Skip to content

AI-Translated Page

This page was translated using artificial intelligence (AI). The content may contain inaccuracies or misinterpretations. For critical or official use, please consult the site administrator to confirm the information.

May 5, 2021

MINUTES 001/2021 - Revision 004

Meeting of the Scientific Committee of the Community Model of the Unified Earth System. At 3:00 pm on May 5, 2021, representatives of INPE (National Institute for Space Research), INMET (National Institute of Meteorology), UFCG (Federal University of Campina Grande), CENSIPAM (Management and Operational Center of the Amazon Protection System), ITA (Technological Institute of Aeronautics), INPA (National Institute for Amazon Research), UFSM (Federal University of Santa Maria), USP (University of São Paulo), LNCC (National Laboratory for Scientific Computing) and UFMS (Federal University of Mato Grosso do Sul) met virtually, with the aim of presenting and initiating the work of the Scientific Committee of the Community Model of the Unified Earth System (MCSTU). These minutes record the meeting and gather the information entered in the chat, such as links and information relevant to the discussions held. Following the opening of the meeting, carried out by Saulo Freitas and Pedro Dias, this document is oriented according to the guidelines established by Saulo Freitas during his presentation.

Opening

Saulo Freitas and Pedro Dias greet all guests and recall that the first meeting of this committee took place on April 8, 2021, when the Director of INPE appointed the members of INPE who make up the Scientific Committee of the Community Model of the Unified Earth System.

Agenda 1 - Introduction of MCSTU

Gilvan Sampaio (General Coordinator of CGCT/INPE) thanks everyone for accepting the invitation, given the challenge of creating a unified model. He highlights the presence of members of CENSIPAM and INMET , but emphasizes that these two institutions are also stakeholders in this community effort. He also appreciates the presence of Fernando II. Regarding the community model, he emphasizes that the objective of the committee - which is scientific - is to debate science, discuss and manage the development of the community model. The decisions that are made must be scientific and have no direct relation to the National Meteorological System. The committee must make decisions based on scientific criteria. These discussions are important, as they will be presented to the other working groups. He emphasizes that this is a very important step for national meteorology. On the part of INPE , represented by the Director, who is committed to this process, the members of the scientific committee will have the support of CGCT (General Coordination of Earth Sciences) and the INPE Board of Directors, whose roles are as facilitators.

Clezio de Nardin (Director of INPE) thanks everyone for their presence, especially the external members, and reiterates INPE's support for the process that is being established. He informs that he will always participate, but that he would like his presence to be requested when necessary.

Saulo Freitas reiterates that the Director's participation is always welcome. This is not a trivial task and the support of the Director of INPE is important. He emphasizes that at some point, the participation of the Director of INMET and the directors and vice-rectors of the participating Brazilian universities will also be pertinent. He states that this activity transcends the walls of INPE, and that the development of the community model of the unified Earth system is a model from Brazil to Brazil.

Saulo Freitas thanks the Director, introduces the scientific committee rapporteur Carlos Bastarz and Fabielle Mota as secretarial support.

