Category: News

In many places, the amount of that rain falls on a particular day or hour is influenced by multiple natural cycles that exist simultaneously on different time scales. One such cycle repeats every 24 hours, and another repeats every year.

The global extent of these cycles can be studied using NASA’s IMERG precipitation estimates that have been generated for 1998 to the present. Most of Earth’s precipitation falls as rain, and the rest falls as snow, hail, or other forms of precipitation.

The length of IMERG’s data record makes it possible to average out some of the random error and unrelated weather fluctuation.  For example, any 100-year storm that occurs around the globe each year would do more to obscure the underlying daily and annual cycle of rainfall in a dataset that was shorter than IMERG’s 28 years. IMERG stands for “Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (GPM) Mission.” The IMERG algorithm is developed by a team of scientists, most of whom are affiliated with NASA Goddard Space Flight Center (Huffman et al. 2023a, 2023b, 2023c).

IMERG precipitation estimates give a more complete picture of Earth’s natural cycles because the dataset has high resolution and global extent. IMERG covers both continents and oceans because it is input data comes from a constellation of satellites that observe the Earth in the microwave, visible, and infrared spectrum. Over decades, individual satellites have joined and dropped out of this constellation like the changing lineup of a professional sports team.

IMERG is available every 30 minutes, globally, at a resolution of no less than 7 miles. IMERG provides estimates every 0.1 degrees of latitude and longitude, which equates to about every 7 miles at the Equator and less near the poles.

As described in an earlier article, you can browse through all of these 30-minute snapshots by visiting NASA’s World View webpage. Natural cycles can be hard to see by flipping through a series of 30-minute estimates. These cycles are easier to see if one first calculates average conditions and then calculates the time-varying deviation from these averages.  This approach is followed here.

The Daily Cycle

During some seasons and in some places, the average amount of precipitation varies with the time of day. Atmospheric scientists call this type of variation the diurnal cycle, and IMERG has been used to study this cycle (e.g., Tan et al. 2019, Watters and Battaglia 2019, Xiong et al. 2025, and Patel 2020). The diurnal cycle during July is shown in the animation below for 70º South to 70º North latitude. A single month of the year was chosen for this animation because, in many places, the diurnal cycle is different in summer and winter.

Precipitation varies with time of day during July, averaging over 1998–2025. Click here to play the animation.

To reduce noise, each frame of the animation covers a 12-hour period. For example, the first frame is centered on 6 AM local time and summarizes IMERG observations for midnight through noon during all Julys between 1998 and 2025. Each 12-hour frame starts 30 minutes later than the prior frame, so the second frame covers 12:30AM at night through 12:30PM in the early afternoon.

For a particular 12-hour period, the image is white where the precipitation rate equals the average rate for the whole day. Deep blue or red indicates large deviations from average conditions. Specifically, deep blue indicates that the 12-hour period has a precipitation rate that is about 50% higher than average and deep red indicates a rate 50% lower than average. The color bar was constructed with units of percent deviation from average conditions. Doing so makes the diurnal cycle easy to see in deserts and other arid locations, not just places that receive more rain.

As you examine the daily-cycle animation, it is important to understand how time is represented. A single frame of the animation shows data for the same range of local times everywhere on Earth. In contrast, a snapshot of the Earth would always show the full 24-hour range of local time as a function of longitude.

The speckle of fine-scale variation that is seen in each of these 12-hour averages demonstrates the high variability of rainfall and snowfall in both space and time. More than a hundred thousand 30-minute IMERG estimates were averaged together to produce each 12-hour image, and even this approach is insufficient to smooth out fine-scale geographic variation due to the influence of individual storms. In the face of nature’s variability, only a broad picture of the global, daily cycle of rainfall emerges from IMERG, a data product that represents NASA’s effort to optimally combine 28 years of observations from a constellation of geosynchronous and polar-orbiting satellites.

The Daily Cycle over Land or Ocean

Below, a single frame is extracted from the previous animation in order to illustrate the rule of thumb that rain tends to fall over land in the late afternoon and over ocean in the early morning.

The physical basis for this rule of thumb is that, unless it’s cloudy, the sun warms the land surface throughout the day, creating maximum atmospheric instability in the afternoon. This instability can trigger storms that persist for several hours as they are carried by the prevailing winds. In contrast, the temperature of the ocean surface changes very little most days because it takes much more energy to increase the temperature of water vs. that of air. Over the ocean, the nighttime cooling of tops of clouds increases atmospheric instability and can trigger rainfall around sunrise. The image below shows these patterns in IMERG data.

Over Asia and North America, the percent difference calculated from the afternoon-evening precipitation rate (noon to midnight local time) minus the daily-average precipitation rate.

The image shows afternoon and evening rainfall, averaging observation made from local noon to midnight during all Julys from 1998 to 2025. In this image, many land areas are dark blue, which indicates it rains there predominantly in the afternoon and evening. In contrast, much of the ocean is dark red, indicating that little precipitation falls there in the afternoon or evening.

Mountains Affect the Daily Cycle of Precipitation

As just described, enhanced rainfall in the afternoon or evening is common over land, but mountain ranges can modify this pattern. For example, the Rocky Mountains have an early afternoon precipitation maximum, but strong storm systems triggered by these mountains often persist for hours as the prevailing wind carries them hundreds of miles to the east. The many storms that follow this track result in peak precipitation occurring later in the day east of the Rockies.

The Rocky Mountains’ impact is easy to see in the following cross section with longitude on its horizontal axis and time on the vertical axis. This kind of diagram is a called a Hovmoller diagram, and the one below is constructed for a west-east slice of the United States. The slice covers 35º to 40º North latitude, which includes San Francisco east to Denver, St. Louis, and Washington DC. The diagram combines all IMERG data for July during 1998–2025.

