction and intensity. The authors could enhance their analysis by exploring the potential drivers of the wetting trend. Specifically, do the authors identify a particular influence of atmospheric circulation variability in this wetting trend?
Response: We focus our discussion for this topic on western North America because it remains an important topic of open discussion in the community and there is substantial independent data for comparison. Sect. 3.7 explores this topic and we invoke SLP anomalies in the North Pacific and resulting changes to atmospheric circulation strength as a likely source of the wetting trend (Fig 10). For Australia we are limited in the available independent proxies to validate results for this region. However, we note that the pattern of the wetting in southern Australia (Fig 7d-f) is potentially consistent with a northward position of the southern westerly winds. This could represent the extratropical response to a northward shift in the position of the ITCZ (Cheng et al., 2012; van der Bilt et al., 2022). We did not consider these results robust enough to include in the manuscript. Future research to integrate isotope proxies from New Zealand with currently unavailable isotope enabled simulations would provide more detailed insight into the dynamics of this region.
Citation: https://doi.org/10.5194/egusphere-2024-746-AC1 -
AC3: 'Reply on RC1', Christopher Hancock, 16 Jul 2024
Thank you for your detailed comments. The responses to each comment are provided below.
Comment: My primary concern is the notable discrepancy between proxy records and transient model simulations in North America and North Africa. As the authors noted, while proxy records suggest divergent trends in these regions, the climate models indicate a common forced response. Since data assimilation depends on modelled covariance to distribute local information from assimilated variables across space and other variables, I question the effectiveness of these models for reconstructing global hydroclimate over the Holocene. The authors attempt to address this issue by applying a localization radius, but this adjustment may limit some of the advantages of data assimilation in understanding hydroclimate variability.Response: As noted in the text, we agree with this concern. However we argue that the effectiveness of models to capture covariance relationships is scale dependent so that global climate relationships are more difficult to accurately simulate than regional ones. This is likely more true of precipitation than temperature. We highlight the advantage of using data assimilation with a multi-model prior which provides more confidence when the two independent simulations agree (e.g., Line 384). 0—Additionally, the use of covariance localization is a common feature in data assimilation (e.g., Osman et al., 2021) to ensure that local proxies are favored over remote proxies when the modeled covariance relationships may be imperfect. Data assimilation also has an advantage over proxy-only reconstructions as it provides a method for considering site specific relationships between lake status and precipitation so that in regions where the two variables covary strongly, the model prior is updated by a greater amount.
Comment: Regarding the mechanisms behind these trends, the authors mainly focus on the Northern Hemisphere, particularly North America. They also mention wetter conditions in southeastern Australia during the mid-Holocene. Many records in mid- and high-latitude regions often reflect changes in wind direction and intensity. The authors could enhance their analysis by exploring the potential drivers of the wetting trend. Specifically, do the authors identify a particular influence of atmospheric circulation variability in this wetting trend?Response: We focus our discussion for this topic on western North America because it remains an important topic of open discussion in the community and there is substantial independent data for comparison. Sect. 3.7 explores this topic and we invoke SLP anomalies in the North Pacific and resulting changes to atmospheric circulation strength as a likely source of the wetting trend (Fig 10). For Australia we are limited in the available independent proxies to validate results for this region. However, we note that the pattern of the wetting in southern Australia (Fig 7d-f) is potentially consistent with a northward position of the southern westerly winds. This could represent the extratropical response to a northward shift in the position of the ITCZ (Cheng et al., 2012; van der Bilt et al., 2022). We did not consider these results robust enough to include in the manuscript. Future research to integrate isotope proxies from New Zealand with currently unavailable isotope enabled simulations would provide more detailed insight into the dynamics of this region.
Referencesvan der Bilt, W.G.M., W.J., D’Andrea, L.T., Oppedal, Bakke, J., Bjune, A. E., Zwier, M.: Stable Southern Hemisphere westerly winds throughout the Holocene until intensification in the last two millennia, Commun Earth Environ, 3, 186, https://doi.org/10.1038/s43247-022-00512-8, 2022.
Cheng, H., Sinha, A., Wang, X., Cruz, F. W., and Edwards, R. L.: The global paleomonsoon as seen through speleothem records from Asia and the Americas, Clim. Dyn., 39, 1045–1062, https://doi.org/10.1007/s00382-012-1363-7, 2012.
Osman, M.B., Tierney, J. E., Zhu, J., Tardif, R., Hakim, G. J., King, J., C.J. Poulsen: Globally resolved surface temperatures since the Last Glacial Maximum. Nature, 599, 239–244, https://doi.org/10.1038/s41586-021-03984-4, 2021
Citation: https://doi.org/10.5194/egusphere-2024-746-AC3
-
AC1: 'Reply on RC1', Christopher Hancock, 16 Jul 2024
-
RC2: 'Comment on egusphere-2024-746', Anonymous Referee #2, 17 Jun 2024
This study provides a new global data assimilation of paleo-hydroclimate change since the LGM by integrating lake status records and transient simulations. The complexity of hydroclimate, both in temporal and spatial scales, surpasses that of temperature change. Consequently, the reconstruction of long-term global hydroclimate is a rare endeavor, making the results from this study a significant advancement in this field. However, I have some concerns on the data used for the assimilation.
