PNNL

The Science 

Mesoscale convective systems produce more than half of tropical rainfall, but the observational records used to track these large, organized storms cover only a few decades. Researchers evaluated whether ERA5, a widely used global reanalysis with a longer record, could support studies of tropical MCS climatology.

The team applied PyFLEXTRKR to hourly ERA5 precipitation and brightness-temperature data from 2007–2020 and compared the resulting storm tracks with those derived from satellite precipitation and infrared observations. Both datasets were analyzed at 0.25° resolution using a consistent MCS definition. The comparison examined storm cloud shields, precipitation features, geographic distributions, lifecycle evolution, seasonal and diurnal timing, and propagation across tropical land and ocean regions. 

The Impact 

ERA5 is widely used to study atmospheric conditions over periods longer than many satellite records. This evaluation showed where researchers could rely on it for studies of large tropical storm systems. ERA5 reproduced the broad locations, seasonal patterns, movement, and cold cloud characteristics of MCSs. However, it produced too much light rain and too little heavy rain. Those biases caused ERA5 to identify fewer storms and underestimate their share of tropical rainfall by 25%–34% in key regions. The results provided practical guidance for selecting suitable data: ERA5 could support cloud-based MCS climatology, but regional rainfall maps and storm-level precipitation estimates required caution, especially over tropical oceans.

Summary 

ERA5 brightness-temperature distributions generally agreed with satellite observations, allowing many cold cloud-shield characteristics to be tracked in a physically meaningful way. Storm lifetime, cloud area, broad seasonal distribution, and propagation direction were reasonably represented. However, the coldest cloud tops over tropical land were approximately 5 K too warm, indicating that ERA5 represented the most intense convective cores less accurately.

Precipitation-related characteristics showed larger discrepancies. ERA5 produced an excess of weak rain and a deficit of heavy rain within tracked MCSs. Its median heavy-rain volume ratio was 44% lower over land and 28% lower over ocean than in the satellite-based dataset. These rainfall biases caused fewer systems to satisfy the study’s precipitation criteria and contributed to a 25%–34% underestimate of the MCS share of total rainfall across the Amazon, tropical Africa, the Indian Ocean, and the Maritime Continent.

ERA5 reproduced the general shape of regional MCS diurnal cycles, but initiation and mature-stage peaks occurred 2–4 hours earlier than observed. It also failed to maintain the observed nocturnal peak in mature systems. The study found that MCS propagation direction was generally consistent between ERA5 and observations.

The researchers modified the PyFLEXTRKR configuration by disabling a function that links neighboring cloud shields through shared precipitation features. In ERA5, excessive weak rain caused this function to combine separate systems into unrealistically large storms. The paper recommended using the modified configuration for ERA5 and other precipitation datasets with similar biases. The study did not evaluate whether ERA5 was suitable for detecting long-term MCS trends.

Contact 

Renu Joseph, Earth and Environmental System Modeling (EESM) Program, Renu.Joseph@science.doe.gov 

L. Ruby Leung, Pacific Northwest National Laboratory, Ruby.Leung@pnnl.gov

Funding 

The U.S. Department of Energy Office of Science, Biological and Environmental Research supported this study through the Regional and Global Model Analysis program area. The DOE Office of Science Distinguished Scientist Fellows Program also provided support. The research used computing resources at the National Energy Research Scientific Computing Center, a DOE Office of Science user facility. The manuscript did not identify a more specific BER project or Scientific Focus Area name.

This text was initially generated using artificial intelligence, and subsequently refined and validated for accuracy, tone, and context by experts at Pacific Northwest National Laboratory.