Nowcasting for Africa: advances, potential and value - CentAUR
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Nowcasting for Africa: advances, potential and value Article Published Version Creative Commons: Attribution 4.0 (CC-BY) Open Access Roberts, A. J. ORCID: https://orcid.org/0000-0002-4970-9032, Fletcher, J. K., Groves, J., Marsham, J. H., Parker, D. J., Blyth, A. M., Adefisan, E. A., Ajayi, V. O., Barrette, R., Coning, E., Dione, C., Diop, A., Foamouhoue, A. K., Gijben, M., Hill, P. G. ORCID: https://orcid.org/0000-0002-9745-2120, Lawal, K. A., Mutemi, J., Padi, M., Popoola, T. I., Rípodas, P., Stein, T. H. M. ORCID: https://orcid.org/0000-0002-9215-5397 and Woodhams, B. J. (2021) Nowcasting for Africa: advances, potential and value. Weather. ISSN 0043-1656 doi: https://doi.org/10.1002/wea.3936 Available at http://centaur.reading.ac.uk/98503/ It is advisable to refer to the publisher’s version if you intend to cite from the work. See Guidance on citing . Published version at: http://dx.doi.org/10.1002/wea.3936 To link to this article DOI: http://dx.doi.org/10.1002/wea.3936 Publisher: Wiley
All outputs in CentAUR are protected by Intellectual Property Rights law, including copyright law. Copyright and IPR is retained by the creators or other copyright holders. Terms and conditions for use of this material are defined in the End User Agreement . www.reading.ac.uk/centaur CentAUR Central Archive at the University of Reading Reading’s research outputs online
Nowcasting for Africa: advances,
potential and value
Alexander J. Roberts1,2 , 11
Numerical Weather and Climate events but poor in terms of existing forecast-
Weather – Month 9999, Vol. 99, No. 99
Prediction Unit, Nigerian Meteorological ing facilities’. At the time, nowcasting in
Jennifer K. Fletcher1,2, Agency (NiMet), Abuja, Nigeria some countries had been performed for
James Groves2, 12
University of Nairobi, Nairobi, Kenya almost a decade, driven in part by the
increasing prevalence of rainfall radars.
John H. Marsham1,2, 13
Ghana Meteorological Agency (GMet),
The lack of radars in the tropics prevented
Accra, Ghana
Douglas J. Parker1, 14
NiMet Regional Training Centre (RTC),
similar nowcasting activities. However, this
period also marks the beginning of satellite
Alan M. Blyth2, Elijah A. Lagos, Nigeria Earth observations. This is not overlooked
Adefisan3, Vincent O. 15
Agencia Estatal de Meteorología in Browning’s book and there are three
(AEMET), Madrid, Spain chapters dedicated to the use of satellites
Ajayi4, Ronald Barrette5, for nowcasting (Kelly et al., 1982; Shapiro
Estelle de Coning6, et al., 1982; Smith et al., 1982)
GCRF African SWIFT
Cheikh Dione3, Satellite Earth observations have been
available in the intervening 39 years.
The Global Challenges Research Fund
Abdoulahat Diop7, Andre (GCRF) African SWIFT (Science for Weather However, there are almost no automated
K. Foamouhoue3, Morne Information and Forecasting Techniques) operational nowcasting systems used in
project is a £9 million programme of work Africa today. Nowcasting performed by
Gijben8, Peter G. Hill9 , led by the National Centre for Atmospheric the South African Weather Service (SAWS)
Kamoru A. Lawal10,11, Science (NCAS) at the University of Leeds, being a notable exception (de Coning
United Kingdom. SWIFT focuses on forecast- et al., 2015; Gijben and de Coning, 2017).
Joseph Mutemi12, ing research for sub-Saharan Africa across The WMO Nowcasting Guidelines (World
Michael Padi13, Temidayo temporal scales from hours to seasons, as Meteorological Organization [WMO], 2017)
outline that: [in] ‘data-sparse regions, “low-
I. Popoola4,14, Pilar well as capacity building for four partner
countries (Figure 1; Senegal, Ghana, Nigeria cost” nowcasting systems are created by using
Rípodas15, Thorwald and Kenya). SWIFT brings together scien- satellite and lightning data (blended with
numerical weather prediction)’. This is not
H.M. Stein9 and Beth J. tists and meteorologists from 10 African
true for most of Africa where nowcasting is
and five UK-based organisations with the
Woodhams1 World Meteorological Organization (WMO) manual: viewing satellite imagery, calculat-
1
Institute for Climate and Atmospheric as an advisory partner (full list here https:// ing storm speeds and predicting storm posi-
Science, University of Leeds, Leeds, UK africanswift.org/about/). Each African coun- tions, usually limited to aviation purposes.
