African Rainfall Research Articles

Synchronization of the Recent Decline of East African Long Rains and Northwestern Eurasian Warming

AbstractThe East African March–April–May (MAM, “long rains”) precipitation decline in recent decades remains a puzzle marked by various proposed large‐scale drivers. Here, the interannual variability of the long rains and their recent drying trend are examined using global model simulations and observations. Comparison of a control simulation and re‐initialized simulations in which land‐surface feedback is suppressed shows that much of the long rains deficit experienced between 1980 and 2014 is synchronized with the warming of the Northwestern Eurasian landmass. In agreement with the modeling results, multiple observational data sets reveal a strong negative correlation between MAM mean East African rainfall amount and the surface temperature over Northwestern Eurasia. Idealized simulations further indicate that warming in Northwestern Eurasia weakens the regional Hadley Cell and diverts the monsoonal transport of moisture away from Eastern Africa toward Europe and southern Africa, highlighting the role of remote land surface warming on the observed precipitation decline.

The modulation of the MJO and tropical cyclone impact on East African rainfall by the Mascarene high during the 2018 “long rains”

The modulation of the MJO and tropical cyclone impact on East African rainfall by the Mascarene high during the 2018 “long rains”

Observations of enhanced rainfall variability in Kenya, East Africa

Understanding local patterns of rainfall variability is of great concern in East Africa, where agricultural productivity is dominantly rainfall dependent. However, East African rainfall climatology is influenced by numerous drivers operating at multiple scales, and local patterns of variability are not adequately understood. Here, we show evidence of substantial variability of local rainfall patterns between 1981 and 2021 at the national and county level in Kenya, East Africa. Results show anomalous patterns of both wetting and drying in both the long and short rainy seasons, with evidence of increased frequency of extreme wet and dry events through time. Observations also indicate that seasonal and intraseasonal variability increased significantly after 2013, coincident with diminished coherence between ENSO (El Nino Southern Oscillation) and rainfall. Increasing frequency and magnitude of rainfall variability suggests increasing need for local-level climate change adaptation strategies.

Constraining Plio‐Pleistocene Shifts in Northwest African Hydroclimate, Ecosystem Distributions, and Marine Productivity: New Paleo‐Records Across the Mid‐Pleistocene Transition

AbstractNorthwest Africa transitioned from a wet/vegetated landscape toward drier/sparser conditions sometime between the late‐Pliocene and the late‐Pleistocene. However, our understanding of the precise timing and nature of this transition is hampered by a paucity of paleo‐records which bridge these two intervals. Here we report new plant‐wax isotope as well as dust and opal flux records from the relatively brief interval ∼1.1–1.0 million years ago (Ma) to evaluate the astronomical timescale controls of Northwest African hydroclimate and vegetation during the Mid‐Pleistocene Transition (MPT) and, in context with published records, the drivers of long‐term climate and ecological trends over the Plio‐Pleistocene. The tempo and amplitude of the Northwest African monsoon rainfall swings closely track low latitude insolation forcings over the last 5 Ma. However, we demonstrate that a pronounced mean state decline in monsoon strength likely occurred following the MPT most likely instigated by increasing Atlantic meridional sea surface temperature gradients or declines in the strength of the meridional overturning circulation. The northward extent of vegetation does not track changes in monsoon strength over the Plio‐Pleistocene and thus may be more strongly influenced by changes in monsoon rainfall extent or ecosystem disturbances. Progressively diminished dust fluxes following a decline in monsoon strength after 1.0 Ma is consistent with reduced production and subsequent depletion of fine‐grained sediments in the Sahara. Synchroneity between dust and opal fluxes across timescales suggests nutrient delivery to the surface ocean via dust plays a key role in marine primary productivity off the coast of Northwest Africa.