Looking at the diagram, it is clear that the rainiest part of the day is early afternoon in Denver and 12 hours later 850 miles east in St. Louis.

Precipitation variation over the United States in July during 1998–2025, as function of longitude and time of day.

The just described geographic variation of the daily cycle has implications for the constellation of satellites necessary to create global precipitation datasets that capture this natural cycle. It would not be possible to study precipitation’s diurnal variation with satellites that only observed the Earth at a fixed set of one or two local times in the afternoon and evening. During the IMERG era, Earth-observing satellites with microwave instruments, fortunately, have been stationed in a wide range of orbits, as depicted in an overflight-time plot generated by NASA.

One way to check that the diurnal patterns that IMERG sees around the globe are reasonable is by comparing them against regional high-quality independent datasets. For example, over the lower 48 states of the United States, there is a network of 150 weather radars operated by the National Weather Service. Data from these radars are shown every day on TV and online weather forecasts. Figure 6 of Sungmin and Kirstetter (2018) shows that this ground-radar data has a similar diurnal pattern to that shown in the IMERG diagram above.

The Annual cycle

Rainy and dry seasons exist are annual occurrences in some parts of the world. The timing of the wet season onset is economically significant and difficult to forecast. Scientists have used IMERG to forecast the onset and spatial pattern of the summer rainy season over India, West Africa, and Florida (Jayasankar and Misra 2025a; 2025b; Misra and Jayasankar 2024; GPM website). Other studies have used IMERG as a reference dataset to evaluate the accuracy of model forecasts of the rainy season over the watershed that supplies the Nile River (Gebremichael et al., 2025). Satellite-based precipitation estimates are the only hope here because there are too few rain gauges and weather radars in Africa for them to verify model forecasts of Nile River flow.

The following IMERG animation shows the annual cycle of precipitation across the world and has frames that each summarize a 30-day period. The first frame, for example, includes data from the first 30 days of 28 years (1998–2025). The next frame starts and ends two days later than the first frame. The yellow dot that moves north and south between 23.5°S and 23.5°N latitude indicates the latitude at which the noon sun is directly overhead during that time of the year.

As the previous section described for daily cycle, the animation below uses a similar red-blue color table to identify dry and wet anomalies in the annual cycle. For a particular 30-day period, the image is red or blue where the precipitation rate is less than or more than the annual average precipitation rate. In this way, this animation shows deviation from average conditions. In contrast, the GPM website’s Precipitation Climatology page shows the annual cycle of precipitation amount rather than the time-varying deviation from average conditions.

The annual cycle, showing how precipitation varies by day of year. To view the animation, click here.

As noted previously about the daily cycle, IMERG’s depiction of the annual cycle also contains strong small-scale variations that show up as speckles in the animation. NASA has expended considerable effort to extend the IMERG dataset back to 1998 in part because a long dataset is necessary to damp down the impact of individual weather events and to reveal underlying regional variation in the annual precipitation cycle.

The annual cycle and the sun

Over continents, there are many places that receive much of their rain during the summer. In both the Northern and Southern Hemispheres, summer starts in the part of the year when the sun is highest in the sky. This north-south migration of large precipitation accumulations can be seen in a time-vs.-latitude diagram, such as the one below. This diagram includes IMERG data from 50º to 60º East longitude, which includes locations from South Africa north to Finland. The line of yellow dots in this diagram shows the latitude at which the sun is directly overhead at noon.

A time-vs.-latitude cross section of the annual cycle of IMERG precipitation through a north-south section through Africa and Europe.

The rule of thumb that summer is the rainy season does not apply to ocean (Fig. 7, Huffman et al. 2023) or even to some land areas. For example, the image below averages data over the United States for all Januarys from 1998 to 2025. Blue over California, Oregon, and Washington indicates that January is typically a wet month on the West Coast. Red over the central United States indicates that little of the year’s precipitation falls there in January most years. The gray over the eastern United States indicates that precipitation in January typically accumulates at close to the average annual rate there.

Average January precipitation over the United States relative to annual-average precipitation.

Another way to explore the seasonal variation of precipitation over the United States is with a time-series diagram. Below is a time-series plot of three regions: the Pacific Northwest, the Great Plains of the central U.S., and the mid-Atlantic. The inset map shows these three regions. The horizontal axis is the day of year and the vertical axis is the long-term average precipitation rate for 30-day periods within the annual cycle. Each region’s median value is shown with dark dots, and the shaded area shows the middle 50% of IMERG values in the region. Cyan is used for Pacific Northwest, tan for the Great Plains, and light green for the mid-Atlantic region.

A time series for three regions of the United States that shows how their seasonal precipitation patterns differ.

Two of these three regions have peak precipitation near the time of the year that the sun is highest in the sky. The summertime peak is more pronounced over the Great Plains than over the mid-Atlantic. The Pacific Northwest, in contrast, has most of its annual precipitation fall in winter when it receives the least solar heating but is soaked by numerous storms coming off the ocean. A similar pattern is seen if the time series is calculated from NOAA ground-radar data (not shown) instead of IMERG data.

Final thoughts

NASA’s IMERG precipitation data contains daily and annual cycles. These natural cycles are easier to see if one examines deviations from average conditions. The daily and annual cycles are present in the 3.2 trillion precipitation values that make up the IMERG dataset for 1998 through the present.[1]

The IMERG algorithm is run and archived at NASA Goddard Space Flight Center. The Center also hosts the team of scientists who develops the IMERG algorithm. GMU’s CEOSR assists NASA both with evaluating the performance of the operational IMERG version and new versions before they become operational.