2
National Centre for Atmospheric try has their national meteorological service The fact that nowcasting technology, data
(NMS) and one university as partners. and guidelines have existed for many years
Science, Leeds, UK
3
African Centre of Meteorological A major research theme of SWIFT is the
improved use of satellite data, especially in
Application for Development (ACMAD),
the context of nowcasting. Nowcasting is
Niamey, Niger the evaluation of the current weather and
4
Federal University of Technology Akure the production of short term predictions to
(FUTA), Akure, Nigeria provide warnings. A more detailed definition
5
Kenya Meteorological Department of African nowcasting is provided in Box 1.
(KMD), Nairobi, Kenya
6
World Meteorological Organisation
(WMO), Geneva, Switzerland
Nowcasting for Africa:
7
Agence Nationale de l’Aviation Civile et de A missed opportunity
la Météorologie (ANACIM), Dakar, Senegal Almost 50 years of nowcasting
8
South African Weather Service –
In the preface to the book Nowcasting
Research, Pretoria, South Africa (Browning, 1982), Keith Browning laments
9
Department of Meteorology, University that ‘the relative lack of material [at the 1981
of Reading, Reading, UK nowcasting symposium] from the Tropics is
10
African Climate and Development regrettable, because nowcasting methods
Initiative (ACDI), University of Cape based on remote sensing are likely to prove Figure 1. Map of African continent with GCRF
Town, Cape Town, South Africa particularly helpful in areas rich in mesoscale African SWIFT partner countries marked. 1
Box 1: A definition of African nowcasting
Nowcasting is the description of the current state of the atmosphere and predictions for the next few hours (WMO, 2017). More
detailed definitions are hard to find, because the process varies regionally and according to the availability of data. The authors
propose that African nowcasting be defined using three criteria:
1. Nowcasting includes analysis of near-real-time observations, to describe current weather conditions.
Nowcasting for Africa
2. Nowcasting forward propagates observed weather features, by extrapolation but also more sophisticated methods using a
wide range of data and dynamical understanding.
3. Nowcasting requires continued monitoring made possible by a rapid workflow of data acquisition, processing and the
dissemination of warnings and updates.
In addition, nowcasting is typically:
• Applied to high impact weather, such as thunderstorms and lightning, fog and atmospheric dust.
• Focussed on stakeholders who are vulnerable to those events such as aviation and fisheries.
Weather – Month 9999, Vol. 99, No. 99
• Focussed on localised forecast domains, so alerts for high-impact weather on the scale of hours can be accurate.
• Not strictly defined by a forward timescale as the utility of nowcasting techniques depends on the meteorological
phenomenon.
• Facilitated by computer systems that bring together disparate data and process them ready for human interpretation by
seasoned forecasters.
without nowcasting becoming widespread, In Africa, NMSs are often closely associ- Satellite Application Facility on Support to
suggests one of two possibilities: (1) there is ated with or part of government depart- NoWCasting and very short range forecasts
little value in African nowcasting or (2) there ments (Tall, 2010). Despite the political (NWC SAF) software, known as NWCSAF/GEO
are barriers to nowcasting. The first of these capital of meteorological themes such as can produce a large number of nowcasting
points is clearly untrue as heavy rainfall, food production, water resource manage- products. It extracts useful information from
strong winds and lightning are abundant ment and public health, funding of NMSs different satellite channels (and NWP data)
in Africa. Therefore, we must conclude that in Africa remains low (WMO, 2019). As such, and generates combined products for now-
there are conditions that prevent African maintaining current systems is prioritised, casting (see Figures 2(a) and (b) for examples
nowcasting. limiting funds available for investment to and Figures S4-S6, S8 and S10 (available in
improve capacity. As such, training oppor- the Supporting Information) for further infor-
tunities are not always abundant. This can mation). These products give greater insight
Barriers for African nowcasting prevent a critical mass of knowledge being than simple visualisations and can be pro-
Often, high impact weather is poorly rep- reached, leaving NMSs vulnerable to the duced rapidly in near-real-time. A number of
resented by numerical weather prediction loss of a few members of staff. products can also be temporally extrapolated.