Agenda 2 - History of INPE's initiatives to develop a unified modeling system

Pedro Dias summarizes the historical facts in the context of model developments in Brazil and what happened at CPTEC . A historical review of the beginning of atmospheric modeling in Brazil until the establishment of CPTEC. The 1980s were marked by research with regional models, simplified global spectral models, as well as oceanic and atmospheric pollution models, also with a focus on research. At that time, it was found that Brazil had the minimum competence to enter the operational area, through LACCAS (Latin American and Caribbean Center for Atmospheric Sciences). The CPTEC project and the choice of a model: global, from COLA and regional, from RAMS (currently BRAMS , with Brazilian developments), ETA (also with Brazilian developments). The choice of ETA was due to the use of the appropriate coordinate for mountainous and abrupt regions. In the 1990s, global and regional models began to operate, as well as ensemble weather forecasting and some improvements in model performance (in collaboration with Jairo Panetta, Saulo Barros, Pedro Dias and other collaborators). The first experiments with data assimilation (in collaboration with JMA). In the 2000s, the coupling between the atmosphere and the ocean occurred (Paulo Nobre and team), and the forecast was extended to 30 days. CPTEC's participation in TIGGE was essential.(providing forecasts over 10 days). In 2009, the Santa Rita do Passa Quatro workshop was held, with the aim of outlining a vision for the future of modeling. Pedro Dias states that, curiously, many of the questions that were asked at that time remain the same today. There have been advances in research, but little progress has been made operationally. From 2011 onwards, the impact of the reduction in financial resources was felt. Some priority questions that arise are: why make global forecasts? Is it strategic? Can't we just use global forecasts from other centers to initialize regional forecasts? Should we or shouldn't we have independence in generating forecast products? Is it correct to depend on forecasts from foreign agencies? Global models are necessary for seasonal climate forecasts. All actors in the meteorological system run regional models. Is all this effort necessary? Who is in charge of developing these models? Only CPTEC (including data assimilation) is fully aware. CPTEC had and has knowledge of the ins and outs of this process. Running models "gives status", but it is important to develop products for specific users. The failure was partly due to management problems in the national meteorological system. This resulted in technical delays, mainly in updating supercomputers, taking advantage of new architectures, using massive parallelism, using co-processors, and seamless prediction (multi-scale, multi-physics vision). The challenge lies in adjusting the models to our reality; in bringing together people from different specialties; in having goals such as internationally competitive products (we need to stop being users); in placing greater emphasis on essential processes and greater influence in the tropics. We need to recognize that we have fallen behind. Part of the problem was financial and the other part was coordination between the actors in the national meteorological system. It is necessary to develop stable partnerships. We need to have multiple partners. In the chat, Ronald Buss questions why the focus should be on the tropical region, given that the country is continental and one of the problems is the relatively low predictability in the southeast region, which is precisely affected by phenomena originating in the subtropics and ocean, which depend on remote links, including with Antarctica. Pedro Dias argues that the goal is to have the best forecast for South America and its surroundings; the oceans adjacent to South America are essential, but the evaluation metric should be the improvement of forecasts for Brazil. In the chat, Pedro Dias adds that if a remote phenomenon has an impact on improving the forecast, it should be included.

Gilvan Sampaio corroborates Pedro Dias' statement and adds that in 2016, some meetings were held at CPTEC, where discussions on a unified model were initiated. Unfortunately, it did not move forward due to several changes at the center. Nevertheless, this focus was maintained, as an agreement was signed with NOAA and ECMWF. As mentioned by Pedro Dias, in 2017, 2018 and 2019, discussions with NOAA continued. In 2019, the agreement with NOAA will be reviewed and reformulated. Historically, these agreements have been little explored. With NOAA, the agreement was related to satellites and currently it will include modeling.

Agenda 3 - Presentation of the Initial Proposal for Scientific Activities within the scope of MCSTU

Paulo Kubota presents an initial proposal of scientific topics to be addressed within the scope of the MCSTU. The objective is to foster the development base of the MCSTU. The initial proposal considers the holding of lectures and round tables on the scientific topics "data assimilation and terrestrial observation systems", "Earth system modeling", "dynamic cores", "high-performance processing", "coupling technologies" and "general topics". Although there is still no initial schedule (it depends on the agendas of the CGCT Coordinator and the INPE Board), the scientific activities are expected to last 5 weeks, with each week dedicated to specific topics. Two or three lectures with a round table are planned per day and the scientific activities committee should meet at the end of the week to prepare a report on the topic of the week, collecting the most important information presented and the results of the discussions. This report should be submitted to the MCSTU scientific committee. Paulo Kubota emphasizes that the committee is open to suggestions from members of the MCSTU scientific committee and that the document with the initial proposal will be submitted for consideration, review and collection of suggestions from the committee members. Saulo Freitas informs that the proposal will be decided at the next meeting of the scientific committee, already with the contributions of the members.

