Bridging the gap between forecasters and operational hydrologists: an OzEWEX summer institute project

Contributed by Melanie Loveridge, Bex Dunn and Yiling Liu

How often do we assume that we understand the users’ needs, which may later be proven untrue? At the recent OzEWEX Australian Climate and Water Summer Institute – held in Canberra, Australia – we got the chance to bridge the gap between forecasters and operational hydrologists.

Fifteen of us were invited to the OzEWEX summer institute, which provides early career researchers the chance to see the current and emerging data and technologies in use within government departments. Following some initial intensive training, we were given the opportunity to use this technology in an innovative but practically useful way.

In the end, we landed a group of three to understand the forecast needs of river and storage operators. We felt that there are so many amazing datasets that it’s unnecessary to create more. Instead, we thought – how do we make existing data more useful to the people who are actually using it? What better way to do this, than going back to basics and defining our end users’ needs.

Discovery of user requirements was performed through an online survey. It was designed to gauge current usage and desirable additions or improvements, and capture associations with different demographics. To date we have received 29 responses throughout Australia – keeping in mind that there is only a small Australian population of river and storage operators. This was then supplemented by follow-up phone interviews and four focus groups.

Our operator’s responses to forecast information have given rise to some very interesting initial findings. Existing forecast information (provided primarily by the Bureau of Meteorology in Australia) is clearly valued, with 79% of users satisfied with the service. Most operators also believed that enhancements could be made to existing forecast information.

More effective communication is key!

Operators wanted to see three key areas of communication addressed. This consistently came up when speaking with operators, as they want to be better enabled to interpret hydro-meteorologic forecasts. Most critically being confidence in forecasts or prediction uncertainties.

Similarly, where multiple forecasts of the same hydro-meteorologic variable is available, they wanted to understand the differences between those estimates. This could also be addressed through an increased understanding of the skill or uncertainty of predictions.

Lastly, they wanted more information about the technique used to create the forecast. This finding was curious as information about the underlying techniques are all provided on the Bureau of Meteorology’s website. It might be more related to the level of understanding of the forecast methods, i.e. maybe it would be more appropriate to have simplified explanations with supporting media (videos, illustrations, etc.).

Additional nationally consistent forecasts.

Communication is often seen as the biggest challenge for operators, however, additional nationally consistent forecasts were sought after. Although already in existence, operators wanted improved reliability for 7-day forecasts and beyond.

One dataset that several operators were particularly passionate about getting is an irrigation demand forecast. Amongst other things, these forecasts can predict the volume of irrigated water required in a catchment, valley or irrigation district. These predictions prove crucial to driving water efficiencies where there is ferocious competition and demand for this valuable resource. Others forecasts included potential and actual evapotranspiration, soil moisture and reservoir levels.

Where to now?

Users are the experts of their own craft, so it is important to understand the goals from their perspective and not our own. These results have suggested that although some additional datasets would be useful, increased communication around existing forecasts is key to increasing user trust (for river and storage operators at least). While not always appropriate, we encourage researchers to engage with end users’ to really understand their genuine needs.

Work continues on this project. However, initial results have already been presented to the Australian Bureau of Meteorology in the hope for future development. Furthermore, we intend to present further conclusions at the forthcoming Hydrology and Water Resources Symposium in Melbourne, Australia.

Posted in case-studies, forecast communication, forecast users, meetings, water management | Leave a comment

Summary of the 2018 HEPEX ‘Breaking the Barriers’ Workshop, Melbourne Australia.

Contributed by David Robertson, James Bennett, QJ Wang, Daehyok Shin, Andy Wood, Maria-Helena Ramos, Ilias Pechlivanidis and Fredrik Wetterhall.

More than 120 HEPEXers from 15 countries descended on Melbourne, Australia, for three days of sunshine, science and applications at the 2018 HEPEX ‘Breaking the Barriers’ Workshop.

The meeting kicked off with a warm welcome from the local organizing team (James Bennett, QJ Wang, David Robertson and DH Shin) and a series of short talks recognizing the importance of water predictions and science, and from an elder of the Wurundjeri people, who stressed the long history of indigenous peoples living in balance with the land and water. The Workshop was supported by CSIRO, the University of Melbourne and the Bureau of Meteorology.

CSIRO’s James Bennett does the honors in opening up the meeting (left), and Uncle Ron Jones welcomes HEPEX to Wurundjeri country (right)

The participants had the opportunity to listen and interact during the 38 oral presentations (including 3 keynote speeches and 10 invited talks), and 40 posters, whilst the open discussion sessions allowed sharing of experiences and insights. Here, we only select and summarize a few out of the numerous high quality presentations.

Hannah Cloke’s keynote talk “Fly me to the moon” set the stage by reviewing the last decade of progress in flood forecasting for the UK and challenges in taking flood forecasts to a global scale. Hannah above all highlighted the need to ‘be brave’ in trying new approaches, which stayed in the minds of participants for the rest of the workshop.

Subsequent presentations on day one reported on projects assessing the quality of global and regional forecasts of flash floods, riverine flooding, drought and seasonal streamflow generated by land system models harnessing and compiling global impact databases to support verification. Florian Pappenberger highlighted that a greater focus on integrating the land surface as part of Earth System prediction is likely to be the key to extending prediction skill.

A healthy diversity of methods were showcased, including multi-model approaches and data assimilation to improve estimates of modeling and forecast uncertainty, the coupling of coarse land surface models (LSMs) to sub-grid routing and high-resolution hydrologic/hydraulic modeling, and the use of GPU-based computing for forecasting.

