the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
29 Jan 2021
29 Jan 2021
Spatially Compounded Surge Events: An Example from Hurricanes Matthew and Florence
- 1Department of Physics and Lt. Col. James B. Near, Jr., USAF, ’77 Center for Climate Studies, The Citadel, Charleston, SC, 29409, USA
- 2Department of Geography, Planning and Environment, East Carolina University, Greenville, NC, 27858, USA
- 3Department of Economics, East Carolina University, Greenville, NC, 27858, USA
- 4Engineering Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, 20899, USA
- 1Department of Physics and Lt. Col. James B. Near, Jr., USAF, ’77 Center for Climate Studies, The Citadel, Charleston, SC, 29409, USA
- 2Department of Geography, Planning and Environment, East Carolina University, Greenville, NC, 27858, USA
- 3Department of Economics, East Carolina University, Greenville, NC, 27858, USA
- 4Engineering Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, 20899, USA
Abstract. The simultaneous rise of tropical cyclone induced flood waters across a large hazard management domain can stretch rescue and recovery efforts. Here we present a means to quantify the connectedness of maximum surge during a storm with geospatial statistics. Tide gauges throughout the extensive estuaries and barrier islands of North Carolina deployed and operating during Hurricanes Matthew (n = 82) and Florence (n = 123) are used to compare the spatial compounding of surge for these two disasters. Moran's I showed the occurrence of maximum storm tide was more clustered for Matthew compared to Florence, and a semivariogram analysis produced a spatial range of similarly timed storm tide that was four times as large for Matthew than Florence. A more limited data set of fluvial flooding and precipitation in eastern North Carolina showed a consistent result – multivariate flood sources associated with Matthew were more concentrated in time as compared to Florence. Although Matthew and Florence were equally intense, they had very different tracks and speeds, which influenced the timing of surge along the coast. We hope this method could be used for other landfalling tropical cyclones to better understand the drivers that lead to spatially compounded surge events.
Scott Curtis et al.
Status: closed
-
RC1: 'Comment on nhess-2021-31', Anonymous Referee #1, 23 Feb 2021
Just a couple minor comments: 1. The goal of the study is somehow not well stated in the introduction. 2. On the abstract, the last sentence belongs more in the conclusions, and not in the abstract. 3. L52 add "and severe drop in atmospheric pressure" after winds.
-
AC1: 'Reply on RC1', Scott Curtis, 24 Mar 2021
- Thank you for your comments. We agree that our goal of characterizing the spatially compounded nature of storm surge was not well articulated in the introduction and this has been remedied.
- The reviewer is correct that the last sentence does not belong in the abstract and has been removed.
- Secondary to winds, a severe drop in atmospheric pressure contributes to storm surge and this has been added as suggested.
-
AC1: 'Reply on RC1', Scott Curtis, 24 Mar 2021
-
RC2: 'Comment on nhess-2021-31', Anonymous Referee #2, 07 Mar 2021
General comments
This paper considers the spatial compounding of storm surge in Eastern North Carolina during Hurricanes Matthew and Florence. While the study provides a useful method for understanding the spatial extent of surge hazards and informing emergency management activities, I am not convinced that all of the data presented actually represent wind-driven storm surge. Hurricane Florence’s winds dropped to 45mph by the end of the day on September 14, and the storm appears to have left the study area by September 15. This suggests to me that the peak water levels associated with Δt>3 in the northern part of the study area in Figure 5 are actually driven by another process, not by wind-driven surge. This would likely mean that the time distributions of the two hurricanes are actually more similar than reported here.
Specific comments
Line 106: It is unclear to me what distribution the K-S test was applied to. The text says it was used to evaluate the distributions of tide magnitude and timing – was this done at each location within the study site, or was the spatial distribution compared at each time after landfall? This is clarified a bit in Tables 1-2 but should be explained in the Methods.
Line 125: The authors never explain why the stable variogram was chosen. Were other variograms tested? The results of the model testing should be presented, if not in the paper then in a supplement.
Line 115: “n is the total number” of what?