- The lake status data are mainly from the Oxford Lake Status (OLS) Databank (n = 98; Street-Perrott et al., 1989), which is really old. Why not the newer one, the Global Lake Status Data Base (GLSDB: Kohfeld and Harrison, 2000, Harrison et al., 2003)( https://pmip2.lsce.ipsl.fr/synth/lakestatus.shtml)? Even though, it is still a little old and there's some controversy on the chronology and proxy used. In Africa and Australia, new compilations have been completed (Cort et al., 2021 QSR; Clerke, 2022 Hydrological Regime of Australian Lakes Over the Late-Quaternary and Holocene). And in Americas and Asia, I’m sure there are more continuous lake level records (Lowry and Morrill, 2019; Li and Zhang, 2020).
- Are TraCE-21k and HadCM3 the only available transient simulations? What is the difference between these two models, such as the temporal resolution and reconstruction skill. I think they should be compared before combined.
Some minor suggestions as follows.
The conclusion is too long.
Line 86: please add “(details in section 2.4)”.
In figure 2(b), What is the purpose of removing the Holocene mean value? And how do you deal with the negative values?
Line 305: “then” should be “than”.
In section 3.2, as Pearson’s correlation coefficient is really not high, the significance level should be mentioned.
In equation S1, no Xa.
Citation: https://doi.org/10.5194/egusphere-2024-746-RC2 -
AC2: 'Reply on RC2', Christopher Hancock, 16 Jul 2024
Thank you for these detailed comments. The responses to each comment are provided below.
Comment: The lake status data are mainly from the Oxford Lake Status (OLS) Databank (n = 98; Street-Perrott et al., 1989), which is really old. Why not the newer one, the Global Lake Status Data Base (GLSDB: Kohfeld and Harrison, 2000, Harrison et al., 2003)( https://pmip2.lsce.ipsl.fr/synth/lakestatus.shtml)? Even though, it is still a little old and there's some controversy on the chronology and proxy used. In Africa and Australia, new compilations have been completed (Cort et al., 2021 QSR; Clerke, 2022 Hydrological Regime of Australian Lakes Over the Late-Quaternary and Holocene). And in Americas and Asia, I’m sure there are more continuous lake level records (Lowry and Morrill, 2019; Li and Zhang, 2020).
Response: Thank you for your detailed list of relevant data sources. Some of these were previously considered but not included for a variety of reasons. One major constraint is that many of these papers (Harrison et al., 2003; Lowry and Morrill, 2019; Li and Zhang, 2019) focus on only the LGM and/or mid-Holocene and therefore only publish lake status values for these narrow time periods. The PSM used in this study requires transient data to compare relative (percentile) changes through time, and the limited published data are therefore insufficient. We also note that using time slice simplifies the designation for “higher” or “lower” as it requires less precise age control; within a single time slice, values can span multiple millennia. Additionally, many of the referenced publications are primarily based on pollen or geochemical interpretations (e.g., Lu et al., 2015; Moreno and León 2002; Sun et at al., 2013; Vélez et al., 2003) which we exclude (Line 125), or the data do not have sufficient duration or resolution to meet our criteria for inclusion in the dataset (Line 131). We evaluate each of the listed references to determine if the publications provide relevant data for our study.
Kohfeld and Harrison, 2000 and Harrison et al., 2003 data available at https://pmip2.lsce.ipsl.fr/synth/lakestatus.shtml reports time slice lake status values for the mid-Holocene and LGM. Although the “core of [GLSDB] is the Oxford Lake-level Data Base” (Qin et al., 1998), the GLSDB (677) nearly doubles the number of lakes within the OLS (360 lakes) mainly by “improvements to the coverage temperate and wet tropical regions…achieved through the incorporation of information from four regional data bases” (Qin et al., 1998). We tracked down three regional databases and found 74 records which match our stated criteria from the former Soviet Union and Mongolia, (FSUDB: Tarasov et al., 1996; n=38); European (ELSDB: Yu and Harrison, 1995; n=16), and Chinese (Yu, 2001; n=20) lake-level data bases. We could not identify data for the African Lake Status Data Base (Jolly et al. 1998), but many of these data would likely be superseded by the new De Cort et al. (2021) compilation.
De Cort et al. (2021) was overlooked because the composite records cannot be used in the data assimilation framework. However, by looking at the original data, we identify 19 sites in eastern and southern Africa which meet our criteria.
Clerke et al. (2022) was not published at the time of data collection. Thank you for bringing it to our attention. There appear to be 5 lakes which fit our criteria.
Lowry and Morrill (2019) and Li and Zhang (2020) report time slice lake status values for the LGM, but we searched the publications for each of the 44 and 37 sites respectively to identify records which we may have overlooked. We found 1 site (Uyuni Basin, Argentina) which we will add to the database.