The second day of the meeting opened with a description of the EDgE Copernicus proof-of-concept, which is comparing the performance LSMs for seasonal to sub-seasonal (S2S) forecasting in Europe and contrasting skill arising from ESP versus GCM-based predictions. The focus pulled back to methodological questions related to downscaling, merging and pre-processing of weather and climate forecasts, investigating predictability at S2S scales, and highlighting the advancement of dynamical national systems in countries such as New Zealand and China.

Talks also delved into using radar-based rainfall and ensemble approaches to flash flood forecasting. Verification was also discussed, with talks showing new systems for meteorological forecast verification in Australia as well as challenging HEPEX’s hydrologists to think more carefully about what we verify, and how event performance may affect user confidence in forecasts and their effectiveness in motivating action.

The keynote talk of the day was a blockbuster, with Dasarath “Jaya” Jayasuriya, the Director of Public Safety for the Bureau of Meteorology, who dispensed rare wisdom on how forecasting fits within Australia’s national services for managing resources and risk, including insights into forecast-related objectives, mindsets and constraints from the producer side to the public user side. Among other topics, his comments on how BoM navigates the path of serving different users while promoting overall acceptance of forecasts (perhaps through hands on case studies that raise awareness in the field), were illuminating.

Healthy discussions over a healthy lunch: intense days of forecast discussions were balanced by plenty of socializing. The Twitter stream #hepex and the WhatsApp ‘HEPEX in Melbourne’ played a key role in keeping participants in Melbourne and from abroad connected.

HEPEX Day 3 began with a focus on S2S forecasts, with an SMHI effort characterizing predictability (through watershed initial conditions and climate) using collections of geophysical attributes, and work in the UK to understand and enhance climate / drought predictability through the incorporation of NAO variability. The conversation turned toward forecast product development and communication, highlighting the importance of co-development of forecast services with users (see also the SWICCA Copernicus proof-of-concept).

The day’s keynote speaker, Matthew Bethune of the Murray Darling Basin (MDB) Authority, provided a bracing real-world overview of the use of hydrologic models to support decision making in the MDB, highlighting the difficult challenges of making robust release decisions to supply customers at lead times (days to weeks) during which weather, climate and river conditions are highly uncertain. Among the methodological needs raised in the talk, the need to know how climate change may impact current methods for prediction was also raised.

The final talk session of the meeting shifted to examples of real-world predictions for operations, both in flood warning and in hydropower operations for several systems. These talks provide inspiring case studies for effective implementations of ensemble techniques for energy management, underscoring the sense that HEPEX-style forecasting is becoming a reality for groups ready to ‘be brave’ and make the effort to implement ensembles. In addition, the speakers described methodological experiments aimed at finding the best strategies, suggesting that many questions are still of interest. Ensemble research is not a solved problem!

The presentations made available by the authors can be downloaded from here.

Maria-Helena Ramos from Irstea (new Division President-elect of EGU Hydrological Sciences for 2019-2021) announced that she is stepping back from co-chairing HEPEX after 4 years of outstanding leadership and irresistible camaraderie. Ilias Pechlivanidis from SMHI now steps up to fill her shoes, bringing his energy, enthusiasm, and new ideas to help lead HEPEX forward.

HEPEX closed out the meeting with breakout groups, an interactive digital survey, and a closing discussion to take stock of where HEPEX should go. Many aims of HEPEX – including the operational adoption of ensemble hydrologic prediction for the benefit of society – have evolved from being a dream in 2004 to being realized operationally in a number of countries. What then are the key challenges HEPEX should pursue in the next 10 years?

The detailed results of this discussion will be summarized in a future blog, but for now it is clear that challenges do remain (particularly in continuing to communicate the value of ensemble systems), and also many opportunities. This will be an ongoing conversation, so start thinking and contributing – where should HEPEX focus its efforts, what are the big challenges, and how can you help make it happen?

HEPEXers touring the Bureau of Meteorology’s National Operations Centre, where overview of BoM services was presented and the daily weather briefing observed, and Melbourne Water’s Sugarloaf Reservoir, pump station and Winneke water treatment plant, where the challenges of delivering water in a highly variable climate were discussed.

Posted in activities, meetings | Leave a comment

Melbourne HEPEX “Breaking the Barriers” convention – Twitter feed

This week, the HEPEX traveling circus will descend on Melbourne, home of the Bureau of Meteorology, the University of Melbourne, Monash University and the CSIRO. HEPEX-ers will be tweeting their way through the convention – and these Tweets will be assembled in below Twitter stream. Feel free to add your thoughts, ideas and observations – just make sure they’re summarized in 140 280 characters max – and don’t forget to include the #hepex tag!

Posted in Unclassified | Leave a comment

Forecasts of Water Variables – interview with Ilias Pechlivanidis

I can’t exactly remember the first time I met Ilias, but I can remember that every time we have the opportunity to sit together, we end up having great talks about hydrology, modelling and Greece (including its food/drink specialities and beautiful places to visit!).

Ilias is a senior hydrologist at the Swedish Meteorological and Hydrological Institute (SMHI) since 2012. He has broad experience in hydrological modelling and forecasting, and has been recently appointed Scientific leader of the theme “Forecasts of Water Variables” at SMHI’s R&D department.

I had the opportunity to ask him some questions about this new challenge and the future of hydrological forecasting for water managers:

Maria-Helena Ramos (MHR): What is the main focus of the research at SMHI on the forecasting of water variables?