Pedro Dias emphasizes that, when it comes to data assimilation, one should not only consider the atmosphere. JEDI (Joint Effort for Data Assimilation Integration) is an important system because it is related to the Earth system as a whole. One should think of a methodology that applies to all systems in the Earth system.

Jairo Panetta comments on the software structure, which must include the global and regional aspects, coupling, and the dynamic insertion of variables from fields of other systems in the Earth system. He asks Paulo Kubota whether these aspects are included in any research topic and suggests combining these aspects with high-performance processing. This is related to the way in which dynamics, physics, etc. are designed.

Pedro Peixoto made two comments on the topics presented by Paulo Kubota, in the context of the statements made by Jairo Panetta. 1. Computational structure is very important and he emphasizes that MetOffice and ECMWF have invested heavily in this. Examples of this are ESCAPE (Energy-efficient Scalable Algorithms for Weather Prediction at Exascale) and GungHo!, which are projects focused on these aspects. In this same context, Pedro Peixoto indicates in the chat the link to an article about the software infrastructure designed by MetOffice, called LFRic (an allusion to scientist Lewis Fry Richardson). It is necessary to design a platform flexible enough to accommodate all of this and this is a very intense software development work. 2. The connection between these aspects. Some decisions have interdependencies, and it is necessary to think about how all groups can benefit. It is not just about coupling between the components, but how to choose them. It illustrates that within the atmospheric system, when optimizations are included among others, it cannot be good for just one component, but has to be good for all other components of the Earth system. Saulo Freitas mentions the name of the researcher Peter Bauer from ECMWF and has the agreement of Pedro Peixoto, as a person to collaborate in the establishment of these aspects.

Haroldo Fraga discusses software refactoring and questions how physics codes already produced can be reused in a new dynamic core. He states that the feasibility of this should be discussed, in recovering what has already been done, as this can help in decision-making. The OLAN (Ocean Land Atmosphere Model), for example, can be recovered, as some time was invested in this project and it is not being used. Would it be a case of reusing what has already been done? He also highlights that the evaluation metric should be the best performance in rain forecasting over Brazil.

Roberto Souto adds to the topic of software refactoring, and proposes the name of researcher Tadeu Gomes from LNCC for the topic of "High Performance Processing" in the lecture and round table series. He mentions that, although he does not work with weather forecasting, he is a software engineer and focuses on scientific computing.

Luiz Flávio emphasizes that the document with the initial proposal for scientific activities was created to receive contributions from everyone, which is essential. He also emphasizes that it concerns the organization of the developments to be carried out. This topic can be more specific to the computing team, and computing is a cross-cutting topic. It is necessary to standardize the software development methodology. With a small team, it is already difficult to control software. With a larger team, it will be even more difficult. If a standard is not established, there is a risk of not being able to develop the software, which would not happen with a larger team. In the context of software engineering, Haroldo Fraga points to the transition from the MM5 model to the WRF as an example of success.

Pedro Dias states that the points raised are fundamental and that this is what will provide reliability and longevity in software development.

Luiz Cândido asks about the scale that the model is intended to be developed on, since he realizes that topics such as cloud microphysics and other processes have not been mentioned. Saulo Freitas reiterates that these topics will be discussed. Physical parameterizations that are suitable for various scales (multiscale and multiphysics) will be necessary. This is an ambitious goal, but one that must be pursued. Pedro Dias states that Brazil has expertise in these areas and that many people can contribute.