Mid-Holocene West African monsoon rainfall enhanced in EC-Earth simulation with dynamic vegetation feedback

Proxy records have shown that the Mid-Holocene was a period of humid conditions across West Africa, with an enhanced West African Monsoon (WAM) and vegetated conditions in areas currently characterized by desert, often referred to as the Green Sahara. However, General Circulation Models regularly struggle with recreating this strengthened Mid-Holocene monsoon in West Africa. Vegetation feedbacks has long been viewed as an essential process modulating the monsoon variability in West Africa, and simulations using prescribed vegetation to recreate a Green Sahara have shown a strengthened WAM and increased rainfall. However, simulations with prescribed vegetation in Sahara represent an idealized vegetation cover and do not take any environmental heterogeneity into account. Furthermore, this only represents a one-directional forcing by the vegetation on the climate rather than the full vegetation feedback. To address this, we have simulated the Mid-Holocene (~ 6 ka) climate using the Earth System Model EC-Earth3-Veg. The results show that coupled dynamic vegetation reproduces an apparent enhancement of the WAM, with the summer rainfall in the Sahel region increasing by 15% compared to simulations with a prescribed modern vegetation cover. Vegetation feedbacks enhance the warming of the Sahara region, deepens the Sahara Heat Low, results in increased rainfall and strengthens monsoonal flow across West Africa. However, the enhancement is still below what can be viewed in proxy reconstructions, highlighting the role of model limitation and biases and the importance of investigating other processes, such as the interactive aerosol-albedo feedback.

Evaluation of Satellite-Based Rainfall Estimates against Rain Gauge Observations across Agro-Climatic Zones of Nigeria, West Africa

Satellite rainfall estimates (SREs) play a crucial role in weather monitoring, forecasting and modeling, particularly in regions where ground-based observations may be limited. This study presents a comprehensive evaluation of three commonly used SREs—African Rainfall Climatology version 2 (ARC2), Climate Hazards Group Infrared Precipitation with Station data (CHIRPS) and Tropical Application of Meteorology using SATellite data and ground-based observation (TAMSAT)— with respect to their performance in detecting rainfall patterns in Nigeria at daily scales from 2002 to 2022. Observed data obtained from the Nigeria Meteorological Agency (NiMet) are used as reference data. Evaluation metrics such as correlation coefficient, root mean square error, mean error, bias, probability of detection (POD), false alarm ratio (FAR), and critical success index (CSI) are employed to assess the performance of the SREs. The results show that all the SREs exhibit low bias during the major rainfall season from May to October, and the products significantly overestimate observed rainfall during the dry period from November to March in the Sahel and Savannah Zones. Similarly, over the Guinea Zone, all the products indicate overestimation in the dry season. The underperformance of SREs in dry seasons could be attributed to the rainfall retrieval algorithms, intensity of rainfall occurrence and spatial-temporal resolution. These factors could potentially lead to the accuracy of the rainfall retrieval being reduced due to intense stratiform clouds. However, all the SREs indicated better detection capabilities and less false alarms during the wet season than in dry periods. CHIRPS and TAMSAT exhibited high POD and CSI values with the least FAR across agro-climatic zones during dry periods. Generally, CHIRPS turned out to be the best SRE and, as such, would provide a useful dataset for research and operational use in Nigeria.

Performance‐based evaluation of NMME and C3S models in forecasting the June–August Central African rainfall under the influence of the South Atlantic Ocean Dipole

AbstractIn this study, hindcasts from eight Copernicus Climate Change Service (C3S) and three North American Multi‐Model Ensemble (NMME) operational seasonal forecast systems, based on dynamical climate models, are employed to investigate the influence of the South Atlantic Ocean Dipole (SAOD) on the predictive skill of Central Africa (CA) rainfall. The focus is primarily on the June–July–August season for 1993–2016. The findings reveal that, when regionally averaged, all models exhibit positive skill in predicting CA rainfall, except for the Geophysical Fluid Dynamics Laboratory (GFDL‐SPEAR) model. Notably, there are significant spatial variations in skill across different regions. Model performance is particularly low (high) in the Central African Republic and Congo Basin (Gabon and Chad) and tends to deteriorate with increasing lead‐time. Models that demonstrate a strong connection between SAOD and CA rainfall tend to exhibit better predictive skills in forecasting rainfall, in contrast to models with weaker connections. This leads to a significant in‐phase relationship between the predictive skills of rainfall and the strength of the SAOD–rainfall connection among the models. Furthermore, the atmospheric circulation responding to SST forcing associated with the El Niño–Southern Oscillation exerts a significant influence on the robust atmospheric circulation associated with the climatological mean of SST over the SAO. This suggests that mean state bias in the SAO/equatorial Pacific region plays a role in modulating the strength of the simulated SAOD–CA rainfall connection and, consequently, the prediction skill of CA rainfall. In general, both NMME and C3S models appear to be valuable tools capable of providing essential seasonal information several months in advance. These insights can aid decision‐makers in the region in making informed decisions regarding adaptation and mitigation measures.