Ilias Pechlivanidis (IP): I would firstly like to thank you, Maria-Helena, for giving me the opportunity to openly share our group’s vision with the broader HEPEX community. Hopefully this blog post will be useful to all friends and colleagues, and particularly inspire the young forecasters.

Going back to your question, I should say that as a service provider at the national, continental and very soon global scale, we produce forecasts of different water (quantity and quality) variables and indicators, such as discharge, soil moisture, groundwater levels, evapotranspiration, snow, sediments etc., at different timescales (at short to seasonal ranges) and hydro-climatic gradients. To ensure usefulness, these forecasts need to be unbiased, reliable and coherent.

We investigate all the components of the hydrological production chain (i.e. selection of meteorological forecasts, pre-processing, hydrological model(s) and setup, calibration and initialization, updating, and generation, evaluation and visualization/communication of forecasts), and evaluate and refine these components for understanding better the sources for predictability. Our efforts focus on identifying the key factors influencing the spatial and temporal variation in hydrological predictability and improving the communication methods bearing also in mind that the user needs are mostly local.

I also find important to note that SMHI is among the very few institutes that acts as a factory and a storage house of many data products, which are further used in our different multi-basin hydrological investigations at the large scale (see our HYPE hydrological model setups).

These spatially broad model applications require a constant interaction with users, in order to learn from their experience of applying our models at the local scale and better comprehend local processes within our work at national to global scales. Despite the challenges of producing and analyzing big data, I consider myself privileged to be given the opportunity to conduct comparative hydrological forecasting analyses and hence complementing the ‘deep’ knowledge from single catchment investigations.

MHR: SMHI has a strong focus on developing forecast products for end-users. In your opinion, is there a “science of hydrological forecasting” and is it currently in phase with operational developments for water (and related risks) management?

IP: Maybe I could say here that a nice lesson learnt working at SMHI’s services is that there is no ‘end-users’ but rather an endless chain of ‘users’.

Indeed, SMHI has been developing various products and services for different users; as an example here for the HEPEX interest, I would point:

  • the Swedish national demonstrator, including from short term to seasonal forecasts, climate predictions and many more,
  • the pan-European short-to-medium range hydrological forecasting service,
  • the pan-European seasonal forecasting demonstrator interface as a proof-of-concept for the Copernicus Climate Change Services (C3S).

I have found very intriguing the different level of understanding between users, and so each service should meet the unique user needs. Nevertheless there is always a continuous need for product and service evolution.

We have been acting as scientific knowledge purveyors incorporating robust new insights and successful outcomes from R&D projects into our products and services. Until recently efforts of the scientific community have been technical and numerical allowing a significant improvement of the services, yet setting unfortunately a knowledge gap between data providers and users.

Despite the deep SMHI in-house knowledge of user needs, application of social science on user engagement has allowed better communication of results and co-evolution of knowledge. We have therefore been investing on co-designing services, providing guidance on ways that our products can address problems at the local and regional scales, and also on ‘teach the teachers’ trainings to ensure that results are adequately communicated.

Another point here is that more and more institutes are nowadays developing similar products and services for different water-related sectors. Although scientific inter-comparisons are very important to improve our process understanding, I believe that different setups have strengths in capturing different aspects of reality; ‘useful information exists in each product/service, so one simply needs to extract it’. To produce the best product for our users, we are currently focusing on operationally implementing multi-model approaches, including different meteorological forecasting systems, pre-processing methods, and/or hydrological models, and further identifying best approaches for multi-model averaging.

MHR: If a young scientist wants to start a PhD thesis today on “forecasting water variables”, what would you suggest as a topic for a 3-year research work, for instance?

IP: As a member of the HEPEX community, I am not sure I could answer this question deterministically. I personally see a lot of potential in different topics, aiming to address fundamental scientific and operational challenges. Without being strict in my selection, I am generally inspired by 3 topics (as I support using 3 arguments in 3 sentences):

  1. Seamless hydrological forecasting; we need to understand how to better bridge between weather, seasonal forecasting and beyond for hydrological purposes, particularly nowadays that integration of available systems (setup for operating at different time horizons and for meeting different types of users) is computationally feasible.
  2. Assimilation of data to advance operational services at the large scale; given that in-situ observations are not usually available in an adequate spatio-temporal resolution, I see the need to explore new Earth Observation technologies and data provided in (near) real time to improve model initializations and hence forecasts.
  3. Hydrological forecasting from an impact perspective for better decision-making; we must bridge the gap between forecast experts and users, driven by the need for a better knowledge/integration of user needs, and highlighting the importance to consider new ways to communicate forecasts and their uncertainty.

MHR: Are there opportunities today for people willing to collaborate with your group?

IP: Absolutely. Our group has been participating in numerous national and international research (and consultancy) projects, with quite major funding coming from the European Commission. This has allowed us to build a strong network with other scientific groups, operational/research institutes, users, and SMEs. Researchers have also been visiting us to work on a targeted challenge. Post-doctoral researchers have also been significantly contributing to our success.

Taking this opportunity, I would kindly like to inform the HEPEX community that we are currently offering a Post-doc contract in the hydrological forecasting scientific theme. For more details please check here.

Thank you, Ilias, for this interview. We are looking forward to hearing more about your challenges and achievements in the future!

Posted in activities, forecast techniques, hydrologic models, interviews, operational systems | 2 Comments

HEPEX 2017 Year in Review

The Hepex Portal published 28 posts in 2017. Here below the year in review, with its highlights.

