ADM Stormwater & Subsoil Drainage Masterplan - Rainfall IDF Curves Update_with Analysis Refs

c:\users\nae16497\documents\my documents\projects\abu dhabi\adm stormwater & subsoil drainage masterplan - rainfall idf curves update_with analysis refs.docx

Abu Dhabi Municipality – Stormwater and Subsoil Drainage Masterplan

Update of Rainfall Intensity-Duration-Frequency Curves

SUMMARY METHODOLOGY 1. Introduction Phase 1 of the project Terms of Reference includes the following task: “Update existing design curves using all available meteorological data”. This refers to the update of the existing rainfall Intensity-Duration-Frequency (IDF) curves in current use in Abu Dhabi Municipality. The project area is defined as the “whole jurisdiction area of the Municipality of Abu Dhabi City”. This area is approximately 11,000km2 and constitutes the area for which existing IDF curves have been updated. 2. Collection & review of data 2.1 Meteorological data All available and relevant raw rainfall data have been requested. NCMS hold daily rainfall totals for a network of rain gauges across Abu Dhabi Emirate and daily data were obtained for the following gauges:

• Abu Dhabi International Airport (1982 – 2013; 32 years)

• Abu Dhabi City (2003 – 2013; 11 years)

• Abu Alaryam (2008 – 2013; 6 years)

• Al Wathbah (2003 – 2013; 11 years)

• Ras Ghanadah (2006 – 2011; 6 years)

• Abu Alabyad (2006 – 2013; 8 years)

• Razeen (2003 – 2011; 9 years)

• Al Khaznah (2003 – 2011; 9 years)

These eight gauges were initially selected as providing comprehensive and available coverage across and immediately surrounding the project area. Data were provided in the form of MS Excel files. No further raw rain gauge data (e.g. sub-daily data) were made available for any rain gauges in Abu Dhabi. 2.2 Existing Design Manuals Existing manuals in use for stormwater design in Abu Dhabi include:

• Sewerage Projects Committee (SPC), 1996 [in current use by ADM]

• Urban Planning Council (UPC), 2011

• Roadway Design Manual (RDM), 2004

IDF relationships are available in all 3 manuals, typically in the form of IDF curves. These relationships have been reviewed in this study as part of the IDF curve update process. Very limited raw rainfall data covering the ADM project area and used in the derivation of the IDF relationships is provided in these manuals. However, the UPC manual gives annual maximum (AMAX) daily rainfall totals for Abu Dhabi International Airport (1982 – 2009) and Swiehan (2003 – 2009).


Neither the SPC manual Chapter 6 nor the RDM manual sections discussing the IDF relationships reproduce any raw rainfall data. However, an SPC Stormwater System Hydrology Report (1989) does provide 3-hourly rainfall totals for one day – 19/2/1988 – when extreme rainfall was recorded at both the International and the Bateen airports. Further, this SPC report provides (in Tables 2.1 – 2.7) a number of ‘peaks over threshold’ (PoT) series for a range of storm durations recorded at the International airport between 1982 and 1989, although only Table 2.5 (the 3-hourly PoT series) has been provided for analysis in this study. 3. Methodology summary There is a paucity of sub-daily data in Abu Dhabi on which to base the derivation and update of IDF curves. The traditional method of analysing a large number of recorded storm events at a range of durations (e.g. from 15 minutes up to 24 hours) has therefore not been possible. Instead, discussion with ADM resulted in the decision to apply the following methodology:

• Update of daily rainfall analysis, to include recent years of data and more local (project area specific) rain gauges than used in previous analyses.

• Use of sub-daily/daily rainfall ratios to derive rainfall depths for durations of less than 24-hours, with ratios assessed from the existing SPC IDF curves as well as IDF work in neighbouring Dubai (due to the very similar climate to Abu Dhabi).

• Comparison of ratios and resulting IDF curves with 3-hour / 24-hour ratios taken from the SPC Hydrology Report.

• Fitting of IDF curves and recommendation of preferred curves for use by ADM.