Uncertainty of the Simulated Mid-Pliocene Changes of Sahel Summer Rainfall in the PlioMIP2 Multimodel Ensemble

Abstract The mid-Pliocene (approximately 3.3–3.0 Ma) was one of the past geological warm periods. Since this past warmer climate was in many respects comparable to near future warming projections, how the regional monsoonal rainfall changed during the mid-Pliocene is an important scientific and socioeconomic concern. Based on phase 2 of the Pliocene Model Intercomparison Project (PlioMIP2), this work examines the simulated changes of Sahel summer rainfall during the mid-Pliocene. The results show the considerable intermodel uncertainty of the simulated mid-Pliocene Sahel rainfall changes in the PlioMIP2 multimodel ensemble, which is due to the uncertainties of both the simulated dynamic and thermodynamic responses to this past warmer climate. In particular, we find that the intermodel spread in the simulated northern North American warming is a major source of the uncertainty of the mid-Pliocene Sahel summer rainfall changes through two processes. One is a direct dynamic process: a stronger warming over northern North America could enhance the meridional temperature gradient between the extratropical and tropical regions, inducing an interhemisphere energy imbalance of the atmosphere. This could lead to a northward shift of the intertropical convergence zone, strengthening the Western African summer monsoon (WASM) circulation and Sahel summer rainfall. Another is an indirect thermodynamic process: the strengthened WASM circulation could further induce anomalous moisture convergence over the Sahel region, increasing local atmospheric moisture at the low-level troposphere, in favor of a wetter Sahel. Our results suggest that an improved warming simulation over northern North America is essential for the hydrological cycle simulation around the Sahel in the mid-Pliocene warmer climate. Significance Statement The Sahel summer rainfall change in a global warmer climate is a widespread scientific and socioeconomic issue because of its important impacts on regional agriculture, ecosystems, food security, water resources, and even cultural environment. However, the present study identifies a nonnegligible intermodel uncertainty of the simulated Sahel rainfall changes during the mid-Pliocene (one of the most recent geological warm periods in Earth’s history) in the Pliocene Model Intercomparison Project phase 2 (PlioMIP2) multimodel ensemble. In particular, such an intermodel uncertainty of the simulated Sahel rainfall changes during the mid-Pliocene is attributed to that of the simulated northern North America warming via both the direct dynamic process and the indirect thermodynamic process. This is quite different from the uncertainty source of near-future projections of Sahel summer rainfall changes. The present results improve our understanding of the underlying physics of the hydrological cycle change around the Sahel region in the mid-Pliocene warmer climate, with implications for the future projections of regional monsoonal rainfall.

Orbitally forced and internal changes in West African rainfall interannual-to-decadal variability for the last 6000 years

Recent variability in West African monsoon rainfall (WAMR) has been shown to be influenced by multiple ocean–atmosphere modes, including the El Niño Southern Oscillation, Atlantic Multidecadal Oscillation and the Interdecadal Pacific Oscillation. How these modes will change in response to long term forcing is less well understood. Here we use four transient simulations driven by changes in orbital forcing and greenhouse gas concentrations over the past 6000 years to examine the relationship between West African monsoon rainfall multiscale variability and changes in the modes associated with this variability. All four models show a near linear decline in monsoon rainfall over the past 6000 years in response to the gradual weakening of the interhemispheric gradient in sea surface temperatures. The only indices that show a long-term trend are those associated with the strengthening of the El Niño Southern Oscillation from the mid-Holocene onwards. At the interannual-to-decadal timescale, WAMR variability is largely influenced by Pacific–Atlantic – Mediterranean Sea teleconnections in all simulations; the exact configurations are model sensitive. The WAMR interannual-to-decadal variability depicts marked multi-centennial oscillations, with La Niña/negative Pacific Decadal Oscillation and a weakening and/or poleward shift of subtropical high-pressure systems over the Atlantic favoring wet WAMR anomalies. The WAMR interannual-to-decadal variability also depicts an overall decreasing trend throughout the Holocene that is consistent among the simulations. This decreasing trend relates to changes in the North Atlantic and Gulf of Guinea Sea Surface Temperature variability.