First of all, two important events for Hepex coming soon:

  • Just a few weeks more to submit your abstract(s) to EGU 2018 (deadline: 10 Jan 2018, 13:00 CET). You can check all sessions proposed under the Hydrological Forecasting sub-division in this Hepex post.
  • The 2018 Hepex Workshop in Melbourne (“Breaking the Barriers”) is coming soon, on 6-8 February 2018. An exciting and intense programme, with 44 oral presentations and 40 posters, is waiting for us. Online registration (no fees) is open until 20 Jan 2018 (check here).

And also blog posts on past events:

  • a post on the event “Ensemble Prediction: past, present and future”, which took place during the ECMWF’s Annual Seminar, from 11-14 September 2017, in Reading (UK), to celebrate 25 years of ensemble weather forecasts. Several interesting presentations made available here.
  • a post on the session “(Ir‑)relevant scales in hydrology: Which scales matter for water resources management?”, convened at this year’s EGU General Assembly 2017.
  • a post on the 15th session of the WMO’s Commission for Hydrology (CHy-15), held in Rome, 7-13 December 2016.
  • a post on the Fourth Pacific Meteorological Council (PMC) and Second Pacific Meteorological Ministers Meeting (PMMM), which was held in Honiara, Solomon Islands, from 14 to 17 August 2017.

Two quizzes and one competition for the readers in 2017:

Publications and more on science and operations:

The HESS special issue on “Sub-seasonal to seasonal hydrological forecasting“, organized after the Hepex workshop in Norrköping, Sweden, in 2015, has gathered 40 papers of excellent quality.

Seasonal forecasts were also the topic of two blog posts, with bias correction (How suitable is quantile mapping for post-processing GCM precipitation forecasts?) and operational systems (Meeting user needs for sub-seasonal streamflow forecasts in Australia) highlighted.

Users were central for two posts on flood forecasting and warning: Community of Users on Secure, Safe and Resilient Societies and Understanding public responses to flood warnings, for instance.

And something to think about concerning the future of ensemble prediction: bridging the gap between hyper-resolution and hydrologic ensemble prediction.

We also published opinions on meteorological prediction and keeping flood memories and historical marks of high waters.

Four new interviews were published in 2017:

All Hepex interviews (15 in total) can be found here.

And, of course, games and decision-making:

A decision making “serious-game”, the Shopkeepers Dilemma Game, and the optimal decision rule were explained by Micha Werner, and an interesting discussion on Risk aversion and decision making using ensemble forecasts was proposed by Marie-Amélie Boucher and Vincent Boucher.

You can also help Hepex to produce more games and teaching material: check here.

Hepex blog portal is running since April 2013: almost five years of a very original and rich content on hydrological forecasting

  • Many more posts and news can be found in our archives and throughout the Portal.
  • For 2018, we invite you all to contribute with your own blog posts (tips can be found here) and to support the organization of Hepex activities and workshops.

Happy holidays! The Portal will be back in 2018!

Posted in activities | Leave a comment

History of Hydrology Wiki – Interview with Keith Beven

Have you seen the wiki “History of Hydrology“? It was created by Professor Keith Beven, Professor Emeritus in the Lancaster Environment Centre, in 2016, and we have recently interviewed him to learn more about it:

What was your motivation behind creating the “History of Hydrology Wiki”?

I have always been interested in the history of the hydrology and hydraulics, which is both long and interesting.

A lot of the early history was summarized in the book on the History of Hydrology, written by Asit K Biswas (North-Holland, 1971) but only up to the end of the 19th Century. There have been a few history studies since, including my own papers on Robert Horton in Journal of Hydrology and Hydrological Processes in 2004, but not a lot.

Both AGU and the International Association of Hydrogeologists have collected some video interviews with eminent hydrologists and hydrogeologists, but clearly there are now very many who we cannot now interview about their work and collaborations.

I remember realizing that I had once met Walter Langbein at a meeting, when he must have been close to retirement. He had worked with Robert Horton but, as a very young hydrologist at that time, I had not realized and certainly had not thought to ask about it. It would be a shame if the history just faded away.

So it is a project I have had in mind for some time, but it is not something that hydrologists are going to get much academic credit for. More of a project for someone who has retired, so I started the site when I took my pension (albeit that I have not exactly retired from research yet).

We have seen that a page was recently posted to Max Adam Kohler, a pioneer in hydrological forecasting. This area of hydrology has long been seen as an “operational” activity. Do you think hydrological forecasting can also be considered as “science” within hydrological sciences?

That is an interesting question (with a long history). Max Kohler, who died only last month at the age of 102, was influential in setting up hydrological forecasting in the US, but also working for the international community as Head of the Hydrological Division at WMO.

As in meteorology, we would want hydrological forecasts to be based on the best available science – but unlike meteorology, hydrological responses are very much dominated by the boundary conditions including catchment characteristics rather than the process representations. In particular in extreme flood events, we cannot be sure that we have a good idea of the volume of inputs from the raingauge and radar data available to the forecaster (and as yet numerical weather prediction cannot provide better estimates). And while we can be pretty sure of observations of water levels, there is a lot of uncertainty associated with estimates of the corresponding discharges. Thus, even the water balance equation may not be that useful in forecasting.

There are certainly cases where there is apparently more discharge than rainfall (and much less than expected when, in mountainous areas, valley bottom raingauges do not reflect that the precipitation falls as snow at higher elevations). Thus in these cases, until we have better data or NWP predictions, data assimilation can be more important as a way of getting improved forecasts  than improved process representations.