(NWP) methods. Recent work has shown The divergence in global uptake of The authors are unaware of NWCSAF/GEO
that global ensemble modelling systems nowcasting in the twentieth century can being run in Africa outside of the SAWS.
typically show less skill than climatology in part be attributed to existing infra- While the SAWS products are for southern
across West and central tropical Africa and structure (television, radio, telecoms). The Africa (up to the equator), and are shared
that skill in eastern and southern Africa is ability to disseminate warnings to almost with nearby nations, there are many nations
also low (Vogel et al., 2018, 2020). In par- the entire population in Europe, North within Africa whose forecasters do not have
ticular, the ability to predict extremes can America and Japan lent itself to now- access to relevant nowcasting products.
be especially poor (Vogel et al., 2020). Even casting in a way not matched in Africa. NWCSAF/GEO is not new; it has been freely
convection permitting simulations (over Thankfully, increased connectivity (espe- available (over several iterations) for over
East Africa) show low skill, with rainfall varia- cially with respect to mobile communica- 10 years. The issues preventing its use within
bility dominated by time of day (Woodhams tions) means that this barrier is far smaller Africa are largely due to the lack of knowl-
et al., 2018). This increases the demand than in previous decades. However, there edge of how to implement and maintain the
for nowcasting as a method of providing are other technological barriers faced by software and how to access satellite data
extreme weather warnings. NMSs in Africa. SWIFT partner organisa- from Preparation for the Use of Meteosat
This is manifest as an aspiration to oper- tions have regular issues of intermittent in Africa (PUMA) systems.
ate rainfall radars (Lamptey et al., 2009). power and disrupted internet connectivity. African NMSs have access to satellite data
However, anecdotally, investment in radar While solutions such as the use of uninter- (as well as a raft of other observational and
systems in Africa is often followed by a ruptible power supplies or mobile internet model data) via the EUMETCast service, a
short operational period. Problems of fund- failovers can provide some mitigation they multicast system for the transmission of
ing repairs or limited local expertise then have associated costs and are untenable data operated by EUMETSAT. Received data
lead to equipment being non-operational solutions for long-term operational use. are generally managed within PUMA sys-
for long periods or indefinitely. Once again, tems (Maathuis, 2017) and are visualised via
the exception is the South African radar net- African nowcasting data (and its Synergie software (Le Gallou, 2017). While
work. However, the maintenance difficulties the software allows users to navigate and
faced by the SAWS are well documented accessibility) visualise a wide variety of data it is clear
(Saltikoff, 2015; du Preez, 2017). Thus, Keith The use of satellite data for nowcasting has from communication with forecasters that
Browning’s prediction that remote sensing been recognised as providing an opportu- there are a number of limitations.
(satellite data) would be useful for tropical nity, not so complex or expensive to preclude Anecdotal evidence from forecasters in
2 nowcasting holds true even today. African NMSs (WMO, 2012). The EUMETSAT SWIFT partner NMSs and the authors’ first-
storm has been shown to be a poor pre-
dictor of its future behaviour (Tsonis and
Austin, 1981; Radhakrishna et al., 2012). As
such, one method to enhance predictabil-
ity is to consider the effects of known geo-
graphic or atmospheric features. Examples
include mountain ranges that enhance pre-
Nowcasting for Africa
cipitation; sea, land and lake breezes pro-
ducing convergence; or dry lines which are
associated with storm triggering (Figure 3).
Short, convection-permitting simulations
are now feasible data sources for nowcast-
ing systems (Bowler et al., 2006). NWP simu-
lations initialised frequently and using data
assimilation techniques are able to provide
Weather – Month 9999, Vol. 99, No. 99
improvements over purely deterministic
forecasts (Sun et al., 2014). Similarly, ensem-
ble systems if initialised at short intervals
have the potential to provide useful infor-
mation. The interpretation of such simula-
tions is inevitably probabilistic, as such this
approach lends itself to the production of
probabilistic risk nowcasts. Given the unpre-
dictability of convective storms, this type of
nowcast is appropriate for Africa but might
prove a challenge to communicate to the
public who often expect yes/no answers to
Figure 2. Examples of products available for nowcasting exercises during the SWIFT testbed 1b. All questions like ‘will it rain?’.
panels show 1800 utc 2 May 2019. (a) is the NWCSAF/GEO rapidly developing thunderstorm (RDT) To predict storm triggering, fuzzy logic
product, (b) is the NWCSAF/GEO convective rainfall rate (CRR) product and (c) is the Met Office systems can be used (Mueller et al., 2003;
provided Meteosat imagery where brightness temperature is shown with cold cloud-top tempera- James et al., 2018). These represent the
tures indicated by a colour scale. inherent uncertainty of the input data and
through the application of rules (informed
hand experiences indicate that once within readily viewable using Synergie and so, for by statistical analyses, expert knowledge
the PUMA system, access to data is diffi- the SWIFT partner NMSs at least, these files and dynamical understanding) give proba-
cult. Users of the system are often unsure continue to go unused and are automati- bilistic predictions of storm initiation.