Atmospheric response to seasonal changes in sea surface temperature during the boreal summer in the Tropical Atlantic

We investigate the atmospheric response to seasonal variations in sea surface temperature (SST) in the eastern tropical Atlantic during the boreal summer, using the Weather Research and Forecasting (WRF) regional atmospheric model. Three ensembles of 11 simulations each are produced with different SST forcings: the control ensemble (CTL) uses the observed climatology of the SST in 2000–2009, while the Frozen North (FzN) and Frozen South (FzS) experiments block the seasonal warming or cooling of the SST from June onwards in a region confined to the eastern tropical Atlantic. The result is a cold SST anomaly in the northeastern tropical Atlantic off the coasts of Senegal and Mauritania in FzN, and a warm anomaly in the southeastern region (Gulf of Guinea and the cold tongue zone in the equatorial Atlantic) in FzS. Comparison with CTL reveals significant impacts of these SST anomalies on the position and intensity of the marine intertropical convergence zone (ITCZ) and on West African rainfall during July and August. Over the ocean, the cold anomaly in NETA suppresses convection on the northern side of the ITCZ (north of 10∘\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^\\circ$$\\end{document}N), while the warm anomaly in the Gulf of Guinea strengthens convection on its southern flank. The latter is also leading to a sharp increase in precipitation in the coastal regions to the northeast of the Gulf of Guinea. These changes are clearly due to variations in surface pressure gradients and the divergence of low-level moisture in response to SST anomalies, which in turn induce changes in deep atmospheric convection through thermodynamic feedback. On the continent, a substantial reduction in precipitation is observed in the western Sahel (particularly Senegal) following the cold anomaly in NETA, and in the eastern Sahel following the warm anomaly in the Gulf of Guinea: both are explained by a positive anomaly in the divergence of moisture transport in the upper troposphere, associated with an acceleration of the African easterly jet along its southern edge. However, the mechanism by which the SST anomalies create this acceleration in both experiments remains to be elucidated.

Analysis of the Hadley cell, subtropical anticyclones and their effect on South African rainfall

This study investigates the behaviour of subtropical high-pressure systems and the Hadley cell, which affect the weather of South Africa, using the ERA-Interim database and ensemble of 14 global circulation models from Phase 6 of the Coupled Model Intercomparison Project (CMIP6). Mass stream function was used to represent the Hadley cell. To analyse the behaviour of the subtropical anticyclones, monthly sea level pressure, the 1018 hPa isobar and the maximum isobar in the study area were used. The seasonal variation of the anticyclones and Hadley circulation is consistent with rainfall over South Africa. During austral summer, a less intense, narrow mass stream function, South Atlantic Subtropical Anticyclone and Mascarene High are located more southwards, causing rainfall over the eastern parts of South Africa. During the austral winter, Hadley circulation, as well as the anticyclones, is stronger and located more northwards, causing rainfall over the southern and southwestern parts of South Africa.

Impact of intra‐decadal variability of meridional heat transport on the rainfall over Southern Africa during austral summer

AbstractThe present study examines the impact of intra‐decadal variability of Meridional Heat Transport (MHT) associated with Indian Ocean Shallow Meridional Overturning Circulation (SMOC) on the rainfall variability over Southern Africa (e.g., Zambia, Malawi, Mozambique, south‐eastern Congo, Zimbabwe, southern Tanzania and Madagascar) during austral summer (December to February). Century long ocean reanalysis, atmospheric reanalysis and rainfall observations are used for the study. Spectrum analysis of MHT index and rainfall anomaly displays intra‐decadal (5–7 years) variability significant at 95% confidence level. The oceanic and atmospheric conditions during strong and weak phases of intra‐decadal MHT variability are examined. Strong (weak) MHT years displayed reduction (increase) in Sea Surface Temperature (SST) in southwest Indian Ocean, which are out‐of‐phase with net heat flux but in phase with ocean circulation. The negative (positive) SST anomaly favours positive (negative) sea level pressure anomaly, which supports the low level moisture divergence (convergence) and anticyclonic (cyclonic) circulation in the lower troposphere. This circulation leads to anomalous subsidence (convection) and below (above) normal moist static energy over the study region resulting in below (above) normal convection, that causes negative (positive) rainfall anomaly over the Southern Africa. The association between intra‐decadal variability of MHT associated with SMOC and rainfall over Southern Africa during austral summer are reported for the first time. The study may pave the way for enhanced predictability of intra‐decadal Southern African austral summer rainfall.

