the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Nov 2024
Interdecadal Cycles in Australian Annual Rainfall
Abstract. The extremes of Australian rainfall have profound economic, ecological and societal impacts; however, the current forecast horizon is limited to a few months. This study investigates interdecadal periodicity in annual rainfall records across eastern Australia. Wavelet analysis was conducted on rainfall data from 347 sites covering 130 years (1890–2020). Prominent cycles were extracted from each site and clustered using a Gaussian Mixture Model. This revealed three principal cycles centred around 12.9, 20.4 and 29.1 years that were highly significant over red noise by t-test (p<0.0001). Overall, the three cycles combined had a mean contribution to total rainfall variance (R2) of 13 % across all sites, but this was up to 29 % at individual sites. Both the 12.9-yr and 20.4-yr cycles were detected at over 95 % of sites. The strength of each cycle varied over time and this amplitude modulation of the signal showed a systematic movement across the area investigated. 86 % of extremely wet years fell within the positive phase of the combined reconstruction, with 80 % of extremely dry years falling in the negative phase. These results indicate underlying periodicity in annual rainfall across eastern Australia, with the potential to build this into long-term forecasts. This concept has been suggested in the past, but not rigorously tested. These findings open new paths for research into rainfall patterns in Australia and internationally. They also have broad implications for the management of water resources across all sectors.
- Preprint
(2711 KB) - Metadata XML
-
Supplement
(4174 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-3149', Anonymous Referee #1, 08 Dec 2024
The manuscript is clearly written and presented. The methods used are standard and clear. The results are appropriately analyzed. The main shortcoming is that there is no effort to explain the results or to add insight to them. For instance, I assume that there are distinct seasons for rainfall in at least part of the region of Australia that was analyzed
Would it make sense to look for composites of atmospheric and oceanic state variables for the +ve and -ve phases of the 3 oscillations and/or to do a similar wavelet-clustering analysis of those variables and/or look for the wavelet coherence of some of those variables with the precipitation time series?
I think doing something like that would help make the paper more convincing
Some similar work is reported in https://www.mdpi.com/2306-5338/10/3/67
Citation: https://doi.org/10.5194/egusphere-2024-3149-RC1 -
AC1: 'Reply on RC1', Tobias Selkirk, 25 Jan 2025
Thank you for taking the time to review this manuscript.
The point raised about possible correlations to the dominant climate modes affecting rainfall in Australia and other oceanic-state variables is an excellent one and we have considered it at great length. Initial results were not included in this paper for several reasons but mostly due to the scope and report length:
- Initial aim and scope: the purpose of this study was to test the previous findings that the 18.6-yr Lunar Nodal Cycle and ~11-yr sunspot cycles were drivers of Australian rainfall using more comprehensive data and rigorous statistical analysis. The unexpected finding was that although the lunisolar cycles did not appear to be present there were other clear cycles of slightly different periodicity.
- Paper length: having discovered the cycles, more emphasis was put on a complete description and exhaustive statistical testing. Much of this work needed to be edited down to fit in a single paper and adding in wider climate variables would have involved either sacrificing some of the evidence required to first substantiate the claim, or increasing the length of the manuscript (which is already quite long). Neither of these seemed like viable options.
It is likely that these cycles are ultimately related to, or mediated by, known climate drivers in some way. Though the influence of the El Niño-Southern Oscillation (ENSO), the Indian Ocean Dipole (IOD) and the Southern Annular Mode (SAM) are significant in Eastern Australia, their impact is not limited to this region alone. We decided that the next step should be to first look at global datasets for other regions which may show similar evidence of influence by these cycles and then consider how this could relate to broader climate variables. Work on this front is currently in progress, and the manuscript will be updated to reflect this rationale. Thank you for the direction to the Amonkar et al (2023) paper, it will be a valuable reference in the consideration of the relationship to climate indices and will be referenced in the revised manuscript as well.