as to what data they have access to and cally deleted by the system. In contrast to these techniques (which are
can be unaware of the management of the driven by a dynamical understanding), there
EUMETCast system. Some products cannot be is now much research being focussed on
visualised via Synergie and so go unnoticed Current nowcasting methods machine learning (Shi et al., 2017; Agrawal
and are automatically deleted without being A feature common to modern operational et al., 2019; Lebedev et al., 2019; Tran and
used. The problem is compounded by the nowcasting systems (including NWCSAF/ Song, 2019), in particular convolutional neu-
automated management of incoming files, GEO) is the automated forward propagation ral networks (CNNs). This field holds great
including changing file names so they are dif- of existing storms through the application of promise for short-term prediction but gives
ferent from the ‘typical file names’ provided storm motion or optical flow vectors (Dixon limited opportunity to develop understand-
by EUMETCast and directory names that defy and Wiener, 1993; Hand, 1996; Mueller ing of physical processes. To the authors’
simple common-sense navigation. Ingesting et al., 2003; Sills et al., 2003; Bowler et al., 2004,
local data (e.g. local surface observations) to 2006; Brovelli et al., 2005; James et al., 2018).
display alongside received data is also com- While extrapolation of this type can work
plicated and so is generally not attempted, very well, it has limitations. It is well suited
hampering analysis and verification of satel- to analysing frontal systems where features
lite or model data. Within Synergie configu- all move in the same direction and at the
rations may be locked for forecasters and same speed or if convective storms do not
settings for complex visualisations cannot be change significantly in size or intensity over
saved to be used repeatedly. This means that the extrapolation period. However, convec-
generating the most useful visualisations can tive storms can vary significantly in their size
be time consuming for forecast staff. and intensity over short periods of time. Also,
In the summer of 2019, 10 years after the extrapolation approach does not predict
NWCSAF/GEO was first released (and fol- the triggering of new storms and as such has
lowing the release of a wide range of a clear predictive limitation, especially in the
near-real-time products by the SWIFT pro- tropics. That said, once storms transition into Figure 3. A SAWS brightness temperature
ject) two products (Rapidly Developing organised systems, forward propagation is difference image (IR12.0–IR10.8) over South
Thunderstorms (RDT) and Convective likely to be a useful technique. Africa on 28 November 2013. A dry line is
Rainfall Rate (CRR)) were made available on To better represent the development of indicated on the image. This feature was
the EUMETCast African service. While this is deep convective storms, some nowcasting associated with the initiation of a very large
a positive step, it has not been particularly systems also represent growth and decay. convective system later the same day. (© 2019
useful. As with other products, they are not However, the past state of a convective EUMETSAT.) 3knowledge, no operational nowcasting sys-
tems yet use CNN techniques. However, it
seems likely that in the future CNNs will be
a powerful tool for nowcasting.
The role of human interpretation is also
important to consider. Local forecasters that
know the system they are using are able to
Nowcasting for Africa
adjust forward projections and account for
spurious features prior to warnings being
issued. Despite technological advances, the
role of the human nowcaster remains an
important step in the production of now-
cast warnings.
Weather – Month 9999, Vol. 99, No. 99
SWIFT progress in the provi-
sion of nowcasting products
Testbeds
SWIFT has a programme of weather fore-
casting ‘testbeds’, bringing researchers and
forecasters together in real-time forecasting
settings (to the authors knowledge the first of
their kind in Africa). Testbeds have been suc- Figure 4. Photos from the GCRF African SWIFT Testbed 1b hosted at the IMTR at the KMD, Nairobi,
cessful in the USA in accelerating the opera- Kenya.
tional use of new tools (Ralph et al., 2013).
The SWIFT testbeds are helping to engender a
sense of shared purpose across meteorologi-
cal institutions from both Africa and the UK.
SWIFT Testbed 1b was hosted in April/
May 2019 at the Kenya Meteorological
Department (KMD), Nairobi (Figure 4)
and in partnership with the WMO-led
High Impact Weather Lake System project
(HIGHWAY). Testbed 1b produced opera-
tional forecasts for East and West Africa
and evaluated the use of data and tech-
niques (Fletcher et al., 2020). The partici-
pants represent a wealth of expertise in
the fields of East and West African mete-
orology. Researchers learnt about the
techniques employed by forecasters and
how meteorological fields are interpreted
in specialised ways and forecasters were
exposed to research ideas, products and
techniques they had not previously used.
This exchange of ideas was only possible
due to the hard work of the participants
and their willingness to share knowledge
and learn from others. Nowcasting was one
of the three main strands of Testbed 1b
along with synoptic forecasting and fore-
cast evaluation. These fed into one another:
forecasts informed nowcasts and nowcast-
ing products enabled the evaluation of
earlier forecasts.