4. Analysis of Daily data The daily data series for all 8 gauges (listed in 2.1 above) have been updated with the last few years of data to provide the longest records available. Summary rainfall statistics have been derived for each gauge (e.g. recorded totals, totals for the same periods of record, mean daily rainfall, maximum annual rainfall, etc). These statistics have been reviewed, together with the location and elevations of each of the gauges, in order to determine climate similarity and appropriateness for use in the IDF updates. Additionally, probability distributions have been fit to the data from each of the gauges and the resultant rainfall quantiles have been compared to identify potential outliers and data concerns. Please see ‘AMAX analysis’ file. ‘AMAX data’ worksheet provides the annual maximum 24-hour series for each of the 8 gauges originally assessed (derived from the raw rainfall data files provided to Hyder). The following 8 worksheets (‘Distribution fits_X’, etc) provide the probability distribution fits and fitting parameters – these have been derived from ‘WINFAP’ – Windows Frequency Analysis Package – software used by Hyder. Screenshots of the programme results are provided separately. Some of the gauges lie outside, or on the edge of, the project area, whilst one lies in a very rural, hinterland, area where there is limited existing or planned development. Some gauges have very short records and produce outlier quantiles when PDFs are fitted. Analysis suggests that these outliers are not the facets of different climatological conditions or elevation and are, in all likelihood, due to the shortness of the records involving the omission of key historic events. The above analysis has resulted in the dismissal of 4 gauges from further analysis, leaving the following 4 gauges to be used in the IDF curve updates for the project area: Please see ‘Summary statistics comparison table’ file. This provides a summary of the key statistics for each gauge (derived from the raw rainfall data files) and shows how certain gauges have been omitted from further analysis.

• Abu Dhabi International Airport (1982 – 2013; 32 years)

• Abu Dhabi City (2003 – 2013; 11 years)

• Al Wathbah (2003 – 2013; 11 years)


• Ras Ghanadah (2006 – 2011; 6 years)

It is assumed that these four gauges adequately describe the daily rainfall characteristics of the project area and that the daily data provided are reliable. Extreme value frequency analysis has been undertaken. Three different probability distributions have been fit to the AMAX data from each of the gauges. The distribution that best fits the data at all gauges is the Generalised Extreme Value (GEV) distribution, with fitting by L-Moments. Please see ‘AMAX analysis’ file. ‘AMAX data’ worksheet provides the annual maximum 24-hour series for each of the 8 gauges originally assessed (derived from the raw rainfall data files provided to Hyder). The following 8 worksheets (‘Distribution fits_X’, etc) provide the probability distribution fits and fitting parameters – these have been derived from ‘WINFAP’ – Windows Frequency Analysis Package – software used by Hyder. Screenshots of the programme results are provided separately. The rainfall quantiles derived from fitting the GEV distribution are summarised in a table in the next worksheet ‘Depth & Intensity plots’. This worksheet also provides an intensity-frequency plot for each gauge. The rainfall quantiles (2, 5, 10, 25, 50 and 100 year return periods) at each gauge have been compared. They are found to be very similar at low return periods but the higher quantiles are significantly lower at Ras Ghanadah. This gauge has a very similar mean annual rainfall to Abu Dhabi City and Al Wathbah gauges, but its short record means it has missed out on some key historic events and consequently has a lower maximum daily rainfall. Although Abu Dhabi International Airport has by far the longest (and perhaps most reliable) record, the pooling of data from all applicable gauges within this region can enhance the derivation of rainfall quantiles. The quantiles have been divided by the mean AMAX for each gauge to produce a dimensionless coefficient for each return period; these have then been weighted by the length of record at each gauge to produce a ‘regional dimensionless coefficient’ for each return period. The mean AMAX from each gauge has also been weighted on record length and multiplied by the regional dimensionless coefficients to give a new, regional, set of rainfall quantiles: Please see ‘AMAX analysis’ file, ‘Regionalisation’ worksheet. The rainfall quantiles for the 4 gauges taken forward for analysis are copied from the ‘Depth & Intensity’ plots worksheet. The procedure outlined above has then been followed and all stages of the calculations can be seen on the ‘Regionalisation’ worksheet. (Note – the mean AMAX at each gauge has been derived from the raw rainfall records). The regionalised depth-frequency plot below is also provided in this worksheet.

