AquaLinks Water management for Southern Africa

Thinks

Thinks by AquaLinks – Nr 3 – 26 September 2022

Stormwater design rainfall in Johannesburg: Need for an update?

Dear AquaThinkers, Ever wondered if the design rainfall of South Africa needs an update because of climate change? In the SANRAL drainage manual1 and the draft Stormwater Design Manual for Johannesburg2, Smithers and Schulzes design rainfall estimates3 are recommended as the most accurate to use to determine rainfall amounts for certain return periods for certain lengths of time. AquaLinks wondered whether the TAHMO rainfall data of Johannesburg over five years indicate that there is an urgent need as well. See the graphs below for daily rainfall, followed by an explanation.

The curves in blue show the design rainfall for different return periods for total daily rainfall only4, against the probability of exceedance that goes with the definition of such return periods. The orange dots show the percentage of hydrological years out of the five year records, that the annual maximum of daily rainfall of TAHMO daily measurements exceeded these design daily rainfall amounts. For example, see the graph of Midrand, the 1:5 year design rainfall amount is by definition on average supposed to be exceeded 20% of the years. However, the orange dot indicates that two out of five hydrological years (i.e. 40%) the maximum rainfall amount in the year was higher than this design rainfall. Of course, the latest five years could coincidentally have higher amounts of maximum rainfall in a year (for the Midrand example, not every five years the percentage that the 1:5 year design rainfall is exceeded should be 20%).

As the orange dots are almost all above the blue lines, the return periods appear to be shorter than currently used for stormwater designs. The consistent deviation between orange dots and blue lines, gives an indication that updating of design rainfall is needed. This has also been a plead of the University of Kwazulu Natal itself and the stormwater engineers in Johannesburg. There is no intention of this Thinks to generate new design rainfall6 as the record of TAHMO data is far too short for this, but it is a support to the plead for an update.

Knowing that climate change is occurring, “Past performance is no guarantee of future results”. Stormwater management designers should realize that design rainfall estimates based on historical data, may still underestimate the occurrence of high amounts of extreme rainfall. Apart from doing an updated analysis of historical data, we may have to counter in a factor in our design rainfall amounts for the uncertain future, or at least realize that the return periods are changing.

Marieke de Groen, 26 September 2022 (marieke@aqualinks.co.za) (correction graphs version 20 September)

Notes:

  1. SANRAL (2013) Drainage Manual Application Guide, published by The South African National Roads Agency SOC Ltd is still widely used for stormwater design. See also note 3.
  2. For City of Johannesburg there is a Draft Stormwater Design Manual (2019) available. This is a report to the City of Johannesburg to assist practitioners in the application of the Johannesburg Stormwater By-Laws. It was prepared by Newtown Landscape Architects, CBA Specialist Engineers, Fourth Element Consulting and the University of Witwatersrand. There is no final approved version. The draft manual mentions “Further research is being conducted on rainfall patterns in Johannesburg. Until this research is finalised and the results verified, it is recommended that the design depth/duration/frequency characteristics of rainfall be derived from the Design Rainfall Estimator of Smithers and Schulze (2012).” See also note 3.
  3. The Design Rainfall Estimator, latest version 2012, is the software that makes the database with country wide design rainfall estimates based on a regional approach, easily available to users. It can be freely downloaded from the UKZN website (link). The database is based on the Smithers and Schulze (2002) Design Rainfall and Flood Estimation in South Africa (project WRC Project No: K5/1060, link). For analysis of daily to seven days rainfall design estimates, the project used data from 1806 rainfall stations in South Africa which have at least 40 years. The design rainfall by Smithers and Schulze is based on time series of Annual Maximum Series; thus, for example a 1:2 year event could have occurred more than once in a year, but that does not make a difference in the analysis. The data used for the analysis were from before 2002, for some locations ending considerably earlier. Updating new design rainfall estimates based on historical records needs long time series. The longer the return period, ideally the longer the record should be.
  4. The analysis of ground data by Smithers and Schulze uses measured rainfall from 8:00 to 8:00 South African time, as is common in South Africa, while TAHMO uses 0:00to 0:00 Coordinated Universal time, which means 22:00 to 22:00 South African time. No effort was made for this Thinks by AquaLinks to convert hourly data to daily data, to overcome this difference in time slots used. If one of the splits had been at the end of the afternoon, when most extreme hourly rainfall usually occurs, a systematic difference could have been caused, but this is not expected now. For smaller property developments, a day is too long a duration to use. As a trial, for Midrand, the hourly statistics were analysed and they give a similar pattern of higher frequency of occurrence then the analysis of daily rainfall is indicating.
  5. TAHMO data quality checks are done, comparing with neighbouring stations and satellite derived rainfall estimates. The data sets used for this analysis were quality checked. TAHMO has five minute-data available for all stations.
  6. One of the reasons to swop x-and y-axis with the normal design curves is to avoid this misinterpretation of the intention. The other ones are readability and the way in which these graphs were derived (from x to y).

