Application of Analytical Hierarchy Process (AHP) and Multi-Criteria Evaluation (MCE) for a case study and scenario assessment of flood risk in the White Volta Basin of the Upper East Region, Ghana

Abstract

In recent years, Ghana, particularly the inhabitants of the Upper East Region, has experienced profound impact of flooding, largely attributable to the complex interplay of climatic factors. This research embarks on a comprehensive assessment of flood risk zones nestled within the White Volta Basin, situated in the Upper East Region. The study employs advanced cartographic methodologies and uses Geographic Information Systems (GIS) in conjunction with the Analytical Hierarchy Process (AHP) to systematically categorize areas susceptible to inundation. Leveraging geospatial datasets acquired from satellites such as Landsat and Sentinel. Topographic, slope, and Land Use/Land Cover (LULC) maps have been constructed. The empirical findings underscore the susceptibility of specific regions, including the Talensi District, territories within Bawku West, and some segments of the Bolgatanga Municipal area, to escalated flood risk. Additionally, the research underscores the high vulnerability of communities such as Nunku, Tolla, Zaare, Pwalugu, Balungu, Winkongo, Biung, and Tongo to the negative impact of inundation. Significantly, the study unveils a pivotal factor in the perpetuation of flood devastation—namely, the role of water discharge. This intrinsic linkage between discharge rates and flood occurrences underscores the pressing need to address this critical component in mitigation strategies to reduce adverse impacts on the basin’s resident communities. The insights derived from the study offer some level of hope for residents, providing essential knowledge concerning flood-prone areas and optimal timing for agricultural activities to safeguard their cherished livelihoods.

Introduction

Flooding is one of the devastating effects of climate change in both developing and industrialized nations [1]. Floods affect an estimated 520 million people and their livelihoods and claim approximately 25,000 lives worldwide [2]. Debarati Guha-Sapir [3] reported that floods affected over 78.1 million people, accounting for approximately 13.7% of all disaster victims in the United States in a calendar year. It affects 60 million people annually in China [1]. Flooding is responsible for the direct economic loss of ten billion dollars globally each year [4].

The risk associated with flooding varies based on region, magnitude, frequency, and duration [5]. Consistent rainfall and snowmelt increase the frequency and intensity of wet conditions, which are essential for flood-generating mechanisms [6] and different forms and shapes [7]. Flash flooding, which is sudden heavy rain, raises the water level and overwhelms rivers, roads, streams, and channels. Changes in riverine floods significantly impact the design of flood protection measures and flood risk assessments [5].

Regarding flood events, Africa is the second most affected continent after Asia [8]. Floods represent 77.3% of all disasters, causing huge losses to life and property [2], in Africa. Extreme occurrences of floods and droughts have serious implications on global agricultural productivity and livelihoods [9]. Floods are preventable natural disasters that are mostly human induced but can be prevented [10], especially, with urban flooding that has occurred in Ghana since 1930 [8]. In August 2007, severe flooding in Ghana affected approximately 350,000 people, with 49 deaths in the northern part of the country [11].

In recent times, on October 15, 2023, the southern part of the White Volta, area of Ghana experienced severe flooding as a result of the spillage from Akosombo Dam causing disastrous consequences in communities downstream [12]. The spillage of the Akosombo and Kpong Dams displaced an estimated 35,857 people including children living along the banks of Lake Volta. School buildings, healthcare facilities, foodstuff, and energy systems were all severely impacted by the spillage coupled with torrential rainfall [13].

The opening of the Bagre dam and strong rains forced 161,000 people from their homes across the nation [11], particularly residents of Volta Lake, where over 55 communities in the Central Gonja in the Savanna Region, and Kwahu East, Kwahu South, and Kwahu North districts in the Eastern Region were severely affected [11].The northern part of Ghana experienced prolonged dry spells, followed by intense rainfall, among Ghana's most flood-prone regions. In 2009, floods displaced more than 121,000 people and destroyed 5104 houses, 13 schools, and 30,000 acres of farmlands [14].

Asare-Kyei et al. [15] conducted a flood risk assessment at the Vea catchment in the Upper East Region of Ghana. Similarly, Kheradmand [16] assessed the impact of perennial floods on agricultural activities and their results showed a decline in food production among farmers. Flood risk is defined as the magnitude of the economic losses as a result of flooding and the probability that such losses will occur with that magnitude [17]. A flood risk analysis involves the identification, estimation, and evaluation of various factors that contribute to the occurrence and mitigation of floods. Flood risks are as a result of intense rainfall, inadequate spatial planning, insufficient waste management practices, and obstructed sewers [18], and indicators for elements at risk determine the number of social, economic, or ecological units that are vulnerable to the impacts of floods.