24-hr Regional

Design Rainfall

Depth (mm)

T

2 18.76

5 38.75

10 55.20

25 80.68

50 103.73

100 130.87

with a 24-hour regional frequency curve:


As would be expected by the length of record, this regional frequency curve is closest to that of Abu Dhabi International Airport analysed as a single-site, although it produces quantiles a little lower. In general, it produces higher quantiles than Abu Dhabi City or Ras Ghanadah. 5. Exploration of Sub-daily data In the absence of sub-daily rainfall data from Abu Dhabi, a number of sources have been used to, firstly, investigate sub-daily/daily relationships and, secondly, to apply these to derive depth-duration-frequency data and, ultimately, IDF curves. 5.1 Dubai Stormwater Design Manual (2000) This manual provides IDF curves and tables for return periods from 2 – 1000 years and durations from 30 minutes – 24 hours. Dubai shares a similar climate with Abu Dhabi, with extreme rainfall predominantly derived from high intensity, short duration, storms. The Dubai ratios are typical of expected extreme rainfall relationships for arid zones with, for example, between 49 and 57% of the 24-hour total falling in just 2 hours. It is therefore considered appropriate, in the absence of Abu Dhabi specific sub-daily data, to explore the use of the ratios between short duration and 24-hour rainfall depths from Dubai in this update of Abu Dhabi IDF curves. Please see ‘Dubai design manual IDF curves & AD SPC ratios’ file. The ‘Dubai IDF curves’ worksheet gives the IDF relationship taken from the Dubai Stormwater Design Manual from which the depth-duration ratios have been tabulated by Hyder. This worksheet also provides the IDF equation and fitting parameters from the manual. Please see ‘AMAX analysis’ file, ‘Depth Duration (Dubai ratios)’ worksheet. This takes the ratios described above and applies them to the Abu Dhabi 24-hour regionalised rainfall quantiles provided in the ‘Regionalisation’ worksheet to generate short-duration rainfall depths for Abu Dhabi. The intensities are then presented, derived from the depths, and unsmoothed IDF curves are plotted. 5.2 SPC Manual (1996) This manual provides IDF curves for return periods from 1 – 100 years and durations from 5 minutes – 24 hours. It is not clear from the manual how these curves have been derived, although the data and discussion provided in the SPC Hydrology Report (1989) [see Section 2.2 above] suggests that these curves may have been produced based on the analysis of a range of storm events with varying durations recorded at the International Airport between 1982 and 1989.

0

20

40

60

80

100

120

140

1 10 100

Ra

infa

ll D

ep

th (

mm

)

Return Period

24-hour ADM Regional Frequency Curve


It appears that sub-daily data was routinely measured and recorded at the airport gauge between these years; it is therefore credible that the relationships recorded between sub-daily and daily may reliably represent those regularly experienced in ADM, albeit that the 8 year record is short. Analysis reveals that the ‘SPC ratios’ are very similar to the ‘Dubai ratios’, although a little lower, e.g.: 50-year: 3 / 24-hour ratio 6 / 24-hour SPC 0.50 0.63 Dubai 0.52 0.67 Please see ‘Dubai design manual IDF curves & AD SPC ratios’ file. The ‘Abu Dhabi SPC IDF curves’ worksheet gives the IDF relationship taken from the SPC Manual from which the depth-duration ratios have been tabulated by Hyder. Please see ‘AMAX analysis’ file, ‘Depth Duration (AD SPC ratios)’ worksheet. This takes the ratios described above and applies them to the Abu Dhabi 24-hour regionalised rainfall quantiles provided in the ‘Regionalisation’ worksheet to generate short-duration rainfall depths for Abu Dhabi. The intensities are then presented, derived from the depths, and unsmoothed IDF curves are plotted. 5.3 SPC Hydrology Report (1989) – 19th February 1988 The 3-hourly rainfall totals for one day – 19/2/1988 – when extreme rainfall was recorded at both the International and the Bateen airports, are insufficient to determine any reliable meaningful relationships. However, using these data, the maximum depths for intervals of 3, 6, 9, 12, 15, 18 and 21 hours have been calculated and ratios determined, as follows:

Ratio: 3 / 24 6 / 24 9 / 24 12 / 24 15 / 24 18 / 24 21 /24 AD Int – 19/2/88

0.38 0.72 0.81 0.86 0.98 1.0 1.0

The 3-hour ratio is lower than either the Dubai or SPC 50-year 3-hour ratios, but the 6-hour is higher than both. However, little can be drawn from this analysis as the 19/2/88 data represents only one day (the highest daily total on record) and is potentially comprised of more than one single storm event. Please see ‘Dubai design manual IDF curves & AD SPC ratios’ file. The ‘AD 19 Feb 1988’ worksheet presents this analysis. 5.4 SPC Hydrology Report (1989) – 3-hourly PoT series The 3-hourly 1982 - 1989 PoT series from the International Airport provides 24 No. 3-hourly maximum rainfall totals, all of which are 8.0mm or over (i.e. 8.0mm has been selected as the threshold). These data have been analysed to determine ratios to 24-hourly totals for each of the event days. The ratios vary from 0.12 on 19/2/88 to 1.00 on several dates, including 5/4/83. The lowest ratio comes from what appears to have been assessed as an independent event to the main storm event on that day, the day that recorded the largest daily total in Abu Dhabi (120mm). The highest ratio comes on a day when the entire 24-hour rainfall (20.2mm) appears to have fallen in just 3 hours. Extreme value analysis indicates that this 24-hour total has a return period just over 2 years; however, the fall of 20.2mm in just 3-hours is the 3rd highest 3-hourly total in the 1982 – 1989 record and is likely to have a higher return period, in the region of 5 years. This discrepancy and high variation in ratios illustrates the inadequacies of utilising daily data analysis and ratios when compared with actual, recorded, sub-daily data. The average 3 / 24-hour ratio from this PoT series is 0.65, which is more in line with the 6 hour ratios derived from the existing SPC and Dubai IDF curves. Please see ‘Dubai design manual IDF curves & AD SPC ratios’ file. The ‘AD 3-hourly PoT’ worksheet presents this analysis. 6. Fitting of IDF curves


Rainfall depths for each duration and return period, using (1) Dubai ratios, and (2) SPC ratios for sub-daily durations, have been converted to rainfall intensities and plotted on log-log paper. In both cases, the Bernard empirical IDF equation has been found to describe the IDF relationships well and appropriate IDF curves have been fit: Please see ‘AMAX analysis’ file, ‘Fit curves – Dubai ratios’ and ‘Fit curves – SPC ratios’ worksheets. These worksheets show the fitting of the Bernard equation to the data determined in the ‘Depth Duration (Dubai ratios) and ‘Depth Duration (AD SPC ratios)’ worksheets. The smoothed IDF curves are presented and the fitting parameters given.


IDF Table for ADM using short duration/24hr depth ratios from 2000 Dubai manual Duration (mins) 5 10 15 30 60 120 180 360 720 1440

Duration (hrs) 0.083 0.167 0.25 0.5 1 2 3 6 12 24

Return period (yrs)

Rainfall Intensity (mm/h)

100 277.89 172.13 130.07 80.57 49.91 30.91 23.36 14.47 8.96 5.55

50 220.26 136.43 103.09 63.86 39.56 24.50 18.51 11.47 7.10 4.40

25 176.10 108.70 81.97 50.60 31.24 19.28 14.54 8.98 5.54 3.42

15 150.82 92.97 70.05 43.18 26.62 16.41 12.36 7.62 4.70 2.90

10 123.61 76.09 57.29 35.26 21.71 13.36 10.06 6.19 3.81 2.35

5 91.71 56.03 41.99 25.65 15.67 9.57 7.18 4.38 2.68 1.64

2 51.89 31.07 23.02 13.78 8.25 4.94 3.66 2.19 1.31 0.79

0

1

10

100

0.01 0.1 1 10 100

Ra

infa

ll I

nte

nsi

ty m

m/h

r

Duration (Hours)