Thus in the Lancaster Flood Forecasting Methodology we use a nonlinear transform of the observed inputs, with a linear transfer function, coupled with a simple data assimilation algorithm to produce forecasts and their associated uncertainties. That does not mean that either the nonlinearity or transfer function should not be considered as scientific – see the hypothetico-inductive arguments in Peter Young’s  article in Water Resources Research, 2013.

If you search for ‘ensemble’ on the wiki then one only gets one hit to a reference, whilst there are 8 results when you check for ‘uncertainty’ – are there trends  or fashions one can observe through history?

I think you have to remember that all the biographies (with one exception who will be 100 years old next month) are of hydrologists who have already died, and that running ensembles of models has only been possible since the later part of the 20th Century (my own first Monte Carlo experiments were carried out in 1980 on a CDC “Mainframe” computer. There will certainly be more mention of both model ensembles and uncertainties in the biographies of those who have been active since then. I am sure that trends and fashions will be evident in the material as it accumulates. That, after all, is what constitutes the history of a subject area, especially when (as in hydrology) dramatic innovations occur only rarely.

Finally, how do you think the HEPEX community can contribute to the “History of Hydrology Wiki”?

As a Wiki site, anyone can register to add or edit material on the site (at the moment I am acting as the moderator for entries), or can send me material. There are templates on the site for the different types of entry which can be biographies, histories of experimental catchments, histories of hydrological institutions and details of historical hydrological textbooks (it is somewhat surprising how texts from the early 20th Century look similar to those of the early 21st Century).

I have also recently added a section on annotated papers about the history of hydrology. The hope is that the materials will provide both a record of our history and a starting point for anyone wanting to do more detailed studies. We particularly need entries from the non-English speaking world which is certainly under-represented as yet. There is no reason why entries should not be in another language (see for example the entry for Eugène Belgrand), but it would be good if an English summary can be provided, with a link to the original (see Pierre Cappus).

There will be a session at EGU in Vienna in 2018 convened Okke Batelaan, Roberto Ranzi, Laurent Pfister and myself. With this session we aim to stimulate the discussion on how we, as a community, develop a historical literacy and integrate this in teaching and research to enhance our science. We warmly invite your contributions that discuss how hydrological concepts have gradually evolved over time; how forgotten methods might have contemporary value; the value of historical datasets of experimental catchments and their management; remarkable contributions of scientists, institutes and organisations.   Contributions from the HEPEX Community would be welcome – including (why not) something on the origins and history of the HEPEX project. A more detailed description of the session and the link for abstract submission can be found here. The abstract submission deadline is 10 January 2018, 13:00 CET.

Thank you for this interview and thoughts.

We encourage all Hepex members to contribute to the wiki and EGU session. Sessions on Hydrological Forecasting (science and applications!) will also be held at EGU 2018 (see them in a previous post) and abstract submissions to them are also welcome.

Posted in activities, historical, interviews, opinion | 1 Comment

Hydropower management in Brazil and water forecasts – Interview with Alberto Assis dos Reis

Alberto is an engineer and hydrologist at Cemig, a Brazilian power company headquartered in Belo Horizonte, the capital of the state of Minas Gerais, and is currently starting a PhD work at the Federal University of Minas Gerais (UFMG). His PhD project involves also collaboration with three other organizations in Europe, strongly involved in Hepex: Irstea (France), Deltares (The Netherlands) and ECMWF (UK). He was recently visiting these organizations and, when he came to France, I took the opportunity to ask him some questions:

Maria-Helena Ramos: In a few words, how do you describe the Brazilian electric system?

Alberto Assis dos Reis: Currently, the system has an installed capacity of about 152 GW, the largest in South America, and it is essentially hydrothermal, with a great part of hydraulic generation (nearly 65% of domestic supply). In fact, Brazil has an electric matrix that is predominantly of renewable sources (74.6% of the electricity of domestic supply), including an important share of biomass (from sugar cane) and wind power. By 2024, it is expected that hydropower and biomass relative part will slightly decrease, although they will remain important sources of power generation. Wind power and solar power are expected to increase their part in the Brazilian electricity generation in the near future.

Today, the system is regulated by the “Operator of the National Electricity System” (ONS), which is a non-profit private entity created in 1998. The ONS is responsible for the coordination and control of the generation and transmission installations in the National Interconnected System (SIN). The SIN comprises the electricity companies in the South, South-East, Center-West, North-East and part of the North region, including Cemig, the company I have been working for since 2002.

MHR: How important are streamflow forecasts to the management of hydropower production in Brazil?

AAR: Brazil has a large capacity for water storage; I think it is one of the greatest, if not the greatest, in the world. The inflows to hydroelectric plants have a considerable weight in the planning of the operation of the electrical system, as well as a large weight in the energy price setting in the short term market, the SDP – Settlement of Differences Price.

The calculation of the SDP is based on what we call MCO – Marginal Cost of Operation, which is limited by a minimum and a maximum price, both set annually by the NEEA – National Electric Energy Agency (in Portuguese, Agência Nacional de Energia Elétrica, ANEEL). The MCO is obtained from computer models run by the ONS. These models optimize the system’s operation, solving the hydrothermal dispatch problem in Brazil.

These models run once a week, every Thursday, with flow forecasts. Forecasts are issued on a weekly basis for the first five weeks and on a monthly basis for the next months. Currently, hydrologic models are used for the first week, ARMA models are used for the following weeks, and synthetic flow series are generated for up to 5 years, based on historic flow records from 1931 to today.

Streamflow flow forecasts and reservoir levels are variables with a strong role as price definers. It is estimated that they account for about 60% of the price definition. It is thus important to have hydrologic forecasts of good quality when taking decisions on selling and buying energy in the market.