Investigating the risks associated with flooding is an essential step in developing effective policies and measures to improve the livelihood and sustainability of the environment. Flood risk assessment is important for disaster management and decision-making. Floods are the outcome of complex interactions between astronomical, geographical, climatic, hydrological, vegetation, and human factors. Therefore, comprehensive flood analysis and risk management require a holistic approach to ensure reliability and accuracy. Intense rainfall causes flooding in Ghana from high runoff [6, 8, 19]. Amoako and Boamah [19] classified the flood hazard of Accra into three broad areas of meteorological (i.e., rainfall and storm surge), hydrological, and land use changes as a result of anthropogenic activities. Areas with low elevation values are high-risk areas for flooding because they generate the largest amounts of runoff [20]. During flood events, staple food crops, such as maize, millet, groundnut, and leafy vegetables, are affected [21]. Similarly, the consequences of floods affect people’s well-being, specifically among the less privileged, through the spread of waterborne diseases, injuries, and animal attacks [22]. According to [23], the winter of 2013–2014 led to widespread river, coastal, and surface water flooding after a period of heavy rainfall.

Flood hazards and risk maps are effective tools for reducing flood damage. Spatial information has become a valuable tool for assessing the risk of flood hazards, contributing to sustainable progress in human society [10, 24]. Flood mapping with satellite data from Sentinel and Landsat, together with high-resolution Shuttle Radar Topographic Mission (SRTM) data, improve flood mapping accuracy [25]. A Geographic Information System (GIS) can incorporate radar rainfall data, high-resolution Digital Elevation Models (DEMs) from the SRTM, soil data, land use/land cover (LULC) [24], and compute the spatial variation of flood-induced parameters in a catchment.

The Analytical Hierarchy Process (AHP) is a technique that uses worldwide decisions for multi-criteria analysis (MCA) [10, 24, 26–28]. Various parameters are assigned weights according to their importance, creating a guide for understanding the parameters that have a high influence on flooding. Hence, this study seeks to use remote sensing and the GIS approach together with AHP to do an appraisal of the flood risk levels of the various zones within the White Volta Basin in the Upper East Region. This study specifically seeks to; (a) classify flood risk zones through mapping, (b) identify the causative factors of flooding and rank their contribution to flooding, and (c) estimate the extent of the flood-prone zones within the basin.

Materials and methods

Description of study area

The study area is a segment of the White Volta Basin with an estimated area of about 4400 km² within the Upper East Region (8842 km²) [29]. The population density in the region is approximately 103 people per km² [30]. The study area has about ten (10) municipalities/districts within the Upper East Region as shown in (Fig. 1) and is bordered by Burkina Faso to the North, Togo to the East, to the south by North East Region and other districts like Kassena-Nankana West district to the west. It lies between longitude 0° and 1° W and latitude 10° and 11° N. The region’s natural vegetation primarily consists of savannah woodland with short, widely spaced trees and grass that endure drought but are vulnerable to bushfires and sun scorching during the long dry season. The area is predominantly covered by Guinea-savannah and Sudan-savannah vegetation. Rainfall follows a unimodal pattern [31], with an average annual rainfall of 921 mm, ranging from 645 to 1250 mm, creating a growing season lasting approximately 5 to 6 months from April/May to September/October, followed by a long dry and hot season lasting 6 to 7 months from October to April. The annual average temperature ranges from a minimum of 15 °C during the dry season (December to February) to a maximum of 45 °C in March and April [30]. The main rivers in the Upper East Region are the White Volta and the Red Volta, while there are other minor rivers such as ‘Agrumatue’ which flows between the cities of Bolgatanga and Zuarungu. One of the major challenges of the area is perennial flooding most often due to spillage from the Bagre Dam in neighboring Burkina Faso. This is exacerbated by the changing rainfall pattern.

Fig. 1 [Images not available. See PDF.]