Citation: https://doi.org/10.5194/egusphere-2024-3149-AC1
-
AC1: 'Reply on RC1', Tobias Selkirk, 25 Jan 2025
-
RC2: 'Comment on egusphere-2024-3149', Anonymous Referee #2, 03 Jan 2025
In this study, the authors investigate interdecadal periodicity in annual rainfall records across eastern Australia. They employ wavelet analysis and a Gaussian Mixture Model to extract and cluster prominent cycles from rainfall data collected at 347 sites covering 130 years (1890-2020). The results confirm the existence of an underlying periodicity in annual rainfall across eastern Australia, with three dominant cycles identified in the rainfall records. This analysis aims at explaining some of the disparate earlier findings and provides the basis for building long-term forecasts. This could open new paths for research into rainfall patterns in Australia and internationally, with broad implications for the management of water resources across all sectors.
The scientific contribution of this paper falls within the scope of Hydrology and Earth System Sciences. The paper is well-written with a clear and well-organized structure. The findings are appropriately discussed and related to previous works on the same topic. The discussion is efficiently supported by figures.
I suggest considering just some minor revisions:
- While the concept of periodicity of rainfall events in Australia is widely and carefully explored, citing a number of previous studies, the choice of using wavelet analysis in this work is not adequately motivated. Is this the first application of wavelet analysis to a hydrological dataset? What are the advantages? Please provide a wider explanation and eventually cite previous papers using this technique to support your choice.
- The visualization of the results is crucial and I found all the figures suitable to convey the different messages about dominant cycles. The only figure that needs some substantial changes is the first. I suggest improving Fig.1 to make it easier to interpret. Furthermore, it appears to have a very low definition, I suggest improving the visualization quality.
- In Fig.5, the letters that indicate the different panels do not correspond to the letters in the caption and in the text where the results are discussed.
- In Fig.4 and 6, I suggest changing the colormap to help the reader distinguish the cycle families and I also recommend making more evident the sites with significant cycles over red noise (I have difficulties in the identification of the points circled in white as it is now).
- Page 5 line 102: possible typo “p”lane.
- Page 15 line 311: typo "20.4-yrcyclecan".
- Page 22 line 427: possible typo “the allowed”.
Citation: https://doi.org/10.5194/egusphere-2024-3149-RC2 -
AC2: 'Reply on RC2', Tobias Selkirk, 25 Jan 2025
Thank you for your detailed and insightful comments on the manuscript.
Regarding the first comment about the use of wavelet analysis, this method is indeed very common in the exploration of periodicity in hydrological data, such as rainfall (Chowdhury & Beecham, 2012; Murumkar & Arya, 2014; Santos & Morais, 2013; Williams et al., 2021) and streamflow (Briciu, 2014; Brown et al., 2021; Gorodetskaya et al., 2024). The paper will be updated to make note of this. Its main advantage over something like a Fourier transform is the ability to identify (and easily visualise) modulation in the period and amplitude of cycles - therefore accommodating some of the non-stationary properties observed by previous researchers (Vines et al, 2004 - see Page 5 line 92). It was also the visualisation of cycle power in the time domain which allowed us to identify that was previously thought to be a "phase-change" was actually a change in the dominant cycle due to amplitude modulation. These advantages of wavelet analysis will be expanded upon in the methods section.
What was novel about the way wavelet analysis was used in this paper was the automated peak extraction and cluster analysis by Gaussian Mixture Model. This was a newly developed method which allowed for the analysis of a much large number of spectra than usual, along with testing statistical significance of the number of sites compared to red and white noise. The results of wavelet analysis are often interpreted visually and independently in a limited region, such as Chowdhury & Beecham (2012)- 53 rain gauge stations, limited to South Australia, Murumkar& Arya (2014) - 4 rain gauge stations, limited to the Nira Catchment in India, or Amonkar et al., (2023) - 24 sites limited to the Ohio River Basin in America. This method allowed for the combined analysis of 347 rain gauge stations covering an area of nearly 4 million km2 - this is the largest area covered by any study that we are aware of into periodicity in rainfall using wavelet analysis. We will modify the discussion to highlight this advantage.
Thank you for the feedback on the figures. It is good to know that most were successful at conveying the message.
Figure 1 - Agreed, we will update the figure and attempt to improve the clarity and resolution.
Figure 4 - Thank you for point this out, the colormap is matched to the cycles throughout the paper but changing the land colour to match the grey in Figure 9 should significantly improve the contrast and clarity of the cycle families. We will do this for all figures which feature blue landmass for consistency.