In total, we identified 99 new records that we will add to the dataset, primarily from Eurasia. Some of these replace previously included data, so the updated dataset will include 216 total sites. We have updated the assimilation and created a new reconstruction, and will revise the text accordingly. Upon initial evaluation of the results, we do not find any major changes that affect our conclusions, although the expanded data coverage allows more insight into some regions, especially Asia, which we will add to the manuscript.
Comment: Are TraCE-21k and HadCM3 the only available transient simulations? What is the difference between these two models, such as the temporal resolution and reconstruction skill. I think they should be compared before combined.
Response: Transient GCM simulations for our time-scale of interest are rare, and we are unaware of additional simulations run with different models.Precipitation values for the two simulations are compared by Hancock et al. (2023) which found that reconstruction skill, as assessed by agreement with proxy values, varies regionally. Results assimilating a model prior sampled from only one model would more closely resemble the model values shown in Fig. S6. Data for both models is available at annual temporal resolution, but we use multidecadal (50 year) mean values for the model prior (see line 235).
Comment: The conclusion is too long.
Response: Agreed. We will make the conclusion more concise.
Comment: Line 86: please add “(details in section 2.4)”.
Response: We will include this reference
Comment: In figure 2(b), What is the purpose of removing the Holocene mean value? And how do you deal with the negative values?
Response: Fig 2 shows the preprocessing steps applied to the proxy values. For data assimilation, values are often converted to anomalies prior to assimilation (e.g. Erb et al., 2022), and we found that our multi-model prior necessitated this step. The lake status percentiles were assigned between 0-100 for each model, but the precipitation values do not have a bias correction. We found significant mean offsets between the TraCE-21ka and HadCM3 which impacted the covariance structure in the assimilation. Therefore, it was necessary to convert the data to anomalies to accurately reconstruct precipitation values. Negative values are not an issue at this step because both the proxy values and the model prior are both converted to anomalies. The difference between these is the innovation which may be positive or negative even without converting to anomalies.
Comment: Line 305: “then” should be “than”.
Response: We will fix this grammatical error.
Comment: In section 3.2, as Pearson’s correlation coefficient is really not high, the significance level should be mentioned.
Response: We will revise the text to mention the median significance level (0.11 for both the prior and reconstruction). We will also mention the percentage of records significantly correlated to the reconstruction (62% of records with a p-value < 0.05). Note that these values may differ in the revised text once additional lake records are included in the analysis.
Comment: In equation S1, no Xa.
Response: We will fix this error.
References
Clerke, L.: Hydrological regime of Australian lakes over the Late-Quaternary and Holocene, Doctoral dissertation, Macquarie University, Sydney, Australia, https://doi.org/10.25949/22662253.v1, 2023.
De Cort, G., Chevalier, M., Burrough, S. L., Chen, C. Y., Harrison, S. P.: An uncertainty-focused database approach to extract spatiotemporal trends from qualitative and discontinuous lake-status histories. Quat. Sci. Rev., 258, 106870, https://doi.org/10.1016/j.quascirev.2021.106870, 2021.
Erb, M. P., McKay, N. P., Steiger, N., Dee, S., Hancock, C., Ivanovic, R. F., Gregoire, L. J., and Valdes, P.: Reconstructing Holocene temperatures in time and space using paleoclimate data assimilation, Clim. Past, 18, 2599–2629, https://doi.org/10.5194/cp-18-2599-2022, 2022.
Hancock, C. L., McKay, N. P., Erb, M. P., Kaufman, D. S., Routson, C. R., Ivanovic, R. F., Gregoire, L. J., and Valdes, P.: Global synthesis of regional Holocene hydroclimate variability using proxy and model data, Paleoceanogr. Paleoclimatology, 38, e2022PA004597, https://doi.org/10.1029/2022PA004597, 2023.
Harrison, S. P., Kutzbach, J. E., Liu, Z., Bartlein, P. J., Otto-Bliesner, B., Muhs, D., Prentice, I. C., and Thompson, R. S.: Mid-Holocene climates of the Americas: a dynamical response to changed seasonality. Clim. Dyn., 20, 663-688, doi:10.1007/s00382-002-0300-6, 2003.
Jolly, D., Harrison, S. P., Damnati, B., Bonnefille, R.: Simulated climate and biomes of Africa during the late quaternary: comparison with pollen and lake status data, Quaternary Science Reviews, 17(6-7), 629-657, https://doi.org/10.1016/S0277-3791(98)00015-8, 1998.
Kohfeld, K. E., and Harrison, S. P.: How well can we simulate past climates? Evaluating the models using global palaeoenvironmental datasets, Quaternary Science Reviews 19(1-5), 321-346, 2000.
Li, Y. and Zhang, Y.: Synergy of the westerly winds and monsoons in the lake evolution of global closed basins since the Last Glacial Maximum and implications for hydrological change in central Asia, Clim. Past, 16, 2239–2254, https://doi.org/10.5194/cp-16-2239-2020, 2020.