Notes

[1] 3.18´10¹² IMERG precipitation values, which is derived from the following expression: (3600 longitude grid boxes) (1800 latitude grid boxes) (48 estimates per day) (365.25 days per year) (28 years).

• *The programs and services offered by George Mason University are open to all who seek them. George Mason does not discriminate on the basis of race, color, religion, ethnic national origin (including shared ancestry and/or ethnic characteristics), sex, disability, military status (including veteran status), sexual orientation, gender identity, gender expression, age, marital status, pregnancy status, genetic information, or any other characteristic protected by law. After an initial review of its policies and practices, the university affirms its commitment to meet all federal mandates as articulated in federal law, as well as recent executive orders and federal agency directives.

During the first three weeks of December 2025,several atmosphere rivers have carried plumes of moisture from the Tropics toward the United States. These plumes of water vapor have contributed to flood-producing storms over Washington State. The illustration below shows the approximate path that the moisture took to reach the United States.

Schematic diagram of an atmospheric river. [High-res image]

NASA’s IMERG algorithm stitches together observations from multiple satellites to estimate the precipitation that falls across the globe. IMERG stands for Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (GPM) mission. NASA’s Goddard Space Flight Center is home to the team of scientists who work to improve this algorithm and for the computing power that enables the algorithm to process more than a petabyte of observations.

To get a more precise idea of how much precipitation fell, the image below shows contours of precipitation accumulation during December 1 through 15. In this image, accumulation is measured in inches and is the sum of the depth of rainfall plus the depth of liquid that would be produced by melting any snowfall. This quantity is called the liquid-equivalent depth of the snowfall.

Precipitation accumulation over Pacific Ocean and the northwestern United States. [High-res image]

The above image shows two parallel trails of rainfall that stretch southwest to northeast across the Pacific Ocean. The northern trail reached Canada and the southern trail reached Washington State. These precipitation trails are evidence of the large amount of water vapor being transported along these routes, compared with the dryer air over nearby ocean regions.

Over the Pacific Ocean, the rainfall produced by these atmospheric rivers was fairly modest at 1 to 6 inches of accumulation during the first half of December 2025. The accumulation was more substantial over land. The large accumulations over North America were due in part to mountain ranges lifting the moist, oceanic, tropical air. IMERG estimated that the two-week precipitation total exceeded 15 inches over Washington’s Olympic Mountains, and further inland, the total exceeded 24 inches over parts of Washington’s Cascade Mountains.

During the first half of December western Washington suffered landslides, flooding, and power outages as a result of these atmospheric rivers. These storms also closed several highways that cross the Cascade Mountains. To see more clearly the geographic variation of rainfall, the zoomed-in image below shows just Washington State.

Close-up of precipitation over Washington State. [High-res image]

In the yellow-dotted region of the Cascade Mountains, a NOAA dataset broadly agreed with the NASA IMERG precipitation estimate, which is notable for two reasons. For one thing, much of the precipitation in this area fell as snow, which is more difficult to estimate from satellite observations than moderate rainfall. Second, any type of precipitation is difficult to estimate over mountainous regions, whether using ground radars, satellites, or rain gauges. NOAA generated its near real-time estimate over the United States using ground radars, rain gauges, and satellites. NOAA’s product is called “Stage IV.” IMERG’s near real-time estimates were generated over the whole globe using satellite observations calibrated against long-term gauge averages. The IMERG precipitation estimates also factored in the storm motion that was estimated from winds in a weather model.

Near the coast, the NASA and NOAA datasets diverged somewhat. Only the NOAA dataset saw a region of 24-inch accumulation over the Cascade Mountains, a compact mountain range west of Seattle. A rain gauge in the Cascades also reported over 24 inches of precipitation during the first half of December 2025, as indicated in the image by the cyan-color square. The two bow-tie-shaped symbols near Seattle indicate locations where the U.S. Geological Survey measured record river height. The rising water at these river gauges was a result of runoff from the heavy rainfall. The record river height occurred on December 11 and 12.

• *The programs and services offered by George Mason University are open to all who seek them. George Mason does not discriminate on the basis of race, color, religion, ethnic national origin (including shared ancestry and/or ethnic characteristics), sex, disability, military status (including veteran status), sexual orientation, gender identity, gender expression, age, marital status, pregnancy status, genetic information, or any other characteristic protected by law. After an initial review of its policies and practices, the university affirms its commitment to meet all federal mandates as articulated in federal law, as well as recent executive orders and federal agency directives.

GMU’s CEOSR team assists NASA with testing their satellite-based precipitation estimation algorithms. Some tests involve checking the algorithms’ performance during major weather events with large storm totals. Other tests investigate light precipitation or precipitation in arid places where rainfall rarely occurs. In related work, a CEOSR team member wrote a paper in 2014 about the seasonality of rainfall over the Saraha Desert.

The Sahara Desert was in the news during August and September 2024 because of several unusually heavy storms there. Most people live in places where it rains much more often than it does in the Sahara Desert. For this reason, it can be challenging to make sense of news reports of major storms in the Sahara. NASA satellite data and rainfall statistics can provide context, as described below. Areas with significant impacts are shown in green in Figure 1.

Figure 1. A map of North Africa showing areas with notable rainfall in August and September 2024. The boundaries of the Sahara and Sahel are based on Olson et al. (2001). [High-res image]

Deeper in the desert, there are no rain gauges or weather radars for hundreds of miles and there is little plant life or infrastructure to be affected by rain. Satellites do fly over these locations and collect data. Satellite data suggest that, in August and September 2024, the greatest deviation from normal conditions occurred inside the Sahara Desert rather than in the less arid areas at its margins. The relevant statistic is calculated by dividing the rainfall in August and September of 2024 by the amount of rain that falls on average during a full year. High values for this statistic occurred in the red rectangles in Figure 1.