Figure 5 - The figure text will be updated to match the panels
Figure 6 - Formatting will be modified to increase visibility of the significant sites over red noise
Page 5 line 102: possible typo “p”lane. - corrected
Page 15 line 311: typo "20.4-yrcyclecan". - corrected
Page 22 line 427: possible typo “the allowed”.- corrected
Citation: https://doi.org/10.5194/egusphere-2024-3149-AC2
Status: closed
-
RC1: 'Comment on egusphere-2024-3149', Anonymous Referee #1, 08 Dec 2024
The manuscript is clearly written and presented. The methods used are standard and clear. The results are appropriately analyzed. The main shortcoming is that there is no effort to explain the results or to add insight to them. For instance, I assume that there are distinct seasons for rainfall in at least part of the region of Australia that was analyzed
Would it make sense to look for composites of atmospheric and oceanic state variables for the +ve and -ve phases of the 3 oscillations and/or to do a similar wavelet-clustering analysis of those variables and/or look for the wavelet coherence of some of those variables with the precipitation time series?
I think doing something like that would help make the paper more convincing
Some similar work is reported in https://www.mdpi.com/2306-5338/10/3/67
Citation: https://doi.org/10.5194/egusphere-2024-3149-RC1 -
AC1: 'Reply on RC1', Tobias Selkirk, 25 Jan 2025
Thank you for taking the time to review this manuscript.
The point raised about possible correlations to the dominant climate modes affecting rainfall in Australia and other oceanic-state variables is an excellent one and we have considered it at great length. Initial results were not included in this paper for several reasons but mostly due to the scope and report length:
- Initial aim and scope: the purpose of this study was to test the previous findings that the 18.6-yr Lunar Nodal Cycle and ~11-yr sunspot cycles were drivers of Australian rainfall using more comprehensive data and rigorous statistical analysis. The unexpected finding was that although the lunisolar cycles did not appear to be present there were other clear cycles of slightly different periodicity.
- Paper length: having discovered the cycles, more emphasis was put on a complete description and exhaustive statistical testing. Much of this work needed to be edited down to fit in a single paper and adding in wider climate variables would have involved either sacrificing some of the evidence required to first substantiate the claim, or increasing the length of the manuscript (which is already quite long). Neither of these seemed like viable options.
It is likely that these cycles are ultimately related to, or mediated by, known climate drivers in some way. Though the influence of the El Niño-Southern Oscillation (ENSO), the Indian Ocean Dipole (IOD) and the Southern Annular Mode (SAM) are significant in Eastern Australia, their impact is not limited to this region alone. We decided that the next step should be to first look at global datasets for other regions which may show similar evidence of influence by these cycles and then consider how this could relate to broader climate variables. Work on this front is currently in progress, and the manuscript will be updated to reflect this rationale. Thank you for the direction to the Amonkar et al (2023) paper, it will be a valuable reference in the consideration of the relationship to climate indices and will be referenced in the revised manuscript as well.
Citation: https://doi.org/10.5194/egusphere-2024-3149-AC1
-
AC1: 'Reply on RC1', Tobias Selkirk, 25 Jan 2025
-
RC2: 'Comment on egusphere-2024-3149', Anonymous Referee #2, 03 Jan 2025
In this study, the authors investigate interdecadal periodicity in annual rainfall records across eastern Australia. They employ wavelet analysis and a Gaussian Mixture Model to extract and cluster prominent cycles from rainfall data collected at 347 sites covering 130 years (1890-2020). The results confirm the existence of an underlying periodicity in annual rainfall across eastern Australia, with three dominant cycles identified in the rainfall records. This analysis aims at explaining some of the disparate earlier findings and provides the basis for building long-term forecasts. This could open new paths for research into rainfall patterns in Australia and internationally, with broad implications for the management of water resources across all sectors.
The scientific contribution of this paper falls within the scope of Hydrology and Earth System Sciences. The paper is well-written with a clear and well-organized structure. The findings are appropriately discussed and related to previous works on the same topic. The discussion is efficiently supported by figures.