MHR: Why have you decided to do a PhD work on ensemble forecasting?

AAR: It is a long story! At Cemig, we have been working on the development of robust seasonal forecasting system for several years. Our aim is to have an efficient operation of the hydropower plants with state-of-the-art products. We have been using medium-range ensemble weather forecasts (up to 15 days) and the ESP (Ensemble Streamflow Prediction) technique with a hydrological model operated in house, calibrated for a large set of catchments in Brazil.

As you know, the operationalization of hydrologic ensemble modeling is a fairly laborious activity to perform manually, because it means repeating several times the data collection activities, the preparation of files for models and their application in hydrological models. We have thus put efforts into the development of tools to automate this process. Since 2014, in collaboration with Deltares and LACTEC, we have the FEWS-Cemig system running, now in operational mode, over 40 operational centers of water forecast.

I have been involved in this project since the beginning and now that everything is running more or less automatically I have more time to spend on experting the forecasts. This means that I have more ideas on how to improve them! I am curious to know if we can improve our systems and make even better forecasts. I am also interested in running a seamless forecasting system, from short- to long-range forecasts in a coherent, and unique, framework.

That’s why I decided to do a PhD work: to explore further techniques to improve the quality of the forecasts in a seamless system and evaluate its impact on the management of hydropower reservoirs. I think this can be a great advantage for us, at Cemig, opening up a range of potential applications as a support tool for decision in the Brazilian scenario of water modeling.

Thank you, Alberto, for this interview and all the best for this exciting PhD work!

Posted in decision making, forecast users, interviews, operational systems, seasonal prediction, water management | Leave a comment

Hydrological Forecasting at EGU 2018: time to write your abstract

You can contribute to advance hydrological predictions and forecasting systems through the presentation of your recent scientific developments, applications and approaches in the operation of hydrologic forecasting systems at the EGU Assembly in 2018.

Why should I go to EGU next year?

  1. Vienna is a beautiful city
  2. We all have a good time in the PICO and Poster sessions
  3. Hepexers always go out and it can be a lot of fun
  4. The oral sessions are a great opportunity to communicate my work
  5. I don’t know, but something tells me I should be there


This year, the sub-division on Hydrological Forecasting, under the umbrella of the Division on Hydrological Sciences, is organizing seven sessions and co-organizing another two. Abstract submission is open until 10 January 2018, 13:00 CET:

Flash floods and associated hydro-geomorphic processes: observation, modelling and warning

[read more]

Convener: Isabelle Braud | Co-Conveners: Marcel Hürlimann, Marco Borga, Jonathan Gourley, Massimiliano Zappa, Jose Agustin Brena Naranjo
Predictive uncertainty estimation and data assimilation for hydrological forecasting and decision making

[read more]

Convener: Oldrich Rakovec  | Co-Conveners: Albrecht Weerts, Hamid Moradkhani, Marie-Amélie Boucher, Rodolfo Alvarado Montero, Joshua K. Roundy
Ensemble hydro-meteorological forecasting

[read more]

Convener: Fredrik Wetterhall  | Co-Conveners: Tomasz Niedzielski, Maria-Helena Ramos, Jan Verkade, Kolbjorn Engeland, Rebecca Emerton
Drought and water scarcity: monitoring, modelling and forecasting to improve hydro-meteorological risk management

[read more]

Convener: Brunella Bonaccorso  | Co-Conveners: Athanasios Loukas, Christel Prudhomme, Micha Werner, Carmelo Cammalleri
Operational forecasting and warning systems for natural hazards: challenges and innovation

[read more]

Convener: Michael Cranston  | Co-Conveners: Jan Szolgay, Ilias Pechlivanidis, Femke Davids
From sub-seasonal forecasting to climate projections: predicting hydrologic extremes and servicing water managers

[read more]

Convener: Louise Crochemore  | Co-Conveners: Henning Rust, Bart van den Hurk, Christopher White, Johannes Hunink, Tim aus der Beek, Louise Arnal
From probabilities to preparedness: early action in response to hazard forecasts

[read more]

Convener: Gabriela Guimarães Nobre  | Co-Conveners: Konstantinos Bischiniotis, Erin Coughlan de Perez, Brenden Jongman, Liz Stephens, Bart van den Hurk
Advances in statistical post-processing for deterministic and ensemble forecasts (main organization by NP Division)

[read more]

Convener: Stéphane Vannitsem  | Co-Conveners: Jakob W. Messner, Daniel S. Wilks
Uncertainty quantification in natural hazard and risk assessments: best practices and lessons learned across different hazards (main organization by NH Division)

[read more]

Convener: Paolo Frattini  | Co-Conveners: Daniela Molinari, Yoshiyuki Kaneda, Ivica Vilibic, Sergiy Vorogushyn

Check the full program of the Division on Hydrological Sciences for more sessions.


  • If you are eligible to apply for financial support and want to take this opportunity, you need to submit an abstract by 1 December 2017.
  • Otherwise, the deadline for abstract submission is 10 January 2018. Don’t miss it!

Detailed information on how to submit an abstract can be found here.

Posted in activities, announcements-events, meetings | Leave a comment

Why are meteorologists apprehensive of ensemble forecasts?

Contributed by Anders Persson, Uppsala, Sweden

A colleague in my world-wide meteorological network made me aware of a CALMet conference in Melbourne, i.e. dealing with meteorological education and training. Through the website you can access the program with more or less extensive abstracts. I have no doubt that most presentations were relevant and interesting, but what surprised me was that a search for the key words “probability” or “ensemble”  gave no hits. “Uncertainty” came up in only one (1) presentation, no 36 “To communicate forecast uncertainty by visualized product” by Jen-Wei Liu and Kuo-Chen Lu from the Central Weather Bureau in Taiwan.