Northern Sudan Flooding in August 2024

In late August 2024, infrastructure was damaged by floods in northeast Sudan, primarily between the Nile River and the Red Sea. Storm runoff caused the collapse of the Arba’at dam late on August 24 (19.8338°N, 36.9396°E; Aljazeera 2024). The dam had supplied drinking water to the city of Port Sudan (United Nations 2024). The late-August floods damaged buildings and roads, making travel difficult between the towns in the region (IFRC 2024; ReliefWeb 2024; Guardian 2024). These impacts occurred, for the most part, between the city of Dongola and the Arb’at Dam, whose locations are marked with a “D” and “A,” respectively, in Figure 2.

Humanitarian agencies described how floods impacted infrastructure but did not specify how much rain fell. For data from prior decades, one can turn to an international rain-gauge archive (GHCNd 2024) to find Sudanese daily rainfall measurements either along the Nile River or Red Sea coast. This archive, however, lacks recent-year data for these rain gauges.

Satellite-based estimates of rainfall are available over most of the globe, including northern Sudan. For example, such estimates are calculated by NASA’s IMERG algorithm and are shown in Figure 2. IMERG stands for Integrated Multi-SatellitE Retrievals for Global Precipitation Measurement (GPM) (Huffman et al. 2023). The IMERG algorithm is used to estimate precipitation rate on a global grid with approximately 7-mile resolution every 30 minutes since January 1998.

Figure 2. IMERG precipitation estimates over northeastern Africa. [High-res image]

Considering how arid the region is, an impressive amount of rain fell during the 24 hours prior to the collapse of the Arba’at dam. IMERG estimated that, upstream of the dam, the Arba’at ravine received an inch of rain in one hour and a total of 2 inches that day. Additional rain had fallen earlier in the month (August 3–11) in northern Sudan. As a result, the monthly total reached 2 to 12 inches (56–316 mm) in many places (left panel of Figure 2). The monthly total for August 2024 was greater than the average-annual accumulation (center panel of Figure 2).

During August 2024, there was also unusually large rain accumulation west of the Arba’at Dam near the Sudan-Libya border. On average, July and August are the rainiest months in this region (Kelley 2014), but the amount that fell in August 2024 was well above average. The statistically unusual rain fell mostly in the 275-by-600-mile red rectangle in Figure 2, an area with no weather radars and no rain-gauge data in the GHCNd international archive. Elsewhere in Figure 2, small white squares mark where there is a rain gauge that has its observations routinely transmitted to the international archive or used in a global-gridded rain-gauge-only data product(GPCC 2024).

Unusually high rainfall totals occurred during August 2024 in the orange, red, and pink areas of the right panel of Figure 2. That month, these areas received over five times as much rain as usual falls during a whole year. Other instances of similarly high rain accumulation are known to have occurred elsewhere in very arid regions (Warner 2004, pg. 348–349). In contrast, it would be inconceivable for a location with a moderate climate to receive five times its average-annual accumulation in a single month. For example, it was considered a 1,000-year storm in 2017 when Hurricane Harvey brought sixty inches of rain to a location near Houston, Texas (Blake and Zelinsky 2018, pg. 6 and 7). Somewhat less, but still large, storm totals occurred over a larger area near Houston (GPM 2020). Even the sixty-inch storm total was not much greater than Houston’s average-annual accumulation. As shown in the center panel of Figure 2, northwestern Sudan has a much lower average-annual accumulation: no more than an inch or so (3–32 millimeters). This statistic means that northwestern Sudan’s average-annual accumulation could be exceeded by just one, fairly mild storm.

Having five years’ worth of rain fall during a single month is a rare occurrence globally, as shown in Figure 3. The black and red areas are where one or two months during the past 24 years had rainfall that exceeded five times average-annual accumulation. Green areas are where a less extreme threshold was reached: at least one 12-month period (January through December) had five times more rainfall than the average year. The top panel of Figure 3 shows where these thresholds were exceeded during 2001 to 2022, and the bottom panel covers January 2023 through September 2024. Figure 3 omits places where IMERG never saw these thresholds exceeded, such as China, Australia, and the nearby ocean.

Figure 3. Locations where IMERG found that a single month or year had precipitation in excess of five times that location’s 2001–2022 average-annual accumulation. [High-res image]

Evaluating the whole globe during the past 24 years, IMERG reports that it is primarily within the southeast portion of the Sahara Desert, along the southern tip of the Arabian Peninsula, and over nearby bodies of water that a monthly accumulation is ever greater than five times a location’s average-annual accumulation. The places where this threshold is reached tend to be very arid and near less-arid regions where strong storms are more common. For example, the Sahara Desert is impacted by a few of the intense convective systems that move westward across tropical Africa. The southern end of the Arabian Peninsula is occasionally impacted by tropical cyclones that form over the Indian Ocean (e.g., GPM 2015, 2018).

So far, it has been shown that the August 2024 rain accumulation in the Sahara Desert near the Sudan-Libya border was a globally rare event because it delivered five times that region’s average-annual accumulation.

Next, we take a look at one of the rainy days in August 2024 to see if the storm clouds over northern Sudan look any different than ordinary storms that occur frequently elsewhere in the world. An easy way to compare satellite observations of storms is to use NASA’s WorldView website. Figure 4 shows a true-color satellite image taken from WorldView and subsequently annotated using other datasets.