Lowry, D. P. and Morrill, C.: Is the Last Glacial Maximum a reverse analog for future hydroclimate changes in the Americas?, Clim. Dyn., 52, 4407–4427, https://doi.org/10.1007/s00382-018-4385-y, 2019.
Lu, Y. B., An, C. B., Zhao, J. J.: An isotopic study on water sys- tem of Lake Barkol and its implication for Holocene climate dynamics in arid central Asia, Environ. Earth Sci., 73, 1377–1383, https://doi.org/10.1007/s12665-014-3492-2, 2015.
Moreno, P. I. and León, A. L.: Abrupt vegetation changes during the last glacial to Holocene transition in mid-latitude South America, J. Quaternary Sci., 18, 787–800, https://doi.org/10.1002/jqs.801, 2003.
Qin, B. J., Harrison S. P., Kutzback, J. E.: Evaluation of modelled regional water balance using lake status data: a comparison of 6ka simulations with the NCAR CCM, Quat. Sci. Rev., 17, 535-548, https://doi.org/10.1016/S0277-3791(98)00011-0, 1998.
Sun, A. Z., Feng, Z. D., Ran, M., and Zhang, C. J.: Pollen-recorded bioclimatic variations of the last ∼ 22,600 years retrieved from Achit Nuur core in the western Mongolian Plateau, Quatern. Int., 311, 36–43, https://doi.org/10.1016/j.quaint.2013.07.002, 2013.
Vélez M.I., Hooghiemstra H., Metcalfe S., Martínez I., Mommersteeg H: Pollen and diatom based environmental history since the Last Glacial Maximum from the Andean core Fúquene-7, Colombia. J Quat Sci 18:17–30, https://doi.org/10.1002/jqs.730, 2003.
Tarasov, P.E., Harrison, S.P., Saarse, L., Pushenko, M.Ya., Andreev, A.A., Aleshinskaya, Z.V., Davydova, N.N., Dorofeyuk, N.I., Efremov, Yu.V., Khomutova, V.I., Sevastyanov, D.V., Tamosaitis, J., Uspenskaya, O.N., Yakushko, O.F., Tarasova, I.V, Ya, M., Elina, G.A., Elovicheva, Ya.K., Filimonova, L.V., Gunova, V.S., Kvavadze, E.V., Nuestrueva, I.Yu., Pisareva, V.V., Shelekhova, T.S., Subetto, D.A., Zernitskaya, V.P.: Lake status records from the Former Soviet Union and Mongolia: documentation of the second version of the database, Paleoclimatology Publications Series Report #5, https://doi.org/10.25921/49ag-ck94, 1996
Yu, G.: Lake status records from Europe: data base documentation, Publications Series Report #3, https://doi.org/10.25921/p5aq-0931, 1995.
Yu, G. Harrison, S.P. Xue, B.: Lake status records from China: data base documentation, MPI-BGC Tech Rep 4, 2001.
Citation: https://doi.org/10.5194/egusphere-2024-746-AC2
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-746', Anonymous Referee #1, 10 May 2024
The authors present a novel reconstruction of global hydroclimate over the Holocene by integrating 130 lake records with two transient climate model simulations through a paleoclimate data assimilation technique. This reconstruction offers significant value to the scientific community, particularly for evaluating the growing number of transient simulations. Although I defer to other reviewers for their expertise on lake records, I found the manuscript clear and the analysis robust. The authors examined the sensitivity of their reconstruction by using two different climate models in their data assimilation method and evaluating the selected records in the final reconstructions. I appreciate their approach of incorporating lake records through a somehow process-based proxy system model rather than solely relying on statistical methods such as simple linear regression.
My primary concern is the notable discrepancy between proxy records and transient model simulations in North America and North Africa. As the authors noted, while proxy records suggest divergent trends in these regions, the climate models indicate a common forced response. Since data assimilation depends on modelled covariance to distribute local information from assimilated variables across space and other variables, I question the effectiveness of these models for reconstructing global hydroclimate over the Holocene. The authors attempt to address this issue by applying a localization radius, but this adjustment may limit some of the advantages of data assimilation in understanding hydroclimate variability.
Regarding the mechanisms behind these trends, the authors mainly focus on the Northern Hemisphere, particularly North America. They also mention wetter conditions in southeastern Australia during the mid-Holocene. Many records in mid- and high-latitude regions often reflect changes in wind direction and intensity. The authors could enhance their analysis by exploring the potential drivers of the wetting trend. Specifically, do the authors identify a particular influence of atmospheric circulation variability in this wetting trend?
Citation: https://doi.org/10.5194/egusphere-2024-746-RC1 -
AC1: 'Reply on RC1', Christopher Hancock, 16 Jul 2024
Thank you for your detailed comments. The responses to each comment are provided below.