Figure 3: I don’t think is it necessary to show an example of a semivariogram. The authors could point out these features on the fitted semivariograms in Figure 6.
Figure 4: Do the crosses indicate that rivers near Cape Fear reached flood stage during both storms, but no rivers near Lake Mattamuskeet ever reached flood stage (since there are no black crosses)? Please explain what “near” means in the caption (near to what?).
Figure 6: These data should be plotted on separate graphs so that the semivariogram for Matthew can be seen more clearly.
Technical corrections
Figure 2: The day/time text is difficult to read because it overlaps the storm track line and other text. Please adjust so this information is more legible.
Wind speeds are reported in knots in the text but m/s in Figure 2. Please be consistent.
Line 161: Replace “boxes” with “squares” to be consistent with the Figure 5 caption.
Line 181: This reference should be for Table 4.
-
AC2: 'Reply on RC2', Scott Curtis, 24 Mar 2021
General comments.
We appreciate the reviewer’s observation and we were remiss in describing the evolution of Hurricane Florence from September 15 to 18. During that time Florence recurved to the north and by 5am EST on the 18th the storm was classified as extratropical and positioned at 41.3N and 75.9W (see daily weather map below). At the time many of the northern coasts of the northern estuaries (Albemarle Sound) were receiving significant SE winds, lowest pressure, and highest surge. An example is given for NCCUR12568. This important information is now included in the manuscript.
Specific comments.
Line 106: The methodology has been clarified.
Line 125: There are several semivariogram models to choose, but over half are inappropriate for the data. For example, the exponential semivariogram, used by Touma et al. (2018), gives an unphysical range for Florence (see Table A below). After a visual inspection and some experimentation, we selected the stable semivariogram. By in large, the results from other candidate models yield the same key result – Matthew has a larger range than Florence. We can add more information as a supplement if necessary.
Table A. Statistics of the exponential semivariogram models of Δtps-lf for Hurricanes Matthew and Florence.
Storm Sill (Co) Shape (s) Range (a) Nugget (b) Matthew 0.047 N/A 36,877 m 0.012 Florence 0.894 N/A 1,369 m 0.0 Line 115: We apologize for the omission. It should be number of tide gauge stations.
Figure 3: We have removed Figure 3 and point out these features in Figure 6.
Figure 4: The crosses are representative rivers in the Eastern Branch of the NCDEM that were affected by the two hurricanes (squares in Figure 5). Only the Cape Fear Lock #1 location contains both a river gauge and a rain gauge. The coloring of the crosses and labeling probably adds to the confusion. The figure and table caption have both been revised and the accompanying text clarified.
Figure 6: We have plotted the graphs separately. However, one reason to plot them on the same scale is to emphasize that Florence has much larger values of γ than Matthew indicating that Florence was more “dissimilar” than Matthew even at short distances between tide gauges.
Technical corrections
We have addressed all the technical corrections.
-
AC2: 'Reply on RC2', Scott Curtis, 24 Mar 2021
Status: closed
-
RC1: 'Comment on nhess-2021-31', Anonymous Referee #1, 23 Feb 2021
Just a couple minor comments: 1. The goal of the study is somehow not well stated in the introduction. 2. On the abstract, the last sentence belongs more in the conclusions, and not in the abstract. 3. L52 add "and severe drop in atmospheric pressure" after winds.
-
AC1: 'Reply on RC1', Scott Curtis, 24 Mar 2021
- Thank you for your comments. We agree that our goal of characterizing the spatially compounded nature of storm surge was not well articulated in the introduction and this has been remedied.
- The reviewer is correct that the last sentence does not belong in the abstract and has been removed.
- Secondary to winds, a severe drop in atmospheric pressure contributes to storm surge and this has been added as suggested.