Figure 4. A true-color MODIS snapshot of northern Sudan during several rainy days in late August 2024. The southeast boundary of the Sahara Desert is marked in orange or pink based on Olson et al. (2001) or AAG (2013). [High-res image]

Figure 4 shows the clouds over northern Sudan at about 2 PM local time (1158 UTC) on August 26, 2024. The isolated white blobs that cast dark shadows are tall convective systems that are producing rain and that are under 50 miles across. The larger areas of less bright clouds are generally not raining and not as tall. This true-color imagery and cloud-height information come from the MODIS instrument on NASA’s Aqua satellite and the VIIRS instrument on NOAA’s NPP satellite. The rainfall rate was estimated using Japan’s Dual-frequency Precipitation Radar (DPR) that flies on NASA’s Global Precipitation Measurement (GPM) satellite.

The GPM satellite’s instruments estimated that the Sudanese convective cells were raining 3–10 millimeters per hour, which are rates typical of run-of-the-mill convective cells in locations with moderate climates. At 2 PM on August 26, the NPP satellite’s OMPS instrument saw airborne dust to the west of the northern Sudan storms (GES DISC 2024).

Two facts suggest that at least some of the storm cells in this system were vigorous. For one thing, GPM radar observations suggested that one of these convective cells had strong updrafts. Strong updrafts were likely present because the radar saw ice-phase hydrometeors lifted to the tropopause, which means a height of about 16.5 km above sea level over Sudan (Tegtmeier et al. 2020, Fig. 8b). The location of this strong convective cell is marked with an “X” in Figure 4.

The other relevant fact is that lightning was observed in many of these cells. Strong updrafts are typically one of the necessary ingredients for lightning to form. At 2 PM on August 26, the European MTG-1 geostationary satellite saw lightning in many convective cells over northern Sudan (small yellow crosses in Figure 4), but no lightning in the clouds further north over Egypt (Weather.us 2024). In Figure 4, black circles mark Sudanese cities impacted by the late-August rainfall. The red circle and line through it represent the Arb’at dam and the stream it dams. As mentioned previously, the Arb’at dam collapsed on August 24 due to large rain accumulation.

It is unclear precisely which of the Sudanese storm cells were in the Sahara Desert because the desert’s boundaries are subject to debate. Orange or pink lines in Figure 4 mark the southeastern edge of the Sahara Desert according to the work of Olson et al. (2001) or AAG (2013), respectively.

Flooding Near the Atlas Mountains in September 2024

Two weeks later and 2,000 miles away, there was another multi-day rain episode with large accumulations on the opposite side of the Sahara Desert. During September 5–8, 2024, IMERG estimated that about 1–4 inches (32–100 millimeters) fell over the Atlas Mountains of Morocco.

The Associated Press reported a similar or greater amount of rain than did IMERG in two locations in the southern foothills of the Atlas mountains (Metz and Ahmed 2024). The Associated Press reported that the Moroccan government measured 100 millimeters of rainfall in one day at Tagounite (29.978°N, 5.583°W) and 250 millimeters over two days at Tata (29.754°N, 7.973°W). Presumably, these measurements were made by the government-operated rain gauges in these cities. These rain gauges are the only government-operated gauges in a region spanning 300 miles (Directorate of Meteorology 2024).

Regardless of how much rain actually fell in early-September, drone and satellite images showed significant runoff from these storms (Earth Observatory 2024). It is clear that significant rainfall did occur, at least by the standards of this arid region (Armon et al. 2024, Fig. 6). Runoff from these storms filled riverbeds and lakebeds that had been dry for decades south of the Atlas Mountains near the Morocco-Algeria border. Photographs of some of these lakes were taken by drones nearly a month after the rain storms and widely circulated online (Metz and Ahmed 2024).

The MODIS instrument on NASA’s Terra satellite collected the two false-color images in Figure 5. The color table is designed to highlight surface water such as streams, lakes, and flooded land. These false-color images are generated from bands 7, 2, and 1 of the MODIS instrument (Earthdata 2024). The top image was observed a few days after the early-September rain storms, and it shows streams in cyan and lakes in dark blue or black. The bottom image was observed a month after the rain storms, and it shows streambeds dry once again but lakes still filled. In Figure 5, dotted white lines mark intermittent streams and white circles mark eight dry lakebeds filled by runoff from the early-September rain storms.

Figure 5. False-color MODIS imagery of the Morocco-Algeria border that was observed either a few days (top) or a month (bottom) after the early-September 2024 storms. The Sahara Desert extends further to the northwest in the classification system of AAG (2013) (yellow contour) vs. that of Olson et al. (2001) (orange contour). [High-res image]

News stories did not tell the full story of the September 2024 rain storms, and strident headlines might mislead people into thinking that a large portion of the Sahara flooded. For example, CNN lead with “Sahara Desert flooded for the first time in decades” (CNN 2024). An internet search on the phrase “Sahara Desert flooding” found news stories with headlines that claim the early-September rain in Morocco’s Atlas Mountains caused the first floods in the Sahara Desert in 50 years.

For one thing, the eight lakes filled by runoff south of the Atlas Mountains covered less than 1/1000^th of this region, as can be seen in Figure 5.

Second, the boundaries of the Sahara Desert are subject to debate. It is open to interpretation how many of these eight lakes were actually within the Sahara Desert rather than in the surrounding steppes. In some land-classification systems, only one of the eight lakes was in the Sahara Desert (Olson et al. 2001; Burgess et al. 2006; Dinerstein et al. 2017; OneEarth 2024). In other land-classification systems, most or all of these lakes were in the Sahara Desert (AAG 2013; Peel et al. 2007, Fig. 4; Warner 2004, pg. 80 and 84; and Nicholson 2011, pg. 294).

Last, the early-September floods in the southern foothills of Morocco’s Atlas Mountains were not the first that the Sahara Desert had experienced in 50 years. On the contrary, flooding had occurred just two weeks earlier in northeastern Sudan, as discussed in the previous section of the present blog post. Northeastern Sudan and the southern foothills of the Atlas Mountains are both at the edge of the Sahara Desert or in the steppes just outside it, depending on what land-classification system you use.