Comment: My primary concern is the notable discrepancy between proxy records and transient model simulations in North America and North Africa. As the authors noted, while proxy records suggest divergent trends in these regions, the climate models indicate a common forced response. Since data assimilation depends on modelled covariance to distribute local information from assimilated variables across space and other variables, I question the effectiveness of these models for reconstructing global hydroclimate over the Holocene. The authors attempt to address this issue by applying a localization radius, but this adjustment may limit some of the advantages of data assimilation in understanding hydroclimate variability.
Response: As noted in the text, we agree with this concern. However we argue that the effectiveness of models to capture covariance relationships is scale dependent so that global climate relationships are more difficult to accurately simulate than regional ones. This is likely more true of precipitation than temperature. We highlight the advantage of using data assimilation with a multi-model prior which provides more confidence when the two independent simulations agree (e.g., Line 384). 0—Additionally, the use of covariance localization is a common feature in data assimilation (e.g., Osman et al., 2021) to ensure that local proxies are favored over remote proxies when the modeled covariance relationships may be imperfect. Data assimilation also has an advantage over proxy-only reconstructions as it provides a method for considering site specific relationships between lake status and precipitation so that in regions where the two variables covary strongly, the model prior is updated by a greater amount.
Comment: Regarding the mechanisms behind these trends, the authors mainly focus on the Northern Hemisphere, particularly North America. They also mention wetter conditions in southeastern Australia during the mid-Holocene. Many records in mid- and high-latitude regions often reflect changes in wind direction and intensity. The authors could enhance their analysis by exploring the potential drivers of the wetting trend. Specifically, do the authors identify a particular influence of atmospheric circulation variability in this wetting trend?
Response: We focus our discussion for this topic on western North America because it remains an important topic of open discussion in the community and there is substantial independent data for comparison. Sect. 3.7 explores this topic and we invoke SLP anomalies in the North Pacific and resulting changes to atmospheric circulation strength as a likely source of the wetting trend (Fig 10). For Australia we are limited in the available independent proxies to validate results for this region. However, we note that the pattern of the wetting in southern Australia (Fig 7d-f) is potentially consistent with a northward position of the southern westerly winds. This could represent the extratropical response to a northward shift in the position of the ITCZ (Cheng et al., 2012; van der Bilt et al., 2022). We did not consider these results robust enough to include in the manuscript. Future research to integrate isotope proxies from New Zealand with currently unavailable isotope enabled simulations would provide more detailed insight into the dynamics of this region.
Citation: https://doi.org/10.5194/egusphere-2024-746-AC1 -
AC3: 'Reply on RC1', Christopher Hancock, 16 Jul 2024
Thank you for your detailed comments. The responses to each comment are provided below.
Comment: My primary concern is the notable discrepancy between proxy records and transient model simulations in North America and North Africa. As the authors noted, while proxy records suggest divergent trends in these regions, the climate models indicate a common forced response. Since data assimilation depends on modelled covariance to distribute local information from assimilated variables across space and other variables, I question the effectiveness of these models for reconstructing global hydroclimate over the Holocene. The authors attempt to address this issue by applying a localization radius, but this adjustment may limit some of the advantages of data assimilation in understanding hydroclimate variability.Response: As noted in the text, we agree with this concern. However we argue that the effectiveness of models to capture covariance relationships is scale dependent so that global climate relationships are more difficult to accurately simulate than regional ones. This is likely more true of precipitation than temperature. We highlight the advantage of using data assimilation with a multi-model prior which provides more confidence when the two independent simulations agree (e.g., Line 384). 0—Additionally, the use of covariance localization is a common feature in data assimilation (e.g., Osman et al., 2021) to ensure that local proxies are favored over remote proxies when the modeled covariance relationships may be imperfect. Data assimilation also has an advantage over proxy-only reconstructions as it provides a method for considering site specific relationships between lake status and precipitation so that in regions where the two variables covary strongly, the model prior is updated by a greater amount.
Comment: Regarding the mechanisms behind these trends, the authors mainly focus on the Northern Hemisphere, particularly North America. They also mention wetter conditions in southeastern Australia during the mid-Holocene. Many records in mid- and high-latitude regions often reflect changes in wind direction and intensity. The authors could enhance their analysis by exploring the potential drivers of the wetting trend. Specifically, do the authors identify a particular influence of atmospheric circulation variability in this wetting trend?Response: We focus our discussion for this topic on western North America because it remains an important topic of open discussion in the community and there is substantial independent data for comparison. Sect. 3.7 explores this topic and we invoke SLP anomalies in the North Pacific and resulting changes to atmospheric circulation strength as a likely source of the wetting trend (Fig 10). For Australia we are limited in the available independent proxies to validate results for this region. However, we note that the pattern of the wetting in southern Australia (Fig 7d-f) is potentially consistent with a northward position of the southern westerly winds. This could represent the extratropical response to a northward shift in the position of the ITCZ (Cheng et al., 2012; van der Bilt et al., 2022). We did not consider these results robust enough to include in the manuscript. Future research to integrate isotope proxies from New Zealand with currently unavailable isotope enabled simulations would provide more detailed insight into the dynamics of this region.