I suggest considering just some minor revisions:
- While the concept of periodicity of rainfall events in Australia is widely and carefully explored, citing a number of previous studies, the choice of using wavelet analysis in this work is not adequately motivated. Is this the first application of wavelet analysis to a hydrological dataset? What are the advantages? Please provide a wider explanation and eventually cite previous papers using this technique to support your choice.
- The visualization of the results is crucial and I found all the figures suitable to convey the different messages about dominant cycles. The only figure that needs some substantial changes is the first. I suggest improving Fig.1 to make it easier to interpret. Furthermore, it appears to have a very low definition, I suggest improving the visualization quality.
- In Fig.5, the letters that indicate the different panels do not correspond to the letters in the caption and in the text where the results are discussed.
- In Fig.4 and 6, I suggest changing the colormap to help the reader distinguish the cycle families and I also recommend making more evident the sites with significant cycles over red noise (I have difficulties in the identification of the points circled in white as it is now).
- Page 5 line 102: possible typo “p”lane.
- Page 15 line 311: typo "20.4-yrcyclecan".
- Page 22 line 427: possible typo “the allowed”.
Citation: https://doi.org/10.5194/egusphere-2024-3149-RC2 -
AC2: 'Reply on RC2', Tobias Selkirk, 25 Jan 2025
Thank you for your detailed and insightful comments on the manuscript.
Regarding the first comment about the use of wavelet analysis, this method is indeed very common in the exploration of periodicity in hydrological data, such as rainfall (Chowdhury & Beecham, 2012; Murumkar & Arya, 2014; Santos & Morais, 2013; Williams et al., 2021) and streamflow (Briciu, 2014; Brown et al., 2021; Gorodetskaya et al., 2024). The paper will be updated to make note of this. Its main advantage over something like a Fourier transform is the ability to identify (and easily visualise) modulation in the period and amplitude of cycles - therefore accommodating some of the non-stationary properties observed by previous researchers (Vines et al, 2004 - see Page 5 line 92). It was also the visualisation of cycle power in the time domain which allowed us to identify that was previously thought to be a "phase-change" was actually a change in the dominant cycle due to amplitude modulation. These advantages of wavelet analysis will be expanded upon in the methods section.
What was novel about the way wavelet analysis was used in this paper was the automated peak extraction and cluster analysis by Gaussian Mixture Model. This was a newly developed method which allowed for the analysis of a much large number of spectra than usual, along with testing statistical significance of the number of sites compared to red and white noise. The results of wavelet analysis are often interpreted visually and independently in a limited region, such as Chowdhury & Beecham (2012)- 53 rain gauge stations, limited to South Australia, Murumkar& Arya (2014) - 4 rain gauge stations, limited to the Nira Catchment in India, or Amonkar et al., (2023) - 24 sites limited to the Ohio River Basin in America. This method allowed for the combined analysis of 347 rain gauge stations covering an area of nearly 4 million km2 - this is the largest area covered by any study that we are aware of into periodicity in rainfall using wavelet analysis. We will modify the discussion to highlight this advantage.
Thank you for the feedback on the figures. It is good to know that most were successful at conveying the message.
Figure 1 - Agreed, we will update the figure and attempt to improve the clarity and resolution.
Figure 4 - Thank you for point this out, the colormap is matched to the cycles throughout the paper but changing the land colour to match the grey in Figure 9 should significantly improve the contrast and clarity of the cycle families. We will do this for all figures which feature blue landmass for consistency.
Figure 5 - The figure text will be updated to match the panels
Figure 6 - Formatting will be modified to increase visibility of the significant sites over red noise
Page 5 line 102: possible typo “p”lane. - corrected
Page 15 line 311: typo "20.4-yrcyclecan". - corrected
Page 22 line 427: possible typo “the allowed”.- corrected
Citation: https://doi.org/10.5194/egusphere-2024-3149-AC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 182 | 59 | 8 | 249 | 28 | 8 | 7 |
- HTML: 182
- PDF: 59
- XML: 8
- Total: 249
- Supplement: 28
- BibTeX: 8
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 121 | 50 |
| Australia | 2 | 17 | 7 |
| China | 3 | 14 | 5 |
| Germany | 4 | 11 | 4 |
| France | 5 | 9 | 3 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 121