The testbeds are a showcase for nowcast-
ing products. A major achievement was the
provision of NWCSAF/GEO products by NCAS Figure 5. (a) The installation and position of the 2.4m C band satellite antenna located at the
and the University of Leeds (Figures 2(a) and Chilbolton observatory. (b) A schematic indicating the location of the hardware and processing
(b) RDT and CRR respectively) and Meteosat required to produce NWCSAF/GEO products and make images available for African forecasters and
derived imagery (Figure 2(c)) and ATDNet researchers.
Sferics (lightning) by the UK Met Office. All
these products were generated in the UK Data is processed using NWCSAF/GEO and are being produced with a typical latency
and made available online for the testbed. the resultant images are then catalogued of 30min (the first time this has ever been
A 2.4m satellite antenna has been installed and made publicly available online (sci.ncas. done for East and West Africa). The use of
4 at the Chilbolton Observatory, UK (Figure 5). ac.uk/swift). At the time of writing, images this system was, in part, to highlight whatis possible for African NMSs to achieve with roducts can be used as a resource to con-
p a change in operating methods. SWIFT is
respect to operational nowcasting as much duct research. working with partner NMSs to help in the
of the equipment required is extant within use of the Common Alerting Protocol (CAP)
NMSs and the rest is readily available and to disseminate warnings through a wide
relatively low cost. Nowcasting potential in Africa range of media. In addition to this, a SWIFT
NWCSAF/GEO products provided by There is great potential for increasing now- extension project is working on an appli-
SWIFT have since been used operationally casting capacity across Africa. The provision cation programming interface (API) for use
Nowcasting for Africa
and during a joint SAtellite and Weather of nowcasting products through SWIFT has with mobile applications for the delivery of
Information for Disaster Resilience in Africa allowed nowcasting to be possible opera- nowcasting warnings (Forecasting African
(SAWIDRA) and the African Centre of tionally across East and West Africa. However, STorms Application (FASTA)), for which
Meteorological Application for Development there is considerable work needed for African African partner institutions will provide local
(ACMAD) forecast demonstration exercise. NMSs to make nowcasting routine. meteorological expertise and warnings.
Feedback is currently being gathered with
the aim of improving visualisation methods
Training Planning for the future
and providing development feedback to
Weather – Month 9999, Vol. 99, No. 99
NWC SAF. A major element of making this a reality is The Meteosat Third Generation (MTG) satel-
training of future forecasters. Educational lites represent a major technological step
centres such as partner universities, ACMAD, forward and could greatly enhance oppor-
Building capacity to implement tunities for nowcasting in Africa (Stuhlmann
the NiMet (Nigerian Meteorological Agency)
nowcasting in Africa Regional Training Centre (RTC), the KMD et al., 2005). Specifically, the new lightning
The experience gained setting up the now- Institute for Meteorological Training and imager will provide information on electri-
casting system described above means that Research (IMTR), Ecole Africaine de la cally active storms across the whole con-
NCAS can provide advice and training to Météorologie et de l’Aviation Civile (EAMAC) tinent. However, the sheer scale of data
build capacity for operational nowcasting and workshops run as part of the Severe generated by the MTG satellites means that
in Africa. In August 2019, training was deliv- Weather Forecasting Programme in regional it is likely that the EUMETCast Africa ser-
ered to staff from the SWIFT partner institu- specialised meteorological centres will not vice (the primary source of forecast/nowcast
tions on how to set up receiving hardware only educate forecasters in partner coun- data for many African NMSs) will only con-
and run NWCSAF/GEO, this covered data tries but more broadly across Africa. This tain a subset of the data produced. SWIFT
preprocessing, software use, visualisation ensures a legacy of nowcasting skills that is working alongside NWC SAF, the SAWS
and automation. The supporting material will outlast the SWIFT funding period. and the WMO to highlight the effects of
for this course was provided as a website downgrading the Africa service and miti-
(https://sites.google.com/ncas.ac.uk/swift- gate the effect of subsetting satellite data to
safnwctraining). Further training scheduled
Development protect nowcasting capability. It is currently
for June 2020 has been postponed due to SWIFT development of additional nowcast- still uncertain whether the introduction of
the Covid-19 global pandemic. Meanwhile ing tools is ongoing; for example, investi- MTG will be beneficial to African users.
technical support for groups managing local gating methods for extending extrapolation
hardware and software is being provided. lead-times for convective systems. It is sus-
Value of satellite and
Alongside support and training there is an pected that large, well-developed systems
ongoing discussion about getting NWCSAF/ that are able to persist for many hours (and nowcasting research for Africa
GEO run operationally within African NMSs. have the highest impact) are likely to be The value of making automated nowcast-
Additionally, four satellite antennas have predictable for several hours. ing products available to African NMSs is
been purchased for partner universities. The While local generation of nowcasting undoubtedly great. The manual approach
Kwame Nkrumah University of Science and products in Africa is a target, there is a used for aviation safety is labour intensive
Technology (KNUST), Ghana have received need to take an holistic approach to now- and so is difficult to expand to national
and set up their receiving station. SWIFT casting. Without a means of interrogating scales. However, augmenting it with the
aims to facilitate African partners in com- incoming data and enabling input from a techniques discussed earlier and introducing
missioning and maintaining such systems. forecaster, there is no way to easily gener- automation will allow forecasters to produce
African ownership of these responsibilities ate meaningful warnings. Similarly, without warnings quickly for much wider regions.
