Atmospheric Sciences Science Update

Torrential Rains and Poor Forecasts Sink Panama’s Infrastructure

Scientists are working to improve the forecasting of heavy rains in Panama following several events over the past decade that caused substantial flooding and damage.

By , Abel Centella-Artola, Maibys Sierra-Lorenzo, and Israel Borrajero-Montejo

December 2010 was the wettest month on record in Panama. So much rain fell so quickly that flooding was widespread across the country. One storm on 7–8 December of that year affected water intake at the Chilibre water treatment plant, leaving Panama City—the capital and largest city in the country—without clean water for 50 days. It also caused the closing of the Panama Canal for just the third time in history. This event, locally called La Purísima, produced more rainfall than any previously observed heavy-rain event in the Panama Canal Watershed [Murphy et al., 2014], costing upward of $150 million in damage.

In late November 2016, another heavy-rain event struck Panama, wreaking havoc. As Hurricane Otto crossed from the Caribbean to the Pacific, it dropped a month’s worth of rain in a day on the country, causing nine deaths, destroying hundreds of homes, closing schools, and interrupting activities at Tocumen International Airport, the international airport of Panama City. To maintain canal operations and lower the water level, the Panama Canal Authority opened 13 of the 14 gates of the Gatún Dam.

In both extreme rainfall events, Panama’s national weather forecast system failed to forecast the areas affected by, or the substantial impacts of, these meteorological phenomena.

The public began clamoring for the development of a more accurate high-resolution precipitation forecast system for Panama. That was the main motivation for a research project proposed by the Centro del Agua del Trópico Húmedo para América Latina y el Caribe (CATHALAC) and the Instituto de Meteorología de Cuba (INSMET) to the Secretaría Nacional de Ciencia, Tecnología e Innovación de Panamá (SENACYT). In June 2018, the project, titled “Análisis del Modelo Numérico WRF-ARW para la predicción de lluvia a escala de cuencas en Panamá” (“Analysis of the WRF-ARW numerical model for basin-scale rainfall prediction in Panama”), was approved and funded by SENACYT. And in February 2020, many of the people working on this project met to share and highlight its outcomes and progress so far and to assess directions for continuing work, discussions we summarize here.

Weather Forecasting in Panama

Despite its relatively small size, Panama, located at the eastern end of the Central America isthmus, plays crucial roles in regional and global economies. It hosts the Panama Canal, through which hundreds of millions of dollars’ worth of cargo pass each year, and is the international flight hub of Copa Airlines, for example. Activities along the canal and at Tocumen International Airport, as well as those related to the lives of almost 3 million inhabitants in Panama City, have been seriously affected by rainfall events that were not well forecast.

Two major circulation systems trigger convective activity near Panama: the Intertropical Convergence Zone and disturbances associated with it and cold fronts penetrating from northern higher latitudes. The isthmus’ long, narrow shape, its complex orography, and its position between two large bodies of water—the Caribbean Sea and Pacific Ocean—that both contribute large amounts of moisture to the atmosphere all contribute to unstable convective activity that leads to cloud formation and to the occurrence of strong rainfall. Because of the atmospheric instability in this region, however, rainfall amounts are very difficult to predict over short timescales.

Panama City is especially prone to damage from significant rainfall because its population growth has outpaced the growth of services infrastructure like stormwater disposal systems. So when heavy rains hit, they can cause services to be paralyzed for long periods, bringing subsequent economic and social damage. Panama’s problems with severe flooding thus arise from a combination of heavy rains, poor infrastructure, and the lack of quality forecasting.

There are two main reasons why weather forecasts fail in Panama. First, Panama’s operational forecasting system prior to the new project consisted of just two daily simulations with relatively coarse spatial resolutions. Second, and more important, was the lack of sensitivity studies used to determine the best physics and atmospheric dynamics parameters to describe the formation of storms over Panama.

Empresa de Transmisión Electrica (ETESA), Panama’s state electrical transmission company, oversees weather forecasts in the country through its meteorology and hydrology unit (HYDROMET). HYDROMET was supposed to rely on the next-generation Weather Research and Forecasting (WRF) model. When we started this project in 2018, though, HYDROMET’s WRF model for Panama was inactive. HYDROMET also did not have a platform by which model outputs could be disseminated, so even if the WRF model had been active, the public had no way to get the information. The other group that used the WRF model for weather forecasting for Panama was CATHALAC.

The Panama Precipitation Prediction Project

The goal of our project is to create an effective forecasting system for Panama using Advanced Research WRF (ARW) models that can identify with sufficient lead time the occurrence of extreme precipitation events, particularly over Panama City. The project, which involves researchers from the Center for Atmospheric Physics at the Institute of Meteorology of Cuba, CATHALAC, ETESA, and the Autoridad del Canal de Panama, comprises three stages.