IDF Curves for ADM using short duration/24hr depth ratios from 2000

Dubai manual

100 year

50 year

25 year

15 year

10 year

5 year

2 year


IDF Table for ADM using short duration/24hr depth ratios from 1996 SPC manual Duration (mins) 5 10 15 30 60 120 180 360 720 1440

Duration (hrs) 0.083 0.167 0.25 0.5 1 2 3 6 12 24

Return period (yrs)

Rainfall Intensity (mm/h)

100 218.87 139.10 106.70 67.81 43.09 27.39 21.01 13.35 8.48 5.39

50 187.67 118.28 90.29 56.90 35.86 22.60 17.25 10.87 6.85 4.32

25 149.05 93.75 71.47 44.95 28.27 17.78 13.56 8.53 5.36 3.37

15 130.66 81.66 62.04 38.77 24.23 15.15 11.51 7.19 4.50 2.81

10 112.76 70.04 53.01 32.93 20.45 12.70 9.62 5.97 3.71 2.30

5 83.40 51.44 38.78 23.92 14.76 9.10 6.86 4.23 2.61 1.61

2 41.97 25.39 18.93 11.45 6.93 4.19 3.12 1.89 1.14 0.69

0

1

10

100

0.01 0.1 1 10 100

Ra

infa

ll I

nte

nsi

ty (

mm

/hr)

Duration (Hrs)

IDF Curves for ADM using short-duration/24hr depth ratios from 1996 SPC

manual

100 year

50 year

25 year

15 year

10 year

5 year

2 year


7. Comparison and final preferred curves The updated IDF curves derived using the ‘Dubai ratios’ produce slightly higher intensities at all return periods and all durations than those produced using the ‘SPC ratios’. Based on the limited sub-daily data available at the International Airport (discussed above in Sections 5.3 and 5.4 above), the higher ‘Dubai ratios’ are to be preferred. They are therefore recommended for use until such time as significant raw sub-daily data becomes available for analysis. A comparison of the preferred IDF curves with those in the SPC manual (1996) shows that the preferred curves give higher intensities for all durations at higher return periods (i.e. less frequent events), but lower intensities for all durations at lower return periods (more frequent events). This is considered likely to be, in part, due to (1) the update of the AMAX series using many more recent years of data from the International Airport where fewer extremes have generally been experienced (see Figure below), and (2) the inclusion of data from additional rain gauges within the project area which have shorter records and do not include the extremes of the 1980s and 1990s. The preferred curves generally give slightly higher or very similar intensities to those from the RDM manual and are not too dissimilar to those in the UPC manual.

8. Further work It is recommended that the preferred IDF curves (Section 6, using ‘Dubai ratios’) are re-worked if and when significant sub-daily data becomes available for one or more gauges within or in close proximity to the project area. Ideally, this would be sub-daily data for the International Airport for the entire record (1982 – 2013). At that time, it would also be appropriate to consider the fitting of an alternative, generalised, 3 or 4 fitting-parameter, empirical IDF equation to describe the extreme rainfall relationships for all return periods. Hyder Consulting February 2014

0

20

40

60

80

100

120

140

01

/01

/19

82

01

/01

/19

84

01

/01

/19

86

01

/01

/19

88

01

/01

/19

90

01

/01

/19

92

01

/01

/19

94

01

/01

/19

96

01

/01

/19

98

01

/01

/20

00

01

/01

/20

02

01

/01

/20

04

01

/01

/20

06

01

/01

/20

08

01

/01

/20

10

01

/01

/20

12

Ra

infa

ll (m

m)

Date

Daily Rainfall Total (mm) - Abu Dhabi International

ADM Stormwater & Subsoil Drainage Masterplan - Rainfall IDF Curves Update_with Analysis Refs

Documents

Transcript of ADM Stormwater & Subsoil Drainage Masterplan - Rainfall IDF Curves Update_with Analysis Refs