will prevent reliance on UK generated prod- a mechanism to rapidly distribute warnings Many industries would benefit from
ucts beyond the lifetime of SWIFT (ending the process of data collection, process- greater access to warnings of imminent
December 2021). ing, analysis and warning production is a high impact weather that are currently not
Communication between the University waste of resources. Therefore, SWIFT is also widely available from African NMSs. For
of Leeds/NCAS and NWC SAF is ongoing involved in solutions to these issues. example: the energy sector (wind, oil and
and there is mutual interest in collabora- NCAS are already working alongside gas), commercial/state logistics and haulage,
tion on the development/evaluation of the UK Met Office in the development ports, fisheries, military and agriculture to
NWCSAF/GEO and the delivery of train- of FOREST (Forecast Observations and name a few. NMSs could provide commercial
ing in Africa. SWIFT NWCSAF/GEO evalua- Research Evaluation and Survey Toolkit). nowcasting services; as such, this represents
tion work conducted at the University of FOREST is a new visualisation tool devel- potential new sources of funds.
Reading (Hill et al., 2020) helps us under- oped by the UK Met Office as part of the By improving their ability to produce and
stand how NWCSAF/GEO products should Weather and Climate Science for Service disseminate warnings there is the poten-
be used over Africa. The role of ACMAD Partnership (WCSSP). The use of such a tool tial to raise the public profile and trust in
within SWIFT will help to generate inter- in Africa would be beneficial as it broad- African NMSs. Skilled human interpreta-
est in using NWCSAF/GEO products and ens access to model products and enables tion of nowcasting fields is highly valuable
provide a pathway for future training in visualisation of near-real-time nowcasting for even the most advanced nowcasting
Africa. Nowcasting products have been products produced locally. systems, as such, experienced forecasters
generated for SWIFT research case studies, Dissemination on timescales useful for play an important role in successful opera-
these will be useful to understand how such nowcasting is a great challenge and requires tional nowcasting. The wealth of forecasting 5expertise within SWIFT partner NMSs means up of new equipment and systems to Lamptey BL, Pandya RE, Warner TT.
they are well placed to take advantage of bring nowcasting to Africa. et al. 2009. The UCAR Africa Initiative. Bull.
Am. Meteorol. Soc. 90: 299-303.
the current market gap.
SWIFT nowcasting work also has a series Le Gallou L. 2017. SYNERGIE-PUMA 2015
of less obvious benefits. Cooperation References Administration Manual. MeteoFrance
International: Toulouse, France.
between African universities and NMSs Agrawal S, Barrington L, Bromberg C.
et al. 2019. Machine learning for precip- Lebedev V, Ivashkin V, Rudenko I,
(including across national boundaries) is a Ganshin A, Molchanov A., Ovcharenko S.
itation nowcasting from radar images,
Nowcasting for Africa
requirement for SWIFT, creating a greater in 33rd Conference on Neural Information et al. 2019. Precipitation nowcasting with
capacity to share experiences, knowledge Processing Systems (NeurIPS 2019), satellite imagery. KDD ’19 Proceedings
and innovations. Centres of education Vancouver, Canada. arXiv: 1912.12132. of the 25th ACM SIGKDD International
Conference on Knowledge Discovery and
can make use of connections to NMSs to Bowler NEH, Pierce CE, Seed AW. 2004. Data Mining. arXiv: 1905.09932.
understand what training is valuable and Development of a precipitation nowcast-
ing algorithm based upon optical flow Maathuis B. 2017. MESA and PUMA
tailor content as required. Similarly, the Station Updates Guide. Monitoring for
techniques. J. Hydrol. 288: 74–91.
use of nowcasting systems in universities Environment and Security in Africa
Bowler NEH, Pierce CE, Seed AW. 2006. (MESA).
for research will produce results useful for
Weather – Month 9999, Vol. 99, No. 99
STEPS: a probabilistic precipitation fore-
NMSs. casting scheme which merges an extrapo- Mueller C, Saxen T, Roberts R. et al.
