Two Weeks of Heavy Rain and a Year of Drought Over Parts of China (July 2021)

China storms and drought (July 2021)
Larger image: click here

During July 17 to 28, 2021, several storm systems brought heavy rain to parts of China and surrounding countries, while a nine-month-long drought persists in an adjacent part of China.  NASA’s multi-satellite precipitation algorithm has been monitoring this rainfall in near real-time, and the estimates are distributed to weather-forecasting agencies and disaster-monitoring organizations.  This algorithm is called IMERG, the Integrated Multi-satellitE Retrievals for GPM. GPM is the NASA / JAXA Global Precipitation Measurement mission, which launched its Core Observatory satellite in 2014.

Two Typhoons

IMERG precipitation estimates from July 17 to 28, 2021, are shown in the image above. Among the weather systems impacting the region were two landfalling typhoons – Cempaka and In-fa.

One can see that over a foot of rain has fallen during this 12 day period along most of the coast from Hong Kong west to Hanoi, Vietnam. This rain came mostly from Typhoon Cempaka during July 17 through 19.

Over a foot of rain has also fallen during this time near Shanghai, which is China’s most populous city.  That rain came mostly from Typhoon In-fa during July 25 to 28.  According to Buckingham (2021), parts of Zhoushan City, near Shanghai, received at least 30 inches of rain from In-fa between July 23 and 27. Over the Pacific Ocean near China, two rain gauges in the Global Historical Climatology Network (GHCN) reported 12-inch rain accumulations during the July 17 to 28 period at locations where NASA’s IMERG algorithm also reported similar accumulations.  These gauges are located on two small islands, Japan’s Nago and Kumejima islands. Point estimates of rainfall are represented in the image as circles that is colored using the same color scale used to display the IMERG data that covers the globe.  Nearby over the ocean, where no rain gauges exist, the IMERG algorithm reported significantly more rain accumulation along the path of Typhoon In-fa, over 30 inches of accumulation.

Two Additional Storm Systems

During the same time, a synoptic system brought over 30 inches of rain accumulation to the Chinese city of Zhengzhou and surrounding areas of Henan Province during July 19 to 22.  According to Childs (2021), over 31 inches fell during a three-day period.

Also separate from the two typhoons, a storm system brought accumulations in excess of 20 inches to the western coast of the Philippines, near Manila, during July 21 to 24.  These accumulations were detected by the IMERG algorithm and are consistent with rain gauge data in Manila.

While such accumulations are impressive, they are not unprecedented.  For example, Hurricane Harvey (2017) stalled over Texas and brought over 60 inches of rainfall to a small area over a period of four days (Blake and Zelinsky, 2018).  IMERG was also used to analyze the rainfall from Harvey

A Persistent Drought

Meanwhile, a drought in Fujian Province in coastal China has persisted since October 2020 with rainfall accumulation less than 25% of the typical accumulation for part of Fujian Province. Deviation from normal rainfall can be estimated using the long-term record of IMERG estimates that stretches back to the year 2000.

Wang (2021) and Ye (2021) report that China’s coast south of the Yangtze River, a region that includes Fujian Province, only received 20% to 50% of the normal amount of rainfall between October 2020 and February 2021. IMERG data shows that summer rains have relieved the drought in the provinces adjacent to Fujian, but not in Fujian itself.

The regional rainfall patterns seen during the past two weeks over China are significantly different than the Meiyu-Baiu pattern of rainfall that IMERG detected at this time last year. Credit: Visualization and caption by O. Kelley.


Rainfall that Matters: A Convective System over Arizona within the North American Monsoon (July 2021)

North American Monsoon region
Figure 1. Geographic extent of the North American Monsoon. Larger Image: click here

There is a monsoon that occurs in the southwestern U.S. each summer, and it brought heavy rain to the deserts of Arizona this week. This monsoon is less well known than India’s Summer Monsoon, but both monsoons have similar causes [1, 2, 3].

In western Mexico and the southern edge of the southwest U.S., most of the year’s rain typically falls in just three months: June, July, and August. The region is shown in light blue in the below climate map, which shows where summer rainfall predominates (Figure 1). This seasonal pattern is known as the North American Monsoon. The map was generated using the long-term record of NASA’s IMERG multi-satellite precipitation analysis. Climatologists and forecasters have used various dates to define the exact start and stop dates of the North American Monsoon [4].