The first stage involved studying how various dynamic factors directly influence the form (i.e., from what type of cloud it falls) and amount of precipitation over Panama using computational experiments and relying on data from many weather stations, as well as weather radar and upper air sounding instrumentation. Such dynamic factors include cloud microphysics (i.e., the necessary conditions for the formation of drops of precipitation), cumulus parameterization (the way the numerical models simulate clouds and their interactions with the environment), and the limit layer parameterizations (the way the atmospheric layer closest to Earth’s surface interacts with the layer where the clouds form).

We examined the combination of cumulus and microphysics parameterization schemes that best reproduce rainfall events in Panama and which model spatial scales yielded good representations of such events. Then we looked at how to implement an efficient and robust forecasting system with available technological and human resources and how to build the key institutional arrangements, research activities, and capacity-building processes to ensure further development of the implemented system.

The second stage of the project has focused on determining which of the parameterization schemes studied in stage 1 has the greatest potential to be implemented [Sierra-Lorenzo et al., 2020] and to provide accurate forecasts in enough time to alert users to the presence of severe weather in the study area, given current computational capacity. We have also been creating a system that gives users the ability to access the output data from these precipitation forecasts in real time. For this step, the project working group decided to use an immediate forecast system tool called Sistema de Pronostico Inmediato, or simply SisPi, which was developed in Cuba and has been used successfully in other Caribbean regions. SisPi allows for the visualization of numerical weather model outputs, providing a series of additional products with which forecasters can quickly assess a specific meteorological situation.

The third and final stage involves determining the necessary requirements for the forecasting system to be used at maximum capacity and figuring out what technological challenges must be overcome to achieve the best forecast in the fastest possible time.

After 18 months of research, including analyzing 150 case studies representing different seasons and 10 synoptic conditions over Central America, we found that no individual model configuration was able to accurately predict all intense rain events with acceptable skill. So we selected the three most skillful WRF-ARW model configurations and incorporated them within the weather forecast system under development. We also decided to build a Panama-specific version of SisPi (SisPi-Panama) within CATHALAC’s facilities and to run it automatically four times a day. This system will be the first in Central America that is based on a robust sensitivity assessment of its configuration, and it will be implemented to share different meteorological products widely through the SisPi-Panama web platform.

Where We Go from Here

During the first 18 months of this project, we fulfilled the first two objectives originally proposed. But to improve upon the work so far and meet the third objective, we submitted a second project proposal to SENACYT. The proposal included an idea from the workshop earlier this year to create an application for cell phones that would show the results of weather forecasts in real time and allow feedback from app users. This feedback should enable us to evaluate the skill of our system objectively and, when the feedback sample size becomes large enough, to evaluate statistics that may offer us the chance to perform bias correction in the models.

At the workshop, participants discussed new ideas for meeting the third project objective that should be taken into account in future efforts. Recommendations for ways to continue to improve forecasts included using emerging methods like machine learning and neural networks and conducting additional experiments in sensitivity studies. Participants were also keen on the possibilities of developing and producing real-time forecasts specific for solar and wind energy potential. We are also looking to partner with a United Nations agency like the World Meteorological Organization or the International Renewable Energy Agency to improve our computing capabilities.

Next steps for the project involve our partner institutions simulating combinations of multiple ARW model ensembles to test which combination most robustly forecasts heavy-rain events in Panama and to improve the SisPi-Panama web tool so that the information is available to all users. These steps will account for the recommendations that came out of the workshop, including developing a cell phone app. Further efforts will be needed to implement a system to assimilate existing high-quality data from satellite images, radar, upper air sounding measurements of atmospheric variables, and weather station data to improve the nowcasting scheme.

With what has been achieved in this project already, Panama is now equipped with a more robust forecasting system to predict severe rainfall events. This system will favor the country’s economy and safeguard human lives.

Acknowledgments

This research was funded through project 2018-4-IOMA17-011, “Análisis del Modelo Numérico WRF-ARW para la predicción de lluvia a escala de cuencas en Panamá,” which was financed by SENACYT and coordinated by CATHALAC.

References

Murphy, M. J., Jr., K. P. Georgakakos, and E. Shamir (2014), Climatological analysis of December rainfall in the Panama Canal watershed, Int. J. Climatol., 34(2), 403–415, https://doi.org/10.1002/joc.3694.

Sierra-Lorenzo, M., et al. (2020), Assessment of different WRF configurations performance for a rain event over Panama, Atmos. Clim. Sci., 10, 280–297, https://doi.org/10.4236/acs.2020.103016.

Author Information

Arnoldo Bezanilla-Morlot ([email protected]), Abel Centella-Artola, Maibys Sierra-Lorenzo, and Israel Borrajero-Montejo, Center for Atmospheric Physics, Institute of Meteorology of Cuba, Havana

Citation: Bezanilla-Morlot, A., A. Centella-Artola, M. Sierra-Lorenzo, and I. Borrajero-Montejo (2020), Torrential rains and poor forecasts sink Panama’s infrastructure, Eos, 101, https://doi.org/10.1029/2020EO150899. Published on 27 October 2020.
Text © 2020. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.