Support from centres such as ACMAD lation nowcast with downscaled NWP. Q. J. 2003. NCAR auto-nowcast system. Weather
R. Meteorol. Soc. 132: 2127–2155. Forecast. 18: 545–561.
(who have created a new Department of
Research and Development; DRD) will allow Brovelli P, Senesi S, Arbogast E. et al. du Preez RG. 2017. Deployment,
2005. Nowcasting thunderstorms with Re-engineering and Risk Analysis of a
for the development of new Africa focused C-band Weather Radar to Build Local
SIGOONS: A significant weather object
nowcasting techniques. Partner universi- oriented nowcasting system. In: World Capacity in South Africa. North-West
ties will be in a stronger position to lead Weather Research Program Symposium University, South Africa. https://
and collaborate on meteorology research on Nowcasting and Very Short Range repository.nwu.ac.za/bitstream/han-
Forecasting. WMO: Toulouse, France, dle/10394/25097/Du_Preez_RG_2017.
and be able to use their research to steer pdf?sequence=1&isAllowed=y.
NWCSAF/GEO development. In the past 7.08.
Browning KA. 1982. Preface, in Radhakrishna B, Zawadzki I, Fabry
NWC SAF focus has mostly been on mid- F. 2012. Predictability of precipita-
Nowcasting. Browning KA (ed). Academic
latitude improvements (reflecting their Press: London, UK, pp ix–xi. tion from continental radar images.
users). Increasing the user base in Africa Part V: growth and decay. J. Atmos. Sci.
de Coning E, Gijben M, Maseko B. et al. 69: 3336–3349.
and active participation in NWC SAF user 2015. Using satellite data to identify and
engagement will inevitably lead to product track intense thunderstorms in South and Ralph FM, Intrieri J, David Andra RA Jr.
improvements for tropical Africa. southern Africa. S. Afr. J. Sci. 111: 1–9. et al. 2013. The emergence of weather-
related test beds linking research and
Dixon M, Wiener G. 1993. TITAN: forecasting operations. Bull. Am. Meteorol.
Concluding remarks Thunderstorm Identification, Tracking, Soc. 94: 1187–1211.
Analysis, and Nowcasting - a radar-based
It is hoped that the nowcasting work of methodology. J. Atmos. Ocean. Technol. 10: Saltikoff E. 2015. The threat to weather
785–797 radars by wireless technology. Bull. Am.
SWIFT can provide a roadmap for how Meteorol. Soc. 97: 1159–1167.
African nowcasting can be achieved. While Enno SE, Sugier J, Labrador L,
Rudlosky S. 2018. Comparison and Shapiro MA, Krueger AJ, Kennedy
we accept that this is an ambitious plan PJ. 1982. 2.5 Nowcasting the position
characterisation of ATDnet versus LIS for
there is a great need and much potential the period of 2008 to 2014. In: Proceedings and intensity of jet streams using a
for improvement. We recognise the risk of the 25th International Lightning Detection satellite-borne total ozone mapping
that advances might be temporary as with Conference/Seventh International Lightning spectrometer, in Nowcasting. Browning
Meteorology Conference. Lauderdale, KA (ed) . Academic Press: London, UK, pp
previous nowcasting projects in Africa that 137–145.
Florida, 12–15 March 2018.
have had initially impressive results (UK Met
Fletcher J, Parker D, Hartley A, Nkiaka Shi X, Gao Z, Lausen L. et al. 2017.
Office [UKMO], 2014; Thiery et al., 2017). The Deep learning for precipitation nowcast-
E, Adefisan E, Clarke S. et al. 2020. GCRF
SWIFT approach of: avoiding complicated or African SWIFT Testbed 1 Report. https:// ing: a benchmark and a new model.
expensive infrastructure, providing techni- africanswift.org/wp-content/uploads/ Adv. Neural Inf. Proces. Syst. 30(2017):
cal training and support and encouraging sites/84/2020/04/Testbed1_Report_Final. 5617–5627
the cooperation of African institutions will pdf [accessed 9 February 2021] Sills D, Driedger N, Greaves B. et al.
Gijben M, de Coning E. 2017. Using satel- 2003. iCAST: A Prototype Thunderstorm
hopefully encourage self-sufficiency and
lite and lightning data to track rapidly Nowcasting System focused on
with it effect long-lasting change. Optimization of the Human-Machine Mix.
developing thunderstorms in data sparse
regions. Atmosphere 8: 64. In: Proceedings of the 25th AMS Conference
on Severe Local Storms. Cloud Physics
Acknowledgements Hand WH. 1996. An object-oriented
technique for nowcasting heavy showers
and Severe Weather Research Section,
Environment Canada.