A map of Ghana showing the location of the study area

Description of datasets

Various datasets were harnessed to create flood-prone maps. These included a high-resolution digital elevation model (DEM) sourced from Shuttle Radar Topography Mission (SRTM), providing detailed elevation, slope, and the extraction of distance from river with the catchment delineation output map. For land use/land cover mapping, the study used World Cover 2021 v200 data, which offers a resolution of 10 m. Total annual rainfall data from 1990, 2000, 2010, and 2020 were obtained from Climate Hazards Group Infrared Precipitation with Station data (CHIRPS), with a resolution of 0.05°, enabling trend analysis and drought monitoring. Additionally, soil type data was sourced from Global Hydraulic Soil Group (HYSOGs250m) with a resolution of approximately 250 m. Discharge data, crucial for understanding river behavior, were acquired from the Ghana Hydrological Service Department (Table 1).

Table 1. Geospatial data, type, source and resolution

S/NN	Name of data	Type of data	Source of the data	Spatial resolution
1	Digital Elevation Model (DEM)	Raster	Earthexplorer.usgs.gov-SRTM	1 arc—Second (30 * 30 m)
2	Landcover 2021	Raster	ESA worldcover.org	10 m
3	Soil type	Raster	https://cmr.earthdata.nasa.gov/	250 m
4	Rainfall	Raster/grided data	https://www.chc.ucsb.edu/data/chirps	0.05°
5	Streamflow	Daily data measurement	Hydrological Services Department, Ghana	N/A

Esri product, ArcMap version 10.4 was used for the production of the various flood prone maps. The. AHP was used to assign weights to the various flood inducing parameters. Figure 2 shows the methodological framework in the generation of the flood prone zonation maps.

Fig. 2 [Images not available. See PDF.]

Methodological framework in flood prone map generation

The Analytical Hierarchy Process (AHP)

The AHP is a comprehensive measurement theory that enables the creation of ratio scales through comparison between discrete and continuous variables [32]. These comparisons can be based on actual measurement or fundamental scale that represents the relative strength of preference and emotions. AHP is a globally recognized technique for assigning weights to different criteria based on their importance [10, 33], aiding in risk assessment and mitigation planning. Researchers commonly use the Multi-Criteria Analysis (MCA) approach for its flexibility and versatility in addressing various factors. The use of the AHP follows a six (6) series of procedures: (1) Breaking a complex, unstructured problem into its component factors, (2) Development of the AHP hierarchy, (3) Paired comparison matrix determined by imposing judgments, (4) Assigning values to subjective judgments and calculating the relative weights of each criterion, (5) Synthesize judgments to determine the priority variables and lastly, (6) Check the consistency of assessments [34]. This approach is summarized in (Fig. 3).

Fig. 3 [Images not available. See PDF.]

Procedures involved in AHP

A fundamental scale by [35] was used for the comparison [32] as in the (Table 2).

Table 2. AHP measurement scale

Intensity of importance on absolute scale	Definition
1	Equal importance
3	Moderate importance
5	Essential or Strong importance
7	Very strong importance
9	Extreme importance
2, 4, 6, 8	Intermediate values between two adjacent judgements
Reciprocal	If activity i has one of the above numbers assigned to it when compared with j, then j has the reciprocal when compared with i

A matrix for pairwise comparison was constructed, where the relative significance of the factors was obtained using the AHP method. The weight assigned to each of the factors was determined based on its relative importance with a rating scale ranging from 1 to 9. Lower values indicate less crucial factors, while higher values indicate more significant factors. Table 3 shows the pairwise comparison matrix of the factors with a 7 by 7 matrix. For example, a scale of 5 was assigned to rainfall as compared with slope, which explains that rainfall was of strong importance as compared to slope.

Table 3. AHP comparison matrix

Parameters	R	E	DR	D	SL	LULC	Soil
R	1	2	2	5	5	3	4
E	1/2	1	2	5	2	3	3
DR	1/2	1/2	1	3	2	3	5
D	1/5	1/5	1/3	1	3	3	2
SL	1/5	1/2	1/2	1/3	1	3	2
LULC	1/3	1/3	1/3	1/3	1/3	1	3
Soil	1/4	1/3	1/5	1/2	1/2	1/3	1

R is rainfall, E is elevation, DR is distance from river, D is discharge, SL is slope, LULC is land use/land cover. Each value of the pairwise matrix table was divided by the summation value of its respective column as shown in Table 4

The Eigen vector was developed using the formula below.

1

$$\text{Vp}=\sqrt[K]{w1\times \cdots \times w}k$$where $k$ is the number of parameters and comparing $\mathit{Wk}$ ratings