This made me again ponder over the question why meteorologists still are apprehensive of ensemble systems (ENS) and probability forecasting.

1. Ensemble forecasting brings statistics into weather forecasting

Since the start of weather forecasting as we know it (in the 1860s), there has always been a rivalry between physical-dynamic-synoptic and statistical methods. Edward Lorenz’s famous 1959 experiment when he discovered the “butterfly effect” was part of a project in the late 1950’s to find out if statistical methods could be as effective in weather forecasting as numerical techniques. The answer was at the time not as clear-cut, but during the 1960’s, the numerical weather prediction (NWP) made much larger advances than the statistical approaches. Statistical methods were thereafter only used to calibrate NWP in what became known as MOS (model output statistics).

Over a lunch at ECMWF Edward Lorenz, on one of his annual visits in the 1990s, told us a parable he had got from the renowned Norwegian meteorologist Arnt Eliassen:

All the world’s birds wanted to compete who could fly the highest. They all set off ascending, but one after the other they had to drop out. Finally, only the great golden eagle was left. But as also he had to stop in order to return, a little sparrow who had been hiding in his feathers came out and managed to beat the eagle by a meter or two. The eagle is the dynamic NWP Eliassen had told Lorenz (who told us), the sparrow the statistical MOS.

To some extent the MOS can deal with uncertainties, but in a limited way since it is based on a deterministic forecast. It can estimate the general uncertainty at a certain range, but not distinguish between more or less predictable flow patterns. This is the strength, the core value of the ENS.

But ensemble forecasts are essentially statistical, probabilistic, and the meteorological education have always avoided to venture into this domain, except for those who wanted to become climatologists which in the old days was looked down upon. The ideal has been a physical-dynamic “Newtonian” approach, where perfect or almost perfect forecasts were seen as possible, if only the meteorological community got enough money to purchase better computers.

Indeed, it has paid of; the predictability range has increased by about one day per decade. Our five day deterministic forecasts today are as good and detailed as the two day forecasts in the 1980s. But also the demands and expectations from the public has increased. Even if we in a few decades from now can make more accurate and detailed seven day forecasts, there will still be questions about their reliability. The problem of uncertainty estimations will always be with us.

2. The ensemble system is a Bayesian system

But also among those meteorologists who are used to statistics, there is another problem. I became aware of that when I traveled on behalf of ECMWF to different Member States. A frequent question was: -How can you compute probabilities from those 50 members when you are not sure that they are equally likely?

My answer then was that we did not know! We did not know the likelihood of every member and we didn’t even know if they were all equally likely (probably they were not). But the verification statistics were good, and they would not have been so good if our assumption had been utterly wrong.

A typical “postage stamp map” from the ECMWF system. These 50 forecasts are not a priori equally likely, but since we do not know the probability of each of them we have to apply Laplace “Principle of insufficient reason” and assume that they are equally likely – an assumption which makes the system Bayesian. Image courtesy of ECMWF.

Only later I was made aware that my answer was the same as Siméon de Laplace had given two centuries earlier, when he was developing what is today known as “Bayesian statistics”: – We do not know, but make a qualified guess and see how it works out. Bayesian statistics, in contrast to traditional “frequentist” statistics, acknowledges the usefulness of subjective probabilities, degrees of belief. Laplace’s answer, which I unknowingly resorted to during my ECMWF days, is known as “Laplace principle of indifference”.

So part of the apprehension to ensemble forecasting cannot be attributed to ignorance, conservatism or “Newtonianism”, but has its basis in a long standing feud between “Bayesian” and “frequentist” statisticians. A “Bayesian” can look at the sky and say “there is a 20% risk of rain” whereas a frequentist would not dare to say that unless he had a diary which showed that in 34 cases out of 170 with similar sky, wind and pressure rain has occurred.

In recent years the gulf between “frequentist” and “Bayesians” has narrowed. Also, the calibration of the ENS data “à la MOS”, “washed away” much of the Bayesian characteristics and provided a more “frequentist” forecast product.

3. What is left for the forecaster?

Bayesian methods should not be alien to experienced weather forecasters. Since weather forecasting started in the 1860s there has been a strong Bayesian element in the routines, perhaps not described as such, but never the less this is how forecasters worked before the NWP. Who else but an experienced forecaster could look at the sky and give a probability estimate of rain? If the forecaster had a weather map to look at, the estimation would be even more accurate. Verification studies in the pre-NWP days in the 1950’s showed that forecasters had a good “intuitive” grasp of probabilities.

But with the advent of deterministic NWP the “unconscious” Bayesianism among weather forecasters evaporated gradually. The NWP could tell very confidently that in 72 hours time it would be +20.7 C, WSW 8.3 m/s and rain 12.4 mm within the coming six hours?

Anybody could read that information, your didn’t need to be a meteorologist. But you needed to be a meteorologist to have an opinion about the quality of the forecast: -Would it perhaps be cooler? The wind weaker? How likely is the rain?

There are currently more weather forecasters around than at any time before, in particular in the private sector where advising customers about their decision making is an important task (Photo from a training course at Meteo Group, Wageningen. Permission to use by Robert Muerau)

The risk was always that this forecast, even against the odds, would verify. So wasn’t it most tactical to accept the NWP? After all, if the forecast was wrong, the meteorologist had something to put his blame on. Some meteorologist took this easy road, but most tried to use their experience, knowledge of the models and meteorological know-how, to make a sensible modification of the NWP, including the reliability of the forecast. If the last NWP runs had been “jumpy” and/or there were large divergences among the available models.Tthis was taken as a sign of unreliability.