Download pdf version: 220926 Thinks by AquaLinks 3 Design Rainfall Johannesburg

______________________________________________________________________

Thinks by AquaLinks – Nr 2 – 17 May 2022

Extreme rainfall in Johannesburg 4-5 February 2022

Fellow Gautengers, While we all feel for those in Kwazulu-Natal who were affected in the events of 11-13 April 2022, Gauteng has also had its share of extreme rainfall this year. There were fatalities and households affected, due to the rain on 4-5 February 2022. Centurion Lake reached the 1:50 year floodline1 and the Jukskei River in Buccleuch water levels went up with 3 metres2. But how extreme was the rainfall? Find below comparisons of 5 minute, 1 hour and 24 hour rainfall measurements of the TAHMO stations3 in Johannesburg, compared with amounts for design rainfall with different return periods4. For the smaller scale of stormwater management, the smaller time scales are important, but these were not extreme.

What can be concluded from the graphs, is that the rainfall measured in Hunter’s Hill (behind Northgate) was remarkably higher than the rest of the stations indicating a > 200 year daily rainfall, while Midrand also had a daily rainfall of 1 in 50 years and Alexandra 1 in 20 years5. Although the maximum rainfall/hour of > 80 mm in Hunter’s Hill was higher than that of high accuracy guaranteed by the equipment, it is clear that the north of Johannesburg had very high rainfall. The measurements fit into the spatial pattern measured by the other stations. Additional information is welcome6.

Unfortunately, the design rainfall estimation in South Africa is already twenty years not updated anymore, although attempts are done by the UKZN7. It would be interesting to know how daily extreme rainfall amounts in 4 to 5 years of TAHMO rainfall data compare to the design rainfall estimates of 2002. Enough inspiration for a next Thinks by AquaLinks!

Notes:

1) Acknowledgement personal communication Chris Brooker (CBE)

2) Acknowledgement personal communication Stuart Dunsmore (Fourth Element)

3) TAHMO stations are stations from the Trans-African Hydro-Meteorological Observatory (TAHMO). This NGO is increasing the density of automatic weather stations in Africa, adding to the national meteorological services and sharing data with academics, NGO’s and clients (www.tahmo.org). In South Africa there are 10 stations currently of which 8 are for the City of Johannesburg, in cooperation with City of Johannesburg Disaster Management. The stations maintenance is supported by AquaLinks Research and Implementation. The sensors are ATMOS41 sensors, which measure 5 minute rainfall and have a tested accuracy for precipitation of 50 mm/hour with +/- 5% accuracy and measurement capacity up to 400 mm/hour. For more information marieke@aqualinks.co.za.

4) Smithers, JC and Re Schulze (2002), Design Rainfall and Flood Estimation in South Africa, WRC Project No: K5/1060, WRC Report No. 1060/1/03. Use of software from project “Rainfall Statistics for Design Flood Estimation in South Africa”  Version 3 (July 2012).

5) Note that the daily rainfall displayed is from 8 a.m. to 8 a.m. to compare with SAWS data and statistics, but it was raining in most stations before 8 a.m., thus the maximum in 24 hours is higher than displayed. Data of TAHMO station in Orange Farm are not displayed. There was zero rainfall measured there, and this is consistent with pentad rainfall of CHIRPS for that location. However, there are some inconsistencies in the sensor there, so it probably needs replacement.

6) Acknowledgement SAWS data Johannesburg Botanical Gardens and Proefplaas Pretoria (Irene) provided by Munyai Thilivhali (COGTA Gauteng; Disaster Management Gauteng). We did not get hold of SAWS Radar estimates and estimates from other SAWS stations and Wunderground data of that date are not available.

7) Acknowledgement personal communication Katelyn Johnson (UKZN) and provision of Probable Maximum Precipitation for daily rainfall using the 2002 study.

(download pdf version: 220517 Thinks2 Precipitation 4-5 Feb 2022 Johannesburg)

_______________________________________________________________________

Thinks by AquaLinks – Nr 1 – 11 March 2022

Rainfall in Johannesburg 2021-2022 

Fellow Joburgers, are you also wondering how much it already rained since the start of the rain season? The graph below shows cumulative rainfall data from 1 October 2021-9 March 2022[1]. Now measured rainfall is already between 937 (Jabulani) and 1240 (Alexandra) mm from 1 Oct 2021 – 9 March 2022, while it already rained end of September 2021. Friday 4 February 2022 had extreme rainfall in the north. There might be some over measurement of rainfall during such extremes. The trend analysis estimates the mean annual precipitation now to be 743 mm/annum (calendar year)[2]. This is down from 848 mm/annum in 1979. The trend will adjust after 2022, but this rain season is split between two calendar years. Enough inspiration for a next Thinks by AquaLinks!

Notes:

1) The date 1 October is the start of the hydrological year. The measurements are TAHMO data of 4 stations in Johannesburg that had a continuous rainfall record. There are in total 8 TAHMO weather stations with 5 minute data in Johannesburg. Contact AquaLinks (marieke@aqualinks.co.za) for more information.

2) ERA5 data reanalysis by ECMWF as visualized by Meteoblue

(download pdf version: 220311 Thinks by AquaLinks 1 – Precipitation 2021-2022 cumulative Joburg)