Referencesvan der Bilt, W.G.M., W.J., D’Andrea, L.T., Oppedal, Bakke, J., Bjune, A. E., Zwier, M.: Stable Southern Hemisphere westerly winds throughout the Holocene until intensification in the last two millennia, Commun Earth Environ, 3, 186, https://doi.org/10.1038/s43247-022-00512-8, 2022.
Cheng, H., Sinha, A., Wang, X., Cruz, F. W., and Edwards, R. L.: The global paleomonsoon as seen through speleothem records from Asia and the Americas, Clim. Dyn., 39, 1045–1062, https://doi.org/10.1007/s00382-012-1363-7, 2012.
Osman, M.B., Tierney, J. E., Zhu, J., Tardif, R., Hakim, G. J., King, J., C.J. Poulsen: Globally resolved surface temperatures since the Last Glacial Maximum. Nature, 599, 239–244, https://doi.org/10.1038/s41586-021-03984-4, 2021
Citation: https://doi.org/10.5194/egusphere-2024-746-AC3
-
AC1: 'Reply on RC1', Christopher Hancock, 16 Jul 2024
-
RC2: 'Comment on egusphere-2024-746', Anonymous Referee #2, 17 Jun 2024
This study provides a new global data assimilation of paleo-hydroclimate change since the LGM by integrating lake status records and transient simulations. The complexity of hydroclimate, both in temporal and spatial scales, surpasses that of temperature change. Consequently, the reconstruction of long-term global hydroclimate is a rare endeavor, making the results from this study a significant advancement in this field. However, I have some concerns on the data used for the assimilation.
- The lake status data are mainly from the Oxford Lake Status (OLS) Databank (n = 98; Street-Perrott et al., 1989), which is really old. Why not the newer one, the Global Lake Status Data Base (GLSDB: Kohfeld and Harrison, 2000, Harrison et al., 2003)( https://pmip2.lsce.ipsl.fr/synth/lakestatus.shtml)? Even though, it is still a little old and there's some controversy on the chronology and proxy used. In Africa and Australia, new compilations have been completed (Cort et al., 2021 QSR; Clerke, 2022 Hydrological Regime of Australian Lakes Over the Late-Quaternary and Holocene). And in Americas and Asia, I’m sure there are more continuous lake level records (Lowry and Morrill, 2019; Li and Zhang, 2020).
- Are TraCE-21k and HadCM3 the only available transient simulations? What is the difference between these two models, such as the temporal resolution and reconstruction skill. I think they should be compared before combined.
Some minor suggestions as follows.
The conclusion is too long.
Line 86: please add “(details in section 2.4)”.
In figure 2(b), What is the purpose of removing the Holocene mean value? And how do you deal with the negative values?
Line 305: “then” should be “than”.
In section 3.2, as Pearson’s correlation coefficient is really not high, the significance level should be mentioned.
In equation S1, no Xa.
Citation: https://doi.org/10.5194/egusphere-2024-746-RC2 -
AC2: 'Reply on RC2', Christopher Hancock, 16 Jul 2024
Thank you for these detailed comments. The responses to each comment are provided below.
Comment: The lake status data are mainly from the Oxford Lake Status (OLS) Databank (n = 98; Street-Perrott et al., 1989), which is really old. Why not the newer one, the Global Lake Status Data Base (GLSDB: Kohfeld and Harrison, 2000, Harrison et al., 2003)( https://pmip2.lsce.ipsl.fr/synth/lakestatus.shtml)? Even though, it is still a little old and there's some controversy on the chronology and proxy used. In Africa and Australia, new compilations have been completed (Cort et al., 2021 QSR; Clerke, 2022 Hydrological Regime of Australian Lakes Over the Late-Quaternary and Holocene). And in Americas and Asia, I’m sure there are more continuous lake level records (Lowry and Morrill, 2019; Li and Zhang, 2020).
Response: Thank you for your detailed list of relevant data sources. Some of these were previously considered but not included for a variety of reasons. One major constraint is that many of these papers (Harrison et al., 2003; Lowry and Morrill, 2019; Li and Zhang, 2019) focus on only the LGM and/or mid-Holocene and therefore only publish lake status values for these narrow time periods. The PSM used in this study requires transient data to compare relative (percentile) changes through time, and the limited published data are therefore insufficient. We also note that using time slice simplifies the designation for “higher” or “lower” as it requires less precise age control; within a single time slice, values can span multiple millennia. Additionally, many of the referenced publications are primarily based on pollen or geochemical interpretations (e.g., Lu et al., 2015; Moreno and León 2002; Sun et at al., 2013; Vélez et al., 2003) which we exclude (Line 125), or the data do not have sufficient duration or resolution to meet our criteria for inclusion in the dataset (Line 131). We evaluate each of the listed references to determine if the publications provide relevant data for our study.