Agenda 4 - Definition of committee members to collect and document MCTSU requirements, demands, and counterparts

Saulo Freitas comments on the definition of the committee members, as a zero-order approximation. The people appointed will be focal points and will be responsible for creating collaboration subnetworks with other partners from the national and international community (stakeholders). Documents should be formatted to include the requirements for physics, dynamics, computational grid, coupling technologies, PAD (High Performance Processing), data assimilation methods, etc. The demands are the applications, the problems to be addressed, the necessary performance and the minimum accuracy. The counterparts are what the Brazilian community can offer to the partners and international institutions. What can the community benefit from the development of the community model? What are the infrastructure demands, the resources for computing equipment and observation of the Earth system? The material produced by the focal points and the subnetworks will be the basis for the work. The focal points should meet with the regional development and research groups to form the subnetworks that should be integrated into the scope of the work. The focal points initially appointed are:

  1. Integrated Modeling System: Pedro Dias/USP
  2. Atmosphere: Saulo Freitas/INPE
  3. Surface and Continental Soils: Antonio Manzi/INPE
  4. Oceans and Continental and Sea Ice: Ronald Buss/INPE
  5. Space Weather: Joaquim Costa/INPE
  6. High Performance Processing and Code Quality: Luiz Flávio/INPE
  7. Earth System Data Assimilation: João Gerd/INPE
  8. Advanced Data Assimilation Methods and Artificial Intelligence Applications: Haroldo Fraga/INPE
  9. Universities of the North/Northeast: Ênio Souza/UFCG
  10. Southeast Universities and Research Institutes: Pedro Dias/USP
  11. Southern Universities: Otávio Acevedo/UFSM
  12. Universities of the Midwest: Vinicius Capistrano/UFMS
  13. CENSIPAM: Ivan Saraiva
  14. INMET: Francisco Quixaba
  15. Suggestions? What's missing?

Ronald Buss, via chat, points out that a better name for the topic "Oceans and Continental and Sea Ice" is "Oceans and Cryosphere". He argues that "cryosphere = sea ice, polar ice caps and third pole (icy mountains) - disregarding permafrost for now". Pedro Dias points out that there is a lack of a focal point for the topic "evaluation". He questions whether this topic should not be considered. The evaluation metric depends a lot on the user's needs. For the energy sector, rainfall (as Haroldo mentioned) may not be the most appropriate metric; it could be wind, radiation, etc. Stakeholders have different needs (agribusiness, the National Water Agency , etc., large users of INMET).

Saulo Freitas agrees and suggests that the committee look for a focal point that can fulfill this coordination and asks who can make this approach, asking the stakeholders what their needs are regarding the evaluation of the system to be developed.

Francisco Quixaba reports that with COSMOS , INMET has served some users (e.g., agribusiness). He questions the absence of the Navy, which has also processed numerical models.

Pedro Dias questions whether CENSIPAM does not encompass this sector; Gilvan Sampaio states that it does, that CENSIPAM is the body that represents the armed forces.

Ivan Saraiva informs that, at this moment, he does not have a certain answer about the Navy's representation in CENSIPAM, but that he will check about this representation.

Gilvan Sampaio points out that Caio Coelho can contribute, as he is a member of the S2S (Subseasonal to Seasonal) evaluation group, and believes he should be included as a representative.

Enio Souza points out that when the focal points were divided, the North and Northeast were combined. He argues that they are two different regions with different groups and stakeholders. He suggests separating the regions, with Luis Cândido taking care of the North and him (Enio Souza) taking care of the Northeast.

Ronald Buss argues that there is an imbalance between the people in the area of ​​atmosphere and oceans and ice. He points out that the scientific community on ocean and ice is poorly represented and that he would like to point out that the physical oceanography community is very large and that there is a lack of people from this area on the MCSTU scientific committee. He argues that he cannot and does not feel safe representing this group alone. The REMO network (Oceanographic Modeling and Observation Network) is an established network and there are some obstacles to this group providing products to society. He says that he expects the community to be involved and represented in the development of the model, and that the model should work both below and above water.

Pedro Dias agrees with Ronald Buss and says that the REMO network is fundamental and this partnership must be institutionalized.

Saulo Freitas reiterates that this is indeed the first meeting. Everyone will be welcomed and there is no "seat limit". The scientific committee is in its first stages, is being structured and the people indicated by Ronald Buss will be invited.