This work was supported by UK Research and thunderstorms. Meteorol. Appl. 3:
31–41. Smith WL, Suomi VE, Zhou FX. et al.
and Innovation as part of the Global 1982. 2.4 Nowcasting applications of
Challenges Research Fund, African SWIFT Hill PG, Stein THM, Roberts AJ. et al. geostationary satellite atmospheric
programme, grant no. NE/P021077/1. 2020. How skilful are Nowcasting Satellite sounding data, in Nowcasting. Browning
Applications Facility products for tropical KA (ed) . Academic Press: London, UK, pp
Marsham was also supported by the Africa? Meteorol. Appl. 27: e1966. 123–136.
NCAS via the NERC/GCRF programme
James PM, Reichert BK, Heizenreder D. Stuhlmann R, Rodriguez A, Tjemkes S.
Atmospheric hazard in developing 2018. NowCastMIX: automatic integrated et al. 2005. Plans for EUMETSAT’s Third
Countries: Risk assessment and Early warnings for severe convection on Generation Meteosat geostationary
Warning (ACREW). The authors would nowcasting time scales at the German satellite programme. Adv. Space Res. 36:
also like to acknowledge the work of col- Weather Service. Weather Forecast. 33: 975–981.
1413–1433.
leagues at SWIFT partner institutions both Sun J, Xue M, Wilson JW et al. 2014.
Kelly GAM, Forgan BW, Powers PE. et al. Use of NWP for nowcasting convec-
inside and outside of Africa to make the
1982. 2.3 Mesoscale observations from a tive precipitation: recent progress and
content of this paper possible. Specifically, polar orbiting satellite vertical sounder. In: challenges. Bull. Am. Meteorol. Soc. 95:
those that have made the SWIFT testbeds Nowcasting. Browning KA (ed). Academic 409–426.
6 run smoothly and are navigating the set Press: London, UK, pp 107–122.Tall A. 2010. Climate Forecasting to Serve cessed ensemble predictions of rainfall Correspondence to: Alexander J. Roberts
Communities in West Africa. World Climate in the tropics. Weather Forecast. 35:
Conference - 3. Procedia Environ. Sci. 1: 2367–2385. a.j.roberts1@leeds.ac.uk
421–431. WMO. 2012. CBS Steering Group SWFDP. © 2021 The Authors. Weather published by
Thiery W, Gudmundsson L, Bedka K. https://www.wmo.int/pages/prog/www/ John Wiley & Sons Ltd on behalf of the Royal
et al. 2017. Early warnings of hazardous CBS-Reports/documents/SG-SWFDP-4_ Meteorological Society
thunderstorms over Lake Victoria. Environ. Final_Report.pdf [accessed 9 February
Res. Lett. 12: 074012. 2021]. This is an open access article under the
terms of the Creative Commons Attribution
Nowcasting for Africa
Tran Q-K, Song S. 2019. Computer WMO. 2017. Guidelines for Nowcasting
vision in precipitation nowcasting: Techniques. WMO, Geneva, Switerland. License, which permits use, distribution and
Applying image quality assessment met- https://library.wmo.int/doc_num. reproduction in any medium, provided the
rics for training deep neural networks. php?explnum_id=3795. [accessed 9 original work is properly cited.
Atmosphere 10: 244. February 2021].
doi: 10.1002/wea.3936
Tsonis AA, Austin GL. 1981. An evalua- WMO. 2019. 4th African Ministerial
tion of extrapolation techniques for the Conference On Meteorology
short-term prediction of rain amounts. (AMCOMET-4): Integrated African Supporting Information
Atmosphere-Ocean, 19: 54-65. Strategy on Weather and Climate
Figures S1–S11. Examples of how now-
Weather – Month 9999, Vol. 99, No. 99
UKMO. 2014. Saving Lives on Lake Victoria. Services. WMO: Geneva, Switerland.
UK Met Office: Exeter, UK. https://www. https://amcomet.wmo.int/sites/default/ casting images routinely generated as part
preventionweb.net/files/40481_140500s files/field/doc/events/draft_african- of the Global Challenges Research Fund
avinglivesonlakevictoria1.pdf [accessed 9 strategy-on-weather_climate_ser- (GCRF) African SWIFT (Science for Weather
February 2021]. vices_ver_5_clean_0.pdf [accessed 9
February 2021].
Information and Forecasting Techniques)
Vogel P, Knippertz P, Fink AH. et al. project were used for operational nowcast-
2018. Skill of global raw and postpro- Woodhams BJ, Birch CE, Marsham JH
et al. 2018. What is the added value ing during the SWIFT testbed 1b.
cessed ensemble predictions of rainfall
over Northern Tropical Africa. Weather of a convection-permitting model for
Forecast. 33: 369–388. forecasting extreme rainfall over tropi- All Supporting Information is available with
cal East Africa? Mon. Weather Rev. 146: the online edition of the article (which is free
Vogel P, Knippertz P, Fink AH. et al. 2757–2780.
2020. Skill of global raw and postpro- for members to access).
7You can also read