The “problem” for the weather forecasters was that with the arrival of the ENS they were deprived of even this chance to show their skill. The “problem” with a meteogram from ENS, compared to a more traditional deterministic from a NWP model, was that “anybody” could read the ENS meteogram! You didn’t need to be a meteorologist, not even a mathematically educated scientist. Einstein’s famous “grandma” could read the weather forecast and understand its reliability!

“You do not really understand something unless you can explain it to your grandmother.” – Albert Einstein

So what is left for the meteorologist?

I will stop here, because this text is already long enough. But the question above is really what educational and training seminars, conferences and workshops should be more focused on. I am personally convinced that the meteorologists have a role to play.

My conviction is based on my experiences from the hydrological forecast community, in particular the existence of this site. Is there any corresponding “Mepex“?

My conviction is also based on my experience as a forecaster myself, how the general public (and not so few scientists) need help to relate the uncertainty information to their decision making.

My conviction is finally based on my experiences from history that new tools always make traditional craftsmen more effective and prosperous – provided they are clever enough to see the new opportunities. Else they will miss the bus . . .

PS. To their credit it must be mentioned that EuMetCal is developing training resources for probabilistic forecasting. Ds.
All images from Thinkstock if not otherwise written.
Posted in ensemble techniques, forecast communication, opinion | 2 Comments

Ensemble prediction: past, present and future

Contributed by Fredrik Wetterhall and Roberto Buizza, ECMWF

The work of producing meteorological ensemble forecasts started 25 years ago at ECMWF and NCEP, and it sparked a revolution in both weather forecasts and its many applications. To celebrate this occasion, more than 100 people from across the world joined the 28 speakers at ECMWF’s Annual Seminar 11-14 September held in Reading, UK. The theme was “Ensemble prediction: past, present and future” and the four days where filled with presentations and discussions on what has been done, where we are and how we in the future can further improve the accuracy and reliability of ensemble-based forecasts.

Thanks to advances in models, data assimilation schemes and the methods used to simulate initial and model uncertainties, today ensembles are widely used to provide a reliable estimate of possible future scenarios. This is expressed for example in terms of probabilities of weather events or of risk indices. Increasingly, ensembles are routinely used to provide forecasters and users with the range of weather scenarios that could happen in the future. An example is given by the ECMWF ensemble-based strike probability of hurricane Irma, issued by ECMWF on 5 September.

The ECMWF ensemble-based strike probability that hurricane Irma would pass within a 120 km radius during the next 10 days, issued on the 5th of September (left panel).

Using ensemble forecasts

Different aspects of ensemble forecasting were discussed during the seminar, and they included the history and theory of ensemble forecasting, initial conditions, model uncertainties, error growth, predictability across scales, verification and diagnostics and future outlook. The full programme including recordings of the talks can be found here. The theme that may be of most interest for the HEPEX community was devoted to applications of ensemble forecasts. The session discussed the various ensemble products that now exist to help decision making (David Richardson, ECMWF), hydrological ensembles including the HEPEX experience (Hannah Cloke, Reading University) and observing and supporting the growing use of ensemble products (Renate Hagedorn, DWD). The session was testament as to how mainstream ensemble forecasts have become, not only in science but also in institutions and authorities that use probabilistic information in decision-making. There is still a lot to do to overcome some of the existing barriers, but the acceptance of ensemble forecast is truly a success story.

Panel discussions and looking forward

The seminar also included a panel discussion which provided an opportunity to explore and discuss in more detail some of the fundamental questions that are currently being tackled by the community, such as:

  • Should we be moving to small ensembles at high resolution, or large ensembles at more moderate resolution?
  • If the most cost-effective ensemble structure changes with lead time, should our ensemble be built so as to give a resolution and ensemble size that changes with lead time?
  • If an ideal ensemble consists of a set of equally likely members, is there a role for an unperturbed/central forecast?
  • What do we expect from the future in terms of our ability to represent model error in ensemble systems, and the representation of perturbations more generally?

It can be interesting to report some of the comments raised during the lively panel discussion:

  • Some users would react also at small probabilities: they would be the ones benefiting more from a size increase;
  • Ensemble size very is important both for the extended/long ranges and for high-resolution ensembles, to be able to capture the fine-scale details;
  • Considering the range of users of the ECMWF ensembles, overall, a size of 50 seems about right; although ECMWF principal aim should be to provide the best raw ensemble forecasts, it should work with the users to develop calibration methods, and understand whether the balance between ensemble size and resolution should be revisited once calibration methods are more widely used;
  • ECMWF should aim to provide the national meteorological services and its users with ensemble-based probabilistic forecasts that could be used by a wide range of users; it will be then up to the national meteorological services and/or third parties to design ‘tailored’ ensemble configurations that can address the needs of specific users;
  • We need more observation-based diagnostic to understand model error, and design better schemes;

Participants of the ECMWF Annual Seminar 2017. Photo: Simon Witter, ECMWF

The HEPEX community was an early advocator of using ensemble forecasts and it is important that we continue to push the boundaries of how ensembles should be used in research and applications. A good way of doing just that is to come to the HEPEX workshop in Melbourne next year!

Posted in activities, data assimilation, ensemble techniques, forecast techniques, forecast users, historical, meetings, operational systems, verification | 1 Comment