-
AC1: 'Reply on RC1', Scott Curtis, 24 Mar 2021
-
RC2: 'Comment on nhess-2021-31', Anonymous Referee #2, 07 Mar 2021
General comments
This paper considers the spatial compounding of storm surge in Eastern North Carolina during Hurricanes Matthew and Florence. While the study provides a useful method for understanding the spatial extent of surge hazards and informing emergency management activities, I am not convinced that all of the data presented actually represent wind-driven storm surge. Hurricane Florence’s winds dropped to 45mph by the end of the day on September 14, and the storm appears to have left the study area by September 15. This suggests to me that the peak water levels associated with Δt>3 in the northern part of the study area in Figure 5 are actually driven by another process, not by wind-driven surge. This would likely mean that the time distributions of the two hurricanes are actually more similar than reported here.
Specific comments
Line 106: It is unclear to me what distribution the K-S test was applied to. The text says it was used to evaluate the distributions of tide magnitude and timing – was this done at each location within the study site, or was the spatial distribution compared at each time after landfall? This is clarified a bit in Tables 1-2 but should be explained in the Methods.
Line 125: The authors never explain why the stable variogram was chosen. Were other variograms tested? The results of the model testing should be presented, if not in the paper then in a supplement.
Line 115: “n is the total number” of what?
Figure 3: I don’t think is it necessary to show an example of a semivariogram. The authors could point out these features on the fitted semivariograms in Figure 6.
Figure 4: Do the crosses indicate that rivers near Cape Fear reached flood stage during both storms, but no rivers near Lake Mattamuskeet ever reached flood stage (since there are no black crosses)? Please explain what “near” means in the caption (near to what?).
Figure 6: These data should be plotted on separate graphs so that the semivariogram for Matthew can be seen more clearly.
Technical corrections
Figure 2: The day/time text is difficult to read because it overlaps the storm track line and other text. Please adjust so this information is more legible.
Wind speeds are reported in knots in the text but m/s in Figure 2. Please be consistent.
Line 161: Replace “boxes” with “squares” to be consistent with the Figure 5 caption.
Line 181: This reference should be for Table 4.
-
AC2: 'Reply on RC2', Scott Curtis, 24 Mar 2021
General comments.
We appreciate the reviewer’s observation and we were remiss in describing the evolution of Hurricane Florence from September 15 to 18. During that time Florence recurved to the north and by 5am EST on the 18th the storm was classified as extratropical and positioned at 41.3N and 75.9W (see daily weather map below). At the time many of the northern coasts of the northern estuaries (Albemarle Sound) were receiving significant SE winds, lowest pressure, and highest surge. An example is given for NCCUR12568. This important information is now included in the manuscript.
Specific comments.
Line 106: The methodology has been clarified.
Line 125: There are several semivariogram models to choose, but over half are inappropriate for the data. For example, the exponential semivariogram, used by Touma et al. (2018), gives an unphysical range for Florence (see Table A below). After a visual inspection and some experimentation, we selected the stable semivariogram. By in large, the results from other candidate models yield the same key result – Matthew has a larger range than Florence. We can add more information as a supplement if necessary.
Table A. Statistics of the exponential semivariogram models of Δtps-lf for Hurricanes Matthew and Florence.
Storm Sill (Co) Shape (s) Range (a) Nugget (b) Matthew 0.047 N/A 36,877 m 0.012 Florence 0.894 N/A 1,369 m 0.0 Line 115: We apologize for the omission. It should be number of tide gauge stations.
Figure 3: We have removed Figure 3 and point out these features in Figure 6.
Figure 4: The crosses are representative rivers in the Eastern Branch of the NCDEM that were affected by the two hurricanes (squares in Figure 5). Only the Cape Fear Lock #1 location contains both a river gauge and a rain gauge. The coloring of the crosses and labeling probably adds to the confusion. The figure and table caption have both been revised and the accompanying text clarified.
Figure 6: We have plotted the graphs separately. However, one reason to plot them on the same scale is to emphasize that Florence has much larger values of γ than Matthew indicating that Florence was more “dissimilar” than Matthew even at short distances between tide gauges.
Technical corrections
We have addressed all the technical corrections.
-
AC2: 'Reply on RC2', Scott Curtis, 24 Mar 2021
Scott Curtis et al.
Scott Curtis et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 427 | 68 | 11 | 506 | 4 | 6 |
- HTML: 427
- PDF: 68
- XML: 11
- Total: 506
- BibTeX: 4
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1