The Phenomenology of West African Coastal Rainfall Events Based on a New Rain Gauge Network over Abidjan (Côte d’Ivoire)

In the District of Abidjan, flooding typically occurs suddenly during intense rainfall events. The individual rainfall event provides the basic element for the study. Its analysis is required to develop solutions for managing the impact of extreme rainfall occurrences. Our study proposes to identify individual rainfall events that occurred in the District of Abidjan from a densified network and analyze some of their characteristics related to the amount of rainfall they provided, their duration, and their level of intensity. A total of 1240 individual rainfall events were identified between 2018 and 2021 using a network of 21 rain gauges. Rainfall events were identified based on criteria such as a minimum inter-event time without rainfall of 30 min, a detection threshold of 0.02 mm/5 min, a minimum duration of 30 min applicable to the average hyetograph, and a minimum of 1 mm of rainfall in at least one rain gauge. The analysis of characteristics related to accumulation, intensity, and duration showed that the rainfall events were essentially convective, with an average duration of more than 2 h and a rainfall of 11.30 mm/event. For 70% of the rainfall events of a mixed nature, the convective episodes last up to 33.33% of the total duration of the event and produce an average of 80% of the cumulative rainfall. The 30-min peak intensities generally occur in the first half of the event. Less than 13.5% of the events have peaks greater than 50 mm/h. The probability of observing more than two, three, or four events per day is high in June and October, the core of the two rainy seasons.

Southern Annular Mode: A New Factor Impacts the East African Short Rains after the Early 1990s

Abstract It is well known that the Indian Ocean dipole (IOD) and El Niño–Southern Oscillation (ENSO) are closely related to rainfall variability over East Africa. This study found that the preceding August–October (ASO) Southern Annular Mode (SAM) can influence East African rainfall from October to December [i.e., East African short rains (EASR)], with a strong and significant relationship during the period of 1992–2019 than 1971–91. This is likely a result of the changes in SAM variations. It is further revealed that, during 1992–2019, the ASO SAM, via air–sea interactions, can excite meridional dipole sea surface temperature anomalies (SSTA) and warm SSTA over the southern Indian and Atlantic Oceans, respectively. In contrast, during 1971–91, the SAM induces warm SSTA over the southern Indian Ocean. Further analysis indicated that these ASO SAM-related SSTA patterns persist through the following October–December and exert a contrasting influence on the EASR. Based on the possible mechanism, the enhanced influence of SAM on EASR appears to be achieved through significant meridional circulation in the Indian Ocean basin, with anomalous sinking motion in the extratropics and the rising motion in the tropical Maritime Continent (MC) region. Correspondingly, strong convection is enhanced over the MC, altering zonal vertical circulation. This process results in anomalous westerlies in the tropical Indian Ocean, which reduces the atmospheric moisture, weakens convection, and suppresses rainfall over the central-western Indian Ocean, including the study region. Further analysis suggests that the prediction skill of the EASR related to the SAM, ENSO, and IOD was enhanced during 1992–2019. Significance Statement Several studies have investigated the influence of tropical climate anomalies, such as El Niño–Southern Oscillation and the Indian Ocean dipole, on the rainfall variability over East Africa. The region continues to be plagued by numerous climate-related disasters, including flooding and drought, and the associated varying impacts have been detrimental. As such, there is a need to understand further the factors and associated physical processes that cause such events and improve forecasting accuracy. This research emphasizes the role of the Southern Annular Mode (SAM), finding a strengthened lagged relationship with East African rainfall in recent decades. This implies that changes in the SAM should be considered in understanding present and future rainfall variability in the region.

Future climate change in the Agulhas system and its associated impact on South African rainfall

South African climate variability has been linked to changes in both the Agulhas system and external forcing (i.e. CO2 and ozone). We analysed future climate change in the Agulhas system volume transport and its associated impacts on South Africa’s precipitation using the Community Climate System Model version 4 as part of the Coupled Model Intercomparison Project, phase 5. Output from one historical and three future greenhouse gas emission scenarios were examined to project various climate storylines. We found that the Agulhas Current volume transport decreases across all three scenarios and that the current displays a strong baroclinic component with an increase in transport at the surface and decrease at intermediate depths. Agulhas leakage was found to increase with historical emissions. Additionally, an east-west dipole pattern for convective precipitation was found over South Africa, with an increase over the eastern region related to an increase in greenhouse gas emissions and a decrease in the western region linked to the location of Hadley cell edge latitude. Moving into the 21st century, future predictions in regional climate variability are shown to be dependent on the intensity of greenhouse gas emissions and are extremely important for South Africa, a region prone to drought and flooding and home to a large population dependent on rain-fed agriculture. Significance: Future climate variability in the Agulhas system and South African region is heavily dependent on changes in external forcing. The Agulhas Current volume transport decreases as the greenhouse gas emissions continue to increase and a strong baroclinic component is found with an increase in transport at the surface and a decrease at the intermediate depths. A strong east-west dipole precipitation pattern is found over South Africa with the increase in the eastern region related to the increase in greenhouse gas emissions and the decrease in the western region related to the location of the Hadley cell edge latitude.