News stories about the early-September rainfall in the vicinity of the Morocco-Algeria border did not comment on the total accumulation for the whole month. In the left panel of Figure 6, the monthly rainfall is seen to reach about 4 inches (100 millimeters) about 300 miles south of the Atlas Mountains, according to IMERG. This area is marked by the red rectangle in Figure 6, and it is far south of the impacts that the media covered from the early-September storms. In the red rectangle, the September 2024 rainfall was twice average-annual accumulation, which is shown in the center panel. The right panel of Figure 6 shows the multiple of average-annual accumulation that fell during September 2024.

Within the red box in Figure 6, there are salt flats (dry lakebeds) that flooded with storm runoff between September 4 and 9, 2024. The resulting lakes are visible in MODIS imagery on the WorldView website and are located within 27.081–27.606ºN, 5.782–4.865ºW. One of the salt flats is called Sebkha Ain Belbela (27.556ºN, 5.295ºW, Teach with GIS 2024). Perhaps, news stories did not mention this flooding because the region appears to be uninhabited. The location of these salt flats is marked by “S” in Figure 6. Also marked are locations further north that were mentioned in the news stories about these storms: Merzouga and Lake Sriji (M), storm runoff reported by the NASA Earth Observatory (E) (Earth Observatory 2024), Tagounite (T), Lake Iriqui (I), and Tata (A).

Figure 6. IMERG precipitation estimates near the Morocco-Algeria border. [High-res image]

The Partial “Greening” of North Africa in 2024

One way to measure the significance of large rain accumulation is by its impact on vegetation. Greater-than-average rainfall may increase crop yields or damage crops. Unusually large accumulation may also increase how green the Earth looks from space. Since the early 1980s, satellites have routinely measured vegetation density using a variable called NDVI, the normalized difference vegetation index (Tucker et al. 1991; NCEI 2024). To calculate NDVI, satellite observations at two wavelengths are compared. NDVI typically has values of 0.1–0.2 for barren ground and 0.7–0.8 for dense vegetation (Ding et al. 2016).

On September 19, 2024, the Washington Post reported that “parts of the Sahara Desert are dramatically greener than usual” because of higher-than-average rainfall (Stillman 2024). The Washington Post supported its claim by showing a map of where NDVI changed between the first half of September 2024 and one year earlier. The figure in the Washington Post article, however, showed that the Sahara Desert’s vegetation was essentially the same in September 2024 and one year earlier. It was in the Sahel, not the Sahara, where the Washington Post figure showed increased vegetation.

To further explore the greening of North Africa, consider the top panel of Figure 7. That panel shows how much the early-October 2024 NDVI deviated from the long-term average for that time of year (USGS 2024). It makes sense to look at NDVI to measure the effect of excess rain from prior months. There can be a delay of a month or two between above-average rainfall and a subsequent increase in plant density (Ugbaje and Bishop 2020). One can check the timing of NDVI changes using the GIMMS data browser.

When discussing NDVI change specifically in the Sahara Desert, it is worth noting that various land-classification systems agree for the most part about what land falls within the Sahel or the Sahara Desert, but they disagree about the exact boundaries (e.g., Olson et al. 2001 vs. AAG 2013). Keeping this in mind, one can still say that, in early October 2024 (day of year 273–288), the vegetation over most of the Sahara Desert was slightly more scarce than average for that time of year, according to MODIS observations. Average conditions were calculated using data from 2000–2023. This decrease in vegetation is indicated by light red in the top panel of Figure 7. NDVI was higher than normal in two woodland areas that are surrounded by the Sahara Desert. These two woodland areas are in Chad’s Tibesti Mountains and Algeria’s Hogar Mountains. The increased vegetation is indicated by shades of green in the top panel of Figure 7.

Figure 7. Terra MODIS NDVI during the first two weeks of October 2024 compared to the median NDVI at the same time of the year during 2000–2023. The yellow and green lines indicate the boundaries of the Sahara Desert and the Sahel (Olson et al. 2001). [High-res image]

In early October 2024, the Sahel’s NDVI was much higher than average for this time of year. The Sahel is the acacia-tree savanna south of the Sahara Desert. The large increase in vegetation is indicated by dark green in the top panel of Figure 7.

The bottom panel of Figure 7 shows the rainfall anomaly estimated by IMERG during the three-month period of July through September 2024. Blue or red indicate where there was more or less rain than usual during these three months of 2024.

Comparing the top and bottom panels of Figure 7, one sees that, in places that already had a moderate amount of vegetation, the excess rainfall during July through September 2024 were associated with denser-than-average vegetation in October 2024.  Such places are the savannas, steppes, and woodlands located in the Sahel, Atlas Mountains, or the Hoggar and Tibesti Mountains.

In contrast, October 2024 did not have denser vegetation than normal in some nearly barren locations, even though those places did receive more rain than usual during the preceding three months. Three such locations are labeled a, b, and c in the bottom panel of Figure 7. The imperfect correlation between the preceding three months of rain and vegetation density indicates that other factors are at work. Such factors might include longer-term rainfall, air temperature, evaporation rate, soil type, and differences in how various ecosystems respond to rainfall.

Conclusion

It can be difficult to make sense of major rain storms and flooding that is occasionally reported in the Sahara Desert. The preceding discussion used NASA data and other datasets to provide context for three news stories published about rainfall in August and September 2024. Long-term averages, single overflights of storms, and monthly statistics were discussed. The only way to see the big picture over this part of the world is to use satellite data because much of the Sahara Desert lacks rain gauges and weather radars. For seeing the big picture, NASA’s IMERG rainfall-estimation algorithm is helpful because it combines infrared and microwave observations from many satellites and calibrates those estimates with rain gauges where available.