Kohfeld and Harrison, 2000 and Harrison et al., 2003 data available at https://pmip2.lsce.ipsl.fr/synth/lakestatus.shtml reports time slice lake status values for the mid-Holocene and LGM. Although the “core of [GLSDB] is the Oxford Lake-level Data Base” (Qin et al., 1998), the GLSDB (677) nearly doubles the number of lakes within the OLS (360 lakes) mainly by “improvements to the coverage temperate and wet tropical regions…achieved through the incorporation of information from four regional data bases” (Qin et al., 1998). We tracked down three regional databases and found 74 records which match our stated criteria from the former Soviet Union and Mongolia, (FSUDB: Tarasov et al., 1996; n=38); European (ELSDB: Yu and Harrison, 1995; n=16), and Chinese (Yu, 2001; n=20) lake-level data bases. We could not identify data for the African Lake Status Data Base (Jolly et al. 1998), but many of these data would likely be superseded by the new De Cort et al. (2021) compilation.
De Cort et al. (2021) was overlooked because the composite records cannot be used in the data assimilation framework. However, by looking at the original data, we identify 19 sites in eastern and southern Africa which meet our criteria.
Clerke et al. (2022) was not published at the time of data collection. Thank you for bringing it to our attention. There appear to be 5 lakes which fit our criteria.
Lowry and Morrill (2019) and Li and Zhang (2020) report time slice lake status values for the LGM, but we searched the publications for each of the 44 and 37 sites respectively to identify records which we may have overlooked. We found 1 site (Uyuni Basin, Argentina) which we will add to the database.
In total, we identified 99 new records that we will add to the dataset, primarily from Eurasia. Some of these replace previously included data, so the updated dataset will include 216 total sites. We have updated the assimilation and created a new reconstruction, and will revise the text accordingly. Upon initial evaluation of the results, we do not find any major changes that affect our conclusions, although the expanded data coverage allows more insight into some regions, especially Asia, which we will add to the manuscript.
Comment: Are TraCE-21k and HadCM3 the only available transient simulations? What is the difference between these two models, such as the temporal resolution and reconstruction skill. I think they should be compared before combined.
Response: Transient GCM simulations for our time-scale of interest are rare, and we are unaware of additional simulations run with different models.Precipitation values for the two simulations are compared by Hancock et al. (2023) which found that reconstruction skill, as assessed by agreement with proxy values, varies regionally. Results assimilating a model prior sampled from only one model would more closely resemble the model values shown in Fig. S6. Data for both models is available at annual temporal resolution, but we use multidecadal (50 year) mean values for the model prior (see line 235).
Comment: The conclusion is too long.
Response: Agreed. We will make the conclusion more concise.
Comment: Line 86: please add “(details in section 2.4)”.
Response: We will include this reference
Comment: In figure 2(b), What is the purpose of removing the Holocene mean value? And how do you deal with the negative values?
Response: Fig 2 shows the preprocessing steps applied to the proxy values. For data assimilation, values are often converted to anomalies prior to assimilation (e.g. Erb et al., 2022), and we found that our multi-model prior necessitated this step. The lake status percentiles were assigned between 0-100 for each model, but the precipitation values do not have a bias correction. We found significant mean offsets between the TraCE-21ka and HadCM3 which impacted the covariance structure in the assimilation. Therefore, it was necessary to convert the data to anomalies to accurately reconstruct precipitation values. Negative values are not an issue at this step because both the proxy values and the model prior are both converted to anomalies. The difference between these is the innovation which may be positive or negative even without converting to anomalies.
Comment: Line 305: “then” should be “than”.
Response: We will fix this grammatical error.
Comment: In section 3.2, as Pearson’s correlation coefficient is really not high, the significance level should be mentioned.
Response: We will revise the text to mention the median significance level (0.11 for both the prior and reconstruction). We will also mention the percentage of records significantly correlated to the reconstruction (62% of records with a p-value < 0.05). Note that these values may differ in the revised text once additional lake records are included in the analysis.
Comment: In equation S1, no Xa.
Response: We will fix this error.
References
Clerke, L.: Hydrological regime of Australian lakes over the Late-Quaternary and Holocene, Doctoral dissertation, Macquarie University, Sydney, Australia, https://doi.org/10.25949/22662253.v1, 2023.
De Cort, G., Chevalier, M., Burrough, S. L., Chen, C. Y., Harrison, S. P.: An uncertainty-focused database approach to extract spatiotemporal trends from qualitative and discontinuous lake-status histories. Quat. Sci. Rev., 258, 106870, https://doi.org/10.1016/j.quascirev.2021.106870, 2021.
Erb, M. P., McKay, N. P., Steiger, N., Dee, S., Hancock, C., Ivanovic, R. F., Gregoire, L. J., and Valdes, P.: Reconstructing Holocene temperatures in time and space using paleoclimate data assimilation, Clim. Past, 18, 2599–2629, https://doi.org/10.5194/cp-18-2599-2022, 2022.