Influence of strong South Atlantic Ocean Dipole on the Central African rainfall’s system

Influence of strong South Atlantic Ocean Dipole on the Central African rainfall’s system

Bias-Corrected CMIP5 Projections for Climate Change and Assessments of Impact on Malaria in Senegal under the VECTRI Model.

On the climate-health issue, studies have already attempted to understand the influence of climate change on the transmission of malaria. Extreme weather events such as floods, droughts, or heat waves can alter the course and distribution of malaria. This study aims to understand the impact of future climate change on malaria transmission using, for the first time in Senegal, the ICTP's community-based vector-borne disease model, TRIeste (VECTRI). This biological model is a dynamic mathematical model for the study of malaria transmission that considers the impact of climate and population variability. A new approach for VECTRI input parameters was also used. A bias correction technique, the cumulative distribution function transform (CDF-t) method, was applied to climate simulations to remove systematic biases in the Coupled Model Intercomparison Project Phase 5 (CMIP5) global climate models (GCMs) that could alter impact predictions. Beforehand, we use reference data for validation such as CPC global unified gauge-based analysis of daily precipitation (CPC for Climate Prediction Center), ERA5-land reanalysis, Climate Hazards InfraRed Precipitation with Station data (CHIRPS), and African Rainfall Climatology 2.0 (ARC2). The results were analyzed for two CMIP5 scenarios for the different time periods: assessment: 1983-2005; near future: 2006-2028; medium term: 2030-2052; and far future: 2077-2099). The validation results show that the models reproduce the annual cycle well. Except for the IPSL-CM5B model, which gives a peak in August, all the other models (ACCESS1-3, CanESM2, CSIRO, CMCC-CM, CMCC-CMS, CNRM-CM5, GFDL-CM3, GFDL-ESM2G, GFDL-ESM2M, inmcm4, and IPSL-CM5B) agree with the validation data on a maximum peak in September with a period of strong transmission in August-October. With spatial variation, the CMIP5 model simulations show more of a difference in the number of malaria cases between the south and the north. Malaria transmission is much higher in the south than in the north. However, the results predicted by the models on the occurrence of malaria by 2100 show differences between the RCP8.5 scenario, considered a high emission scenario, and the RCP4.5 scenario, considered an intermediate mitigation scenario. The CanESM2, CMCC-CM, CMCC-CMS, inmcm4, and IPSL-CM5B models predict decreases with the RCP4.5 scenario. However, ACCESS1-3, CSIRO, NRCM-CM5, GFDL-CM3, GFDL-ESM2G, and GFDL-ESM2M predict increases in malaria under all scenarios (RCP4.5 and RCP8.5). The projected decrease in malaria in the future with these models is much more visible in the RCP8.5 scenario. The results of this study are of paramount importance in the climate-health field. These results will assist in decision-making and will allow for the establishment of preventive surveillance systems for local climate-sensitive diseases, including malaria, in the targeted regions of Senegal.

The Role of Water Vapor Observations in Satellite Rainfall Detection Highlighted by a Deep Learning Approach

West African food systems and rural socio-economics are based on rainfed agriculture, which makes society highly vulnerable to rainfall uncertainty and frequent floods and droughts. Reliable rainfall information is currently missing. There is a sparse and uneven rain gauge distribution and, despite continuous efforts, rainfall satellite products continue to show weak correlations with ground measurements. This paper aims to investigate whether water vapor (WV) observations together with temporal information can complement thermal infrared (TIR) data for satellite rainfall retrieval in a Deep Learning (DL) framework. This is motivated by the fact that water vapor plays a key role in the highly seasonal West African rainfall dynamics. We present a DL model for satellite rainfall detection based on WV and TIR channels of Meteosat Second Generation and temporal information. Results show that the WV inhibition of low-level features enables the depiction of strong convective motions usually related to heavy rainfall. This is especially relevant in areas where convective rainfall is dominant, such as the tropics. Additionally, WV data allow us to detect dry air masses over our study area, that are advected from the Sahara Desert and create discontinuities in precipitation events. The developed DL model shows strong performance in rainfall binary classification, with less false alarms and lower rainfall overdetection (FBias <2.0) than the state-of-the-art Integrated MultisatellitE Retrievals for GPM (IMERG) Final Run.