Agenda 5 - INPE-ECMWF Agreement in the Context of MCSTU

Saulo Freitas introduces the INPE-ECMWF agreement, in the context of the community model. He invites Fernando II and Caio Coelho to present.

Fernando II listed the possible topics for collaboration with ECMWF. He pointed out that one of the reasons for ECMWF's success is that it has a single objective, which is to develop the medium- to long-term forecast model. All of the agreements that ECMWF makes are aimed at improving the model itself. ECMWF has an adequate and up-to-date structure for its operational and development activities. Cooperation between INPE and ECMWF has been going on since 1988 and the areas of cooperation are reviewed every year. ECMWF can cooperate with INPE in the context of developing the community model, with lectures and workshops on modeling, and training. In the agreement with ECMWF, there is always a "free chair" for any course. There is an agreement that includes modeling and Copernicus ( C3S - Copernicus Climate Change Service for atmospheric monitoring, climate change, etc.), artificial intelligence, scalability, OpenIFS programme (ECMWF model, open for research and teaching - the previous version and without data assimilation), cloud computing - European Weather Cloud and ClimetLab - access software for meteorological datasets.

Pedro Dias requests clarification on the possibility of people outside INPE participating in the training offered by ECMWF, within the scope of the INPE agreement. Caio Coelho and Fernando II point out that it is possible, with INPE's consent.

Saulo Freitas states that the courses can be taken online. He highlights that OpenIFS physics can be used as a basis for developing the community model, ECMWF analyses for validation, and MetView for viewing unstructured grids.

Haroldo Fraga argues that it is important to negotiate a specific work plan for the community model within the INPE-ECMWF agreement.

Fernando II points out that the INPE-ECMWF agreement is renewed every 5 years and that the work plans are annual.

Caio Coelho adds that ECMWF's focus is also only on forecasting between a few days and a few months (extended forecasting - up to 10 days, sub-seasonal - up to 5 weeks, seasonal - up to 6 months, using 50% of the computational resources available for research activities to develop the model for these three scales). They do not do nowcasting or climate change projections. Their contribution to climate change projections is to provide the tools for storing and accessing projections produced by other centers for use by the community in general.

Other Matters

Haroldo Fraga, via chat, suggests using the RNP (National Education and Research Network) Conferênciaweb platform to record meetings.

During Gilvan Sampaio's speech, after Pedro Dias' presentation in the context of "Agenda 2 - History of INPE's initiatives to develop a unified modeling system", Francisco Quixaba questions Pedro Dias about the INMET-CPTEC agreement; this discussion can be seen in the recording chat from 58'46" .

During discussions about the computational infrastructure to be designed for community development, in the context of "Agenda 3 - Initial Proposal Presentation of Scientific Activities within the scope of MCSTU", Carlos Bastarz suggests standardizing the development environment with the use of containers.

Actions for the Next Meeting

  1. Send the document with the initial proposal of the names listed for the lectures and round tables for consideration and collection of suggestions by the scientific committee;
  2. Find a focal point for the community model evaluation subnetwork;
  3. Consider the name of researcher Tadeu Gomes/LNCC for the theme of "High Performance Processing" in the lecture and round table cycles (suggestion by Roberto Souto/LNCC) and researcher Peter Bauer/ECMWF as a participant in the discussions within the scope of scientific activities;
  4. Reinforce the need for data assimilation to be considered in all components of the Earth system (suggestion by Pedro Dias/USP);
  5. Consider the suggestion by Ronald Buss/INPE to change the name of the theme "Oceans and Continental and Sea Ice" to "Oceans and Cryosphere";
  6. Include/detail software engineering proposals in the development of the community model (suggestions Roberto Souto/LNCC, Pedro Peixoto/USP, Haroldo Fraga/INPE and Luiz Flávio/INPE);
  7. Review the needs of the scientific committee within the scope of the INPE-ECMWF agreement;
Attachments