In one part of the Sahara Desert, the rain accumulation during just August 2024 was more than five times as great as what usually falls there during a full year. At the edge of the Sahara Desert or in the steppes adjacent to it, the rain accumulation was less statistically extreme (as a fraction of average-annual accumulation), yet it was still significant. In these edge cases, the rainfall was sufficient to produce floods that damaged infrastructure and runoff that filled dry lakebeds.

Satellites saw little change in the nearly barren Sahara Desert despite the higher-than-average rainfall there in 2024. In contrast, satellites saw a dramatic increase in vegetation in October 2024 in the Sahel. The Sahel is a less arid region of acacia-tree savanna that lies to the south of the Sahara Desert. Most of the Sahel received higher than normal rainfall during July through September 2024, preceded the increase in vegetation.

Credits: Text and images by Owen Kelley (GMU/NASA). Several people provided feedback on an earlier version of this story including Jim Tucker (NASA), George Huffman (NASA), Jorge Pinzon (SSAI/NASA), Ed Pak (SSAI/NASA), and Jason West (KBR/NASA).

References

AAG, 2013: A New Map of Standardized Terrestrial Ecosystems of Africa. special supplement to the African Geographical Review, Association of American Geographers, 24 pp., https://pubs.usgs.gov/publication/70045097. For a TIFF file to display this land classification, see Sayre, R., 2013: Africa Terrestrial Ecosystems. USGS, https://www.sciencebase.gov/catalog/item/63e4622ed34e9fa19a9c007a.

Aljazeera, August 26, 2024: More than 60 reported killed in northeast Sudan dam collapse, https://www.aljazeera.com/news/2024/8/26/a-dam-collapsed-in-northeast-s…. Mentions that local media reported the dam burst on Saturday night (August 24, 2024).

Armon, M., A. Jan de Vries, F. Marra, N. Peleg, and H. Wernli, 2024: Saharan rainfall climatology and its relationship with surface cyclones. Weather and Climate Extremes, 43, https://doi.org/10.1016/j.wace.2023.100638.

Blake, E. S., and D. A. Zelinsky, 2018: Tropical Cyclone Report Hurricane Harvey (AL092017). National Hurricane Center., 77pp., https://www.nhc.noaa.gov/data/tcr/AL092017_Harvey.pdf.

Burgess, N., et al., 2004: Terrestrial Ecoregions of Africa and Madagascar: A Conservation Assessment. Island Press, 526 pp., https://www.researchgate.net/publication/292588815_Terrestrial_eco-regi….

CNN, October 14, 2024: Meteorologist explains rare weather event that left lakes in one of the driest places on earth. video, featuring meteorologist Derek Van Dam, https://www.cnn.com/2024/10/14/weather/video/sahara-weather-derek-van-d…. See also CBS Evening News, October 12, 2024: Sahara Desert is flooding after receiving a year’s worth of rain in 2 days, https://www.youtube.com/watch?v=xmiua1GWWPc.

Dinerstein, E., et al., 2017: An ecoregion-based approach to protecting half of the terrestrial realm. BioScience, 67, https://academic.oup.com/bioscience/article/67/6/534/3102935.

Ding, Y., X. Zheng, K. Zhao, X. Xin, and H. Liu, 2016: Quantifying the Impact of NDVIsoil Determination Methods and NDVIsoil Variability on the Estimation of Fractional Vegetation Cover in Northeast China. Remote Sensing, 29, doi:10.3390/rs8010029.

Directorate of Meteorology, 2024: map of rain gauge locations. Moroccan government’s General Directorate of Meteorology, https://www.marocmeteo.ma/fr/observation.

Earthdata, February 23, 2024: Worldview Snapshots: What do the different band combinations mean for the corrected reflectance images? NASA Earthdata Forum, https://forum.earthdata.nasa.gov/viewtopic.php?t=5230.

Earth Observatory, September 10, 2024: A deluge for the Sahara. https://earthobservatory.nasa.gov/images/153320/a-deluge-for-the-sahara. The MODIS (Moderate Resolution Imaging Spectroradiometer) on NASA’s Terra satellite captured these false-color images comparing 14 August 2024 to 10 Sept 2024 in the vicinity of the boundary of Morocco and Algeria.

Guardian (UK), August 26, 2024: Flood surge in Sudan bursts dam, destroying villages and killing dozens. https://www.theguardian.com/world/article/2024/aug/26/flood-surge-in-su….

GES DISC, 2024: Ozone Mapping and Profiling Suite (OMPS) Nadir-Mapper (NM) instrument on the Suomi-NPP satellite. L2 Aerosol Index swath orbital V2 data product, doi:10.5067/40L92G8144IV.

GHCNd, 2024: Global Historical Climatology Network daily. https://www.ncei.noaa.gov/products/land-based-station/global-historical…. A global archive of rain-gauge 24-hour observations spanning over a century.

GPCC, 2024: Global Precipitation Climatology Centre gridded rain-gauge data product. GPCC daily or monthly first-guess daily product. Documentation: https://gpcc.dwd.de/ and https://opendata.dwd.de/climate_environment/GPCC/html/gpcc_monitoring_v…. Download near real-time files from https://opendata.dwd.de/climate_environment/GPCC/.

GPM, November 10, 2015: Yemen hit by 2nd tropical cyclone in a week. Extreme Weather News, NASA Global Precipitation Measurement (GPM), https://gpm.nasa.gov/extreme-weather/yemen-hit-2nd-tropical-cyclone-week.

GPM, June 5, 2018: Two cyclones hit Somalia and Oman within a week. Extreme Weather News, NASA Global Precipitation Measurement (GPM), https://gpm.nasa.gov/extreme-weather/two-cyclones-hit-somalia-and-oman-….

GPM, September 16, 2020. Hurricane Sally brings extremely heavy rain to the northern Gulf Coast. Extreme Weather News, NASA Global Precipitation Measurement (GPM), https://gpm.nasa.gov/applications/weather/hurricane-sally-extremely-hea…. In the section titled “Hurricane Sally’s Rainfall In Context”, see the comparison of Hurricane Harvey (2017) over Houston, Texas, and Hurricane Sally (2020) over Florida. See also https://gpm.nasa.gov/extreme-weather/nasas-imerg-shows-rainfall-accumul….

Huffman, G. J., et al., 2023: NASA Global Precipitation Measurement (GPM) Integrated Multi-satellitE Retrievals for GPM (IMERG) Version 07. Algorithm theoretical basis document (ATBD), NASA, https://pps.gsfc.nasa.gov/Documents/IMERG_V07_ATBD_final.pdf.

IFRC, September 5, 2024: Flood Sudan 2024. update by A. Aghaei, International Federation of Red Cross and Red Crescent Societies (IFRC), https://go.ifrc.org/field-reports/17233.

Kelley, O. A., 2014: Where the least rainfall occurs in the Sahara Desert, the TRMM radar reveals a different pattern of rainfall each season. J. Climate, 27, DOI: 10.1175/JCLI-D-14-00145.1.

Metz, S., and B. Ahmed, October 8, 2024: Water gushes through sand dunes after a rare rainfall in the Sahara Desert. Associated Press, https://apnews.com/article/sahara-flooding-oasis-rain-morocco-2b6c28d24…. The news article shows lakes photographed on October 2 and mention 24-hour 100 mm storm-total in village of Tagounite, Morocco (29.979°N, 5.584°W), but that rain must have been the early-September storm a month before the photo was taken because no major storm after that according to IMERG data and visible/infrared data from other satellites.

NCEI, 2024: NOAA Climate Data Record (CDR) of AVHRR Normalized Difference Vegetation Index (NDVI), Version 5. https://www.ncei.noaa.gov/access/metadata/landing-page/bin/iso?id=gov.n….

Nicholson, S. E., 2011: Dryland Climatology. Cambridge University Press, 516pp.

Olson, D. M., et al., 2001: Terrestrial ecoregions of the world: A new map of life the Earth. BioScience, 51, 933–938, https://academic.oup.com/bioscience/article/51/11/933/227116. For display in QGIS, download shapefiles of ecoregions from the USGS: https://www.sciencebase.gov/catalog/item/508fece8e4b0a1b43c29ca22.

OneEarth, 2024: Announcing the release of Ecoregion Snapshots. https://www.oneearth.org/announcing-the-release-of-ecoregion-snapshots/. See similar ecoregions in https://ecoregions.appspot.com/.

Relief Web, September 6, 2024: Sudan Floods 2024 – DREF Operation (MDRSD034). situation report, https://reliefweb.int/report/sudan/sudan-floods-2024-dref-operation-mdr….

Stillman, D., September 19, 2024: Why parts of the Sahara Desert are turning green this month. Washington Post, https://www.washingtonpost.com/weather/2024/09/19/sahara-desert-rain-af….

Teach with GIS, 2024: ESRI online GIS layer. Documentation at https://teachwithgis.co.uk. Download data at https://www.arcgis.com/apps/Profile/index.html?appid=f0a2a2a3e1964129b2…. See also the online geographic name database at https://geonames.nga.mil/geonames/GeographicNamesSearch/.

Tegtmeier, S., et al., 2020: Temperature and tropopause characteristics from reanalyses data in the tropical tropopause layer. Atmos. Chem. Phys., 20, https://doi.org/10.5194/acp-20-753-2020.

Tucker, C. J., H. E. Dregne, and W. W. Newcomb, 1991: Expansion and contraction of the Sahara Desert from 1980 to 1990. Science, 253, 299–301.

Ugbaje S. U., and T. F. A. Bishop, 2020: Hydrological Control of Vegetation Greenness Dynamics in Africa: A Multivariate Analysis Using Satellite Observed Soil Moisture, Terrestrial Water Storage and Precipitation. Land, 9, doi:10.3390/land9010015.

United Nations, 26 August 2024: SUDAN: Collapse of Arba’at Dam in Port Sudan, Red Sea State Flash Update No. 01 (26 August 2024). United Nations Office for the Coordination of Humanitarian Affairs (UNOCHA). https://reports.unocha.org/en/country/sudan/flash-update/6cwSvn6aEpmuMQ….

USGS, 2024: MOD13C1 v061 MODIS/Terra Vegetation Indices 16-Day L3 Global 0.05 Deg CMG. For documentation, click on the documentation link at https://lpdaac.usgs.gov/products/mod13c1v061/. Download data from https://e4ftl01.cr.usgs.gov/MOLT/MOD13C1.061/.

Warner, T., 2004: Desert Meteorology. Cambridge University Press, 595pp.

Weather.us, 2024: online browsing of METOSAT’s MTG-L1 lightning data and infrared observations. https://weather.us/satellite/northern/satellite-color-15min–li/2024082….

• *The programs and services offered by George Mason University are open to all who seek them. George Mason does not discriminate on the basis of race, color, religion, ethnic national origin (including shared ancestry and/or ethnic characteristics), sex, disability, military status (including veteran status), sexual orientation, gender identity, gender expression, age, marital status, pregnancy status, genetic information, or any other characteristic protected by law. After an initial review of its policies and practices, the university affirms its commitment to meet all federal mandates as articulated in federal law, as well as recent executive orders and federal agency directives.