The African Holocene Humid Period in the Tibesti mountains (central Sahara, Chad): Climate reconstruction inferred from fossil diatoms and their oxygen isotope composition

The climate of the African Holocene Humid Period (AHHP) is reconstructed in the Tibesti Volcanic Massif (TVM) in the central Sahara from well-preserved diatomaceous deposits in the two crater palaeolakes of Trou au Natron at Pic Toussidé and Era Kohor at Emi Koussi. The two records cover the period from ∼9500 to 4500 cal yr BP. Climate and palaeoenvironmental changes during this period were inferred from diatom assemblages, interpretation of variations in their oxygen isotope composition (δ18Odiatom), reconstruction of lake water conductivity from diatom-based transfer functions, and estimation of the lake water balance (Evaporation/Inflow ratio, E/I). Our findings provide evidence for two distinct lacustrine episodes. During the early to mid-Holocene transition, low δ18Odiatom values, high percentages of planktonic diatoms, low lake water conductivity and a positive water balance (E/I < 1) suggest wet conditions, which were likely related to the optimum of the AHHP. From the mid-to late Holocene transition, an aridification trend is revealed by increasing δ18Odiatom values, high percentages of benthic diatoms and a negative water budget (E/I > 1), occurring as early as 6500 cal yr BP and intensifying after 5300 cal yr BP. Moreover, our data show on average a decrease in precipitation amounts of ∼35% between the peak and the end of the AHHP in the Tibesti region. This timing of the AHHP in the mountainous Tibesti is consistent with the aridification of the central Sahara recorded at lowland sites, which has mainly been related to the southward retreat of the Intertropical Convergence Zone (ITCZ) and the associated African monsoonal rainfall belt, following the gradually declining summer insolation that led to the termination of the AHHP. Our results prove the existence of Holocene lakes in the TVM craters that developed contemporaneously with the lakes of the Chadian basin and the Libyan Sahara. On a broader scale, our data share similar hydroclimatic patterns with studies from the eastern and northern Sahara.

The Potential of Deep Learning for Satellite Rainfall Detection over Data-Scarce Regions, the West African Savanna

Food and economic security in West Africa rely heavily on rainfed agriculture and are threatened by climate change and demographic growth. Accurate rainfall information is therefore crucial to tackling these challenges. Particularly, information about the occurrence and length of droughts as well as the onset date of the rainy season is essential for agricultural planning. However, existing rainfall models fail to accurately represent the highly variable and sparsely monitored West African rainfall patterns. In this paper, we show the potential of deep learning (DL) to model rainfall in the region and propose a methodology to develop DL models in data-scarce areas. We built two DL models for satellite rainfall (rain/no-rain) detection over northern Ghana from Meteosat TIR data based on standard DL architectures: Convolutional neural networks (CNNs) and convolutional long short-term memory neural networks (ConvLSTM). The Integrated Multi-satellitE Retrievals for the Global Precipitation Measurement (GPM) mission (IMERG) and Precipitation Estimation from Remotely Sensed Imagery Using an Artificial Neural Network Cloud Classification System (PERSIANN-CCS) products are used as benchmarks. We use rain gauge data from the Trans-African Hydro-Meteorological Observatory (TAHMO) for model development and performance evaluation. We show that our models compare well against existing products despite being considerably simpler, developed with a small training dataset—i.e., 8 stations covering 2.5 years with 20.4% of the data missing—and using TIR data alone. Concretely, our models consistently outperform PERSIANN-CCS for rain/no-rain detection at a sub-daily timescale. While IMERG is the overall best performer, the DL models perform better in the second half of the rainy season despite their simplicity (i.e., up to 120 k parameters). Our results suggest that DL-based regional models are a promising alternative to state-of-the-art global products for providing regional rainfall information, especially in meteorologically complex regions such as the (sub)tropics, which are poorly covered by ground-based rainfall observations.