This year’s monsoon season has officially begun, and the week of July 12-16 has been an active monsoon week for Southern Arizona [5].

On the morning of July 15, 2021, the National Weather Service issued a warning for severe thunderstorms over Arizona’s southern desert with rain rates of 1 inch per hour over ground that was already saturated by recent storms [6]. Under these circumstances, flash flooding is a concern. The location of this warning is indicated by the red circle on the climate map (31.72N latitude and 112.13W longitude in Figure 1).

The NASA / JAXA GPM Core Observatory satellite captured a 3D view of a convective storm system at this location at 8:37 a.m. local time on July 15. The data was collected by the satellite’s Dual-frequency Precipitation Radar (DPR). The 3D view is shown in Figure 2. This overflight of the storm system showed several convective cells embedded within it with precipitation rates at the Earth’s surface in excess of 1.4 inches per hour. The strongest of these cells contained updrafts that were sufficiently strong to lift ice precipitation overshooting through the top of the troposphere. The troposphere is the lowest layer of the atmosphere that normally contains the weather.

Arizona Storm (2021)
Figure 2. 3D observation of a convective system over Arizona contributing to the North American Monsoon. Larger image: click here.

Shown in red in Figure 2, the Dual-frequency Precipitation Radar detected a 20-dBZ radar-reflectivity signal at an altitude 14 kilometers above the Earth’s surface and over 14.8 km above sea level. In Figure 2, the yellow and green areas identify where 20-dBZ radar reflectivity was detected at 11 and 8.5 km altitude, respectively. Commercial airlines cruise at 9 to 13 kilometers. At lower altitudes, this convective cell contained radar reflectivity in excess of 50 dBZ, suggesting hail was present.

This GPM satellite overflight of the storm system occurred at 8:37 a.m., approximately half an hour before the National Weather Service issued its thunderstorm discussion for the area [6].

This portion of Arizona receives only 10 to 15 inches of precipitation in an entire year, on average, according to the long-term record from NASA’s IMERG multi-satellite precipitation algorithm. Local accumulation exceeded 2 inches in places due to the convective storm system that passed through this area on July 15, 2021.  This storm total was estimated both by the near real-time IMERG algorithm and also by the National Weather Service’s preliminary analysis of ground-radar data [].

It is typical of this area’s climate that heavy rainfall from convective systems such as this one contribute much of the area’s annual rainfall accumulation. According to the IMERG record, this portion of Arizona receives approximately 42% to 50% of its annual accumulation in June through August.  Visualization and caption by Owen Kelley (NASA/GMU).


(1) NASA, 2017: NASA Looks at the North American Monsoon. 1:26 minute video featuring NASA Scientist Dr. George Huffman,

(2) Branon, M., 2021 June 17: How the Weather Service cleared the air about Southwest monsoon season. Capital Weather Gang, Washington Post,….

(3) Adams, D. K., and A. C. Comrie, 1997: The North American Monsoon. Bulletin of the American Meteorological Society, 78, 2197-2213.

(4) June 15 is currently the National Weather Service’s official start date for the North American Monsoon.  Adams and Comrie (1997) propose June-August or July-September as possible date ranges for the monsoon.

(5) Arizona Daily Star, 2021 July 14: Tucson passes 2020 monsoon rainfall total; flash flood warning in effect,….

(6) NWS, 2021 July 15: Mesoscale Precipitation Discussion #0546 issued at 9:10AM EDT on Thursday, July 15, 2021. Weather Prediction Center (WPC), College Park, MD,….


Australia’s Heavy Rainfall: Desert and Coast (March 2021)

Australia Flooding (March 2021)

larger image: click here

During the week ending on March 23, 2021, two locations in Australia experienced unusually high rainfall totals. In the news, a persistent system brought flooding rains to Australia’s east coast from Brisbane to Sydney and points further south.

The preliminary estimate from NASA’s multi-satellite global precipitation analysis is that more than 24 inches fell just off the coast of Australia in 7 days with accumulations in coastal areas exceeding 16 inches. Near the Strzelecki Desert in central Australia, a storm system brought 8 inches of precipitation during the same 7-day period. Most of the rain fell during a 3-day period (0000 UTC on 20 March to 2359 UTC on 22 March).