Hancock, C. L., McKay, N. P., Erb, M. P., Kaufman, D. S., Routson, C. R., Ivanovic, R. F., Gregoire, L. J., and Valdes, P.: Global synthesis of regional Holocene hydroclimate variability using proxy and model data, Paleoceanogr. Paleoclimatology, 38, e2022PA004597, https://doi.org/10.1029/2022PA004597, 2023.
Harrison, S. P., Kutzbach, J. E., Liu, Z., Bartlein, P. J., Otto-Bliesner, B., Muhs, D., Prentice, I. C., and Thompson, R. S.: Mid-Holocene climates of the Americas: a dynamical response to changed seasonality. Clim. Dyn., 20, 663-688, doi:10.1007/s00382-002-0300-6, 2003.
Jolly, D., Harrison, S. P., Damnati, B., Bonnefille, R.: Simulated climate and biomes of Africa during the late quaternary: comparison with pollen and lake status data, Quaternary Science Reviews, 17(6-7), 629-657, https://doi.org/10.1016/S0277-3791(98)00015-8, 1998.
Kohfeld, K. E., and Harrison, S. P.: How well can we simulate past climates? Evaluating the models using global palaeoenvironmental datasets, Quaternary Science Reviews 19(1-5), 321-346, 2000.
Li, Y. and Zhang, Y.: Synergy of the westerly winds and monsoons in the lake evolution of global closed basins since the Last Glacial Maximum and implications for hydrological change in central Asia, Clim. Past, 16, 2239–2254, https://doi.org/10.5194/cp-16-2239-2020, 2020.
Lowry, D. P. and Morrill, C.: Is the Last Glacial Maximum a reverse analog for future hydroclimate changes in the Americas?, Clim. Dyn., 52, 4407–4427, https://doi.org/10.1007/s00382-018-4385-y, 2019.
Lu, Y. B., An, C. B., Zhao, J. J.: An isotopic study on water sys- tem of Lake Barkol and its implication for Holocene climate dynamics in arid central Asia, Environ. Earth Sci., 73, 1377–1383, https://doi.org/10.1007/s12665-014-3492-2, 2015.
Moreno, P. I. and León, A. L.: Abrupt vegetation changes during the last glacial to Holocene transition in mid-latitude South America, J. Quaternary Sci., 18, 787–800, https://doi.org/10.1002/jqs.801, 2003.
Qin, B. J., Harrison S. P., Kutzback, J. E.: Evaluation of modelled regional water balance using lake status data: a comparison of 6ka simulations with the NCAR CCM, Quat. Sci. Rev., 17, 535-548, https://doi.org/10.1016/S0277-3791(98)00011-0, 1998.
Sun, A. Z., Feng, Z. D., Ran, M., and Zhang, C. J.: Pollen-recorded bioclimatic variations of the last ∼ 22,600 years retrieved from Achit Nuur core in the western Mongolian Plateau, Quatern. Int., 311, 36–43, https://doi.org/10.1016/j.quaint.2013.07.002, 2013.
Vélez M.I., Hooghiemstra H., Metcalfe S., Martínez I., Mommersteeg H: Pollen and diatom based environmental history since the Last Glacial Maximum from the Andean core Fúquene-7, Colombia. J Quat Sci 18:17–30, https://doi.org/10.1002/jqs.730, 2003.
Tarasov, P.E., Harrison, S.P., Saarse, L., Pushenko, M.Ya., Andreev, A.A., Aleshinskaya, Z.V., Davydova, N.N., Dorofeyuk, N.I., Efremov, Yu.V., Khomutova, V.I., Sevastyanov, D.V., Tamosaitis, J., Uspenskaya, O.N., Yakushko, O.F., Tarasova, I.V, Ya, M., Elina, G.A., Elovicheva, Ya.K., Filimonova, L.V., Gunova, V.S., Kvavadze, E.V., Nuestrueva, I.Yu., Pisareva, V.V., Shelekhova, T.S., Subetto, D.A., Zernitskaya, V.P.: Lake status records from the Former Soviet Union and Mongolia: documentation of the second version of the database, Paleoclimatology Publications Series Report #5, https://doi.org/10.25921/49ag-ck94, 1996
Yu, G.: Lake status records from Europe: data base documentation, Publications Series Report #3, https://doi.org/10.25921/p5aq-0931, 1995.
Yu, G. Harrison, S.P. Xue, B.: Lake status records from China: data base documentation, MPI-BGC Tech Rep 4, 2001.
Citation: https://doi.org/10.5194/egusphere-2024-746-AC2
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 431 | 161 | 35 | 627 | 91 | 29 | 19 |
- HTML: 431
- PDF: 161
- XML: 35
- Total: 627
- Supplement: 91
- BibTeX: 29
- EndNote: 19
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Christopher L. Hancock
Michael P. Erb
Nicholas P. McKay
Sylvia G. Dee
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2154 KB) - Metadata XML
-
Supplement
(2191 KB) - BibTeX
- EndNote
- Final revised paper