These estimates come from the near-realtime version of NASA’s IMERG algorithm which combines microwave and infrared satellite observations with climatological datasets including rain-gauge
observations. In addition to near global coverage, another advantage of the IMERG algorithm is that it provides estimates covering approximately a 20-year period. This long-term record allows for comparisons of current precipitation events with long-term averages.

According to IMERG records, the past week’s rainfall accumulation along Australia’s eastern coast is more than 25% of the average annual accumulation there. The past week’s rainfall accumulation near the Strzelecki Desert is greater than the average annual accumulation there.

Data from the Australian Bureau of Meteorology confirms at least the main conclusions from these IMERG observations. Central Australia typically receives only 4 to 8 inches (100 to 200 millimeters) of precipitation a year, while a thin band of land along Australia’s eastern coast typically receives about ten times as much precipitation: 40 to 60 inches (1000 to 1500 millimeters) in a year.

In terms of 7-day storm totals, the Australian Bureau of Meteorology reports that there is a 1% chance each year of an event exceeding 498 or 177 millimeters (20 or 7 inches) of precipitation accumulation in Sydney or in the desert where New South Wales, Queensland, and South Australia meet. The desert rainstorm during the week ending on March 23, 2021, exceeded this threshold, while the flooding over Australia’s east coast did not.

Credits: IMERG data from NASA at Average annual accumulation from the Australia Bureau of Meteorology at One percent annual chance of 7-day rainfall accumulation from the Australian Bureau of Meteorology’s intensity-depth-frequency (IDF) data in the Design Rainfall Data System (2016) at Visualization and caption by O. Kelley.


Hurricane Sally in Context (Sept. 2020)

Hurricane Sally (2020) and Hurricane Harvey (2017)

One way to put a storm into context is to ­­­­­compare its precipitation to the average annual accumulation at that location. The IMERG multi-satellite precipitation algorithm estimates that Hurricane Sally brought over 20 inches of precipitation to the coast of Florida’s Panhandle during the 7 days ending September 18, 2020, at 0000 UTC. That accumulation was equivalent to 25% to 33% of that area’s average annual precipitation. This area is indicated in purple on the left half of the image.

The coast of Florida’s Panhandle is the wettest area along the entire US East Coast and Gulf Coast. It receives approximately 60 to 68 inches of precipitation in an average year, based on 19 years of estimates made by the IMERG algorithm.

When it comes to hurricane flooding along the Gulf Coast, Hurricane Sally does not top Hurricane Harvey which made landfall in late August, 2017. Like Sally, Harvey also dumped over 20 inches of rainfall, but over a larger area that included a major metropolitan area. Specifically, this area of southeastern Texas includes Houston, Texas, a city that typically averages approximately 55 to 60 inches of precipitation a year. In Houston, the 7-day accumulation from Hurricane Harvey was 50% to 75% of the city’s average annual accumulation. This area is indicated in red on the right half of the image. Houston area flooding contributed to Hurricane Harvey being one of the most costly hurricanes in US history.

Our ability to interpret near real-time estimates of precipitation over both land and ocean is assisted by the availability of a reference data set of 19 years of global precipitation estimates made by the same IMERG algorithm. Visualization and caption by O. Kelley.


A Year’s Worth of Rain in a Week for a City in Pakistan (Aug. 2020)

Pakistan Flood (August 2020)

In the last week of August 2020, Pakistan’s largest city, Karachi, received over 8 inches of rainfall according to NASA’s IMERG dataset, causing destructive flooding in the region. The amount of rain that fell that week is roughly equivalent to the amount that Karachi typically receives in an entire year, based on IMERG’s 19-year global climatology. In a typical year, most of Karachi’s rain will fall in July and August, but the rainfall during the week of August 23rd was unusually heavy.

The top panel of the three panels in this image shows the depth of the 7-day rainfall accumulation in inches (August 23 to 30, 2020). The light green color indicates at least 8 inches of rainfall in Karachi and locations further inland and to the east.

The middle panel divides the rain that fell during this seven day period by the average annual accumulation at each 0.1 x 0.1 degree grid box of the IMERG product.  Values in excess of 1 (red) indicate locations where a full year’s worth of rain fell in this one-week period. When such an extreme event occurs, it can be difficult for the built environment to function normally.

The bottom panel gives a sense of what is normal for the last week of August.  It shows what fraction of the annual total precipitation falls during the last week of August of a typical year.  Regions where at least 1/24 of a “year’s worth” of rain typically falls during the last week of August  normally experience part of their rainy season during the last week of August.  This is the case for the eastern portion of Pakistan and India.  Regions where less than 1/100 of a “year’s worth” of rain typically falls in the last week of August normally experience part of their dry season during the last week of August.  This is the case for western Pakistan and Afghanistan.

IMERG rainfall estimates are automatically generated in near real-time as part of NASA’s effort to monitor the Earth. The estimates are based on observations from an international constellation of satellite including the GPM satellite. Several months after an event occurs, the rainfall estimates are improved when NASA re-runs the IMERG algorithm using additional data sets that are not available in near real-time.


IMERG Reveals Two Weeks of Heavy Rain over the Eastern Mediterranean (Jan. 2020)

IMERG estimate of two weeks of heavy rainfall over the Mediterranean

In the two weeks starting on December 25, 2019, several heavy storm systems impacted countries along the eastern Mediterranean. While January is typically one of the rainiest months of the year in this region, the rainfall totals are truly impressive and far above typical for this region and season.

NASA’s satellite-based near-realtime precipitation-estimation algorithm reports over 30 inches (760 mm) of accumulation during this period in isolated locations just off the coast of Syria and Cyprus, with accumulation in excess of 24 inches (610 mm) over a patch of northern Israel. News stories from cities in these regions report severe flooding, including in several cities in Israel, Latakia in Syria, and Chrysochous in Crete. Several low-pressure centers over the eastern Mediterranean have contributed to the sea-to-land flow of moist air that has fed the flooding.

NASA’s algorithm that combines precipitation estimates from a international fleet of satellites is called IMERG.  IMERG stands for Integrated Multi-satellitE Retrievals for GPM.  Visualization by O. Kelley.


Estimating Tropical Cyclone Imelda’s Rain Total over Texas (Sept. 2019)

Tropical Storm Imelda over Texas (Sept 2019)

By Friday morning, September 20, the rainfall from the remnant of Tropical Storm Imelda had increased to over 24 inches in some areas near the Gulf of Mexico coast between Beaumont and Houston, Texas. This rainfall was in excess of what had been forcasted a few days earlier and was due to Imelda’s forward motion ceasing for approximately 24 hours between Wednesday and Thursday afternoon. The image shows, with large “L” symbols, the location estimated by the National Hurricane Center for Imelda’s low-pressure center of rotation at various times over the past three days.

This near-realtime rain estimate comes from the NASA’s IMERG algorithm, which combines observations from a fleet of satellites, in near-realtime, to provide near-global estimates of precipitation every 30 minutes.

If one compares the IMERG satellite-based rain estimate to that from a National Weather Service ground radar, one sees that IMERG correctly identified the large region of heavy rainfall near Beaumont, but IMERG failed to resolve an extremely narrow band of heavy rainfall along Galveston Island. Such good detection of large rain features in realtime would be impossible if the IMERG algorithm merely reported the precipitation observed by the periodic overflights of various agencies’ satellites. Instead, what the IMERG algorithm does is “morph” high-quality satellite observations along the direction of the steering winds to deliver information about rain at times and places where such satellite overflights did not occur. Information morphing is particularly important over the majority of the world’s surface that lacks ground-radar coverage.

An “R” symbol on the image indicates a place where the rainfall from the remnant of Imelda caused a US Geological Survey river gauge to swell to “major flood” stage. “Major” flood generally means that nearby homes and roads were flooded. The river-gauge data shown here is intended merely to give a hint of what areas experienced flooding and is not intended to portray the complete extent of flooding. In addition, there were several preliminary reports of Imelda-spawn tornados on Wednesday and Thursday, September 18-19. Red circles on this image indicate the location of these tornado reports, as provided by NOAA’s Storm Prediction Center. Visualization by NASA Goddard.  Visualization and caption by O. Kelley.


Hurricane Dorian, from the Caribbean to Canada (Sept. 2019)

Hurricane Dorian (2019) brought heavy rain to the Caribbean, along the US East Coast, and up to Canada.  NASA satellite-based precipitation estimates tracked the storm throughout its lifetime, as shown by the sequence of images below.

September 3, 2019: Hurricane Dorian over Grand Bahama and Abaco Islands

In the early hours of Tuesday, September 3, Hurricane Dorian had been stationary over the island of Grand Bahama for 18 hours, most of the time as a category 5 hurricane.  Storm-total rain accumulation over parts of Grand Bahama and Abaco islands have exceeded 24 inches according to NASA satellite-based estimates. On early Tuesday morning, Dorian’s central pressure has risen and its wind intensity had dropped to category 4 on the Saffir-Simpson scale.  In addition, Dorian had experienced an eyewall replacement cycle on September 2, so by Tuesday morning, the geographic extent of its tropical-storm-force winds had expanded.

These rain estimates come from the NASA IMERG algorithm, which combines observations from a fleet of satellites, in near-realtime, to provide global estimates of precipitation every 30 minutes. The storm-total rainfall at a particular location varies with the forward speed of the hurricane, with the size of the hurricane’s wind field, and with how vigorous the updrafts are inside of the hurricane’s eyewall.

The graphic also shows the distance that tropical-storm force (39 mph) winds extend from Hurricane Dorian’s low-pressure center, as reported by the National Hurricane Center.  The symbols H and TS represent a hurricane of various Saffir-Simpson categories or a tropical storm, respectively.  Visualization by Owen Kelley (NASA/GMU).

September 6, 2019: Hurricane Dorian Brings Tornadoes and Floods to the US East Coast

By Friday morning, September 6, Hurricane Dorian was located off the coast of North Carolina, having generated tornadoes the previous day as the northern rainband came ashore in North Carolina. NASA’s satellite-based realtime precipitation estimates suggest that, during the past day, most of the areas experiencing over 10 inches of rain accumulation remained offshore, while Dorian did drop heavy rain on South Carolina and North Carolina.

The preliminary reports of tornadoes were obtained from NOAA’s Storm Prediction Center, and are shown on the graphic as red circles. Since storm spotters are land based, these reports rarely capture any water spouts (tornado-like events over water) that might occur. As Hurricane Dorian interacted with the U.S. East Coast, the only tornado reports occurred from 4:50 AM to 5:00 PM EDT on September 5 in North and South Carolina.  Scientists think of a hurricane as a heat engine that converts the warmth of the sun-warmed ocean into the kinetic energy of the hurricane’s strong, horizontal wind. When these strong winds reach land, the increased friction of the land surface vs. the ocean surface can convert some of this kinetic energy into tornadoes within the hurricane.

The near-realtime rain estimates come from the NASA’s IMERG algorithm, which combines observations from a fleet of satellites, in near-realtime, to provide global estimates of precipitation every 30 minutes.  The storm-total rainfall at a particular location varies with the forward speed of the hurricane, with the size of the hurricane’s wind field, and with how vigorous the updrafts are inside the hurricane.  This graphic only shows precipitation that fell starting at 0000UTC on September 1, and therefore does not show the precipitation that fell in late August, prior to Hurricane Dorian’s approach to The Bahamas.

The graphic shows the distance that tropical-storm force (39 mph) winds extend from Hurricane Dorian’s low-pressure center, as estimated by the National Hurricane Center.  The Saffir-Simpson intensity category is the number following the “H” in the label on the image.  Visualization by Owen Kelley (NASA/GMU).

September 9, 2019: Dorian Reaches Canada

On Monday morning, September 9, Hurricane Dorian was a post-tropical storm after a mid-latitude weather front and cold seas had altered its tropical characteristics over the weekend. On Saturday and Sunday, Hurricane Dorian struck eastern Canada, causing wind damage and bringing heavy rainfall.  According to the Associated Press, a peak of 400,000 people were without power in Nova Scotia, Canada, because of Dorian.

This graphic shows precipitation that fell during the almost two-week period from August 27 to the early hours of September 9. The near-realtime rain estimates come from the NASA’s IMERG algorithm, which combines observations from a fleet of satellites, in near-realtime, to provide near-global estimates of precipitation every 30 minutes.

This year, NASA began running a improved version of the IMERG algorithm that does a better job estimating precipitation at high latitudes, specifically north of 60 degrees North latitude. The post-tropical remnant of Hurricane Dorian was approaching this cold region at the end of the period shown in this image. While the IMERG algorithm is still unable to estimate precipitation falling over ice-covered surfaces (such as Greenland), IMERG can now give a more complete picture of the water cycle in places such as Canada, which is, for the most part, free of snow cover at this time of year.

At one-day intervals, the image shows the distance that tropical-storm force (39 mph) winds extended from Hurricane Dorian’s low-pressure center, as estimated by the National Hurricane Center.  The Saffir-Simpson hurricane-intensity category is the number following the “H” in the label on the image. “TS” or “PT” indicate times when the storm was either at tropical storm strength or when the storm was categorized as post-tropical.

Red circles over North Carolina indicate preliminary reports of tornados from 4:50 AM to 5:00 PM EDT on September 5, provided by NOAA’s Storm Prediction Center.  Since storm spotters are land based, these reports rarely capture water spouts (tornado-like events over water) that may occur.  Lines of latitude and longitude are curved on this map projection, as would be seen by an observer in Earth orbit, so that both tropical and arctic regions can be shown on the same map with minimal distortion. By combining NASA precipitation estimates with other data sources, we can gain a greater understanding of major storms that effect our planet. Visualization by O. Kelley.


Alaska’s Wildfires, Precipitation, and Lightning (summer 2019)

NASA’s satellite-based estimate of global precipitation can provide valuable information to officials monitoring the many wildfires in Alaska this summer. Wildfires occur in Alaska each summer, but July 2019 was a particularly active month. Few rain gauges exist in the large tracts of Alaskan wilderness, making satellite-based precipitation estimates potentially more valuable for monitoring wildfire. Many wildfires in the Alaskan wilderness are monitored but allowed to burn themselves out. Monitoring triggers firefighting efforts when a fire threatens life, infrastructure, or locations with critical ecological resources.

The movie shows NASA’s IMERG precipitation estimates for May through September, 2019. The total accumulation since May 1 is shown in millimeters (inches) on the left half, while the accumulation during a 3-hour period is shown on the right half. The locations of likely fires are shown in red, based on thermal anomalies observed by the VIIRS instrument on the Suomi NPP polar-orbiting satellite. The VIIRS “hot spot” data has a resolution of approximately 0.25 square kilometers and is based on infrared brightness temperature. Locations of lightning strikes are shown in yellow, as detected by the network of ground sensors that make up the World Wide Lightning Location Network. A flash is detected when five or more WWLLN stations around the world detect a radio-frequency atmospheric signal from the same lightning flash. A gray circle along the southern coast or center of Alaska represents the cities of Anchorage or Fairbanks, respectively.

The first part of the movie covers May 2019, and it is a period of little precipitation, little lightning, and few wildfires.

June 2019 shows an increasing amount of lightning but still few large fires. During June, the storms that do pass through central Alaska deliver only about half of the climatological normal amount of precipitation according to NOAA’s Climate Prediction Center.

At end of June and into July, things start to heat up. Numerous wildfires are present in Alaska even though regional storms reduced the intensity of some fires or put them out. One such storm passes through Alaska’s west coast on June 27 and another on July 1 near Fairbanks. During the first half of July, many wildfires burn. There is an absence of large storms coupled with significant lightning activity, which together contribute both to a dry fuel supply and lightning to ignite it.

By July 25, the U.S. Bureau of Land Management (BLM) reported that over 2 million acres of Alaska forests had burned so far this year, 98% of which were ignited by lightning rather than man-made fire sources. The area burned is equivalent to a square with sides 58 miles long.

Credits: For IMERG data, visit NASA’s Precipitation Measurement Missions ( The Visible Infrared Imaging Radiometer Suite (VIIRS) hot-spot data was downloaded from NASA’s Fire Information for Resource Management System ( Lightning data provided by University of Washington ( Climatological data provided by NOAA’s Climate Prediction Center ( Fire statistics from the Bureau of Land Management ( Visualization by Owen Kelley at NASA Goddard’s Precipitation Processing System.


Large 7-day Accumulation over Western India (August 2019)

In early August 2019, a depression formed in the Bay of Bengal that moved over India contributing to heavy rainfall on India’s west coast. NASA’s satellite data analysis suggests that for August 5 though 11, two feet of rain fell in some places. This estimate is from the realtime multi-satellite algorithm called IMERG, which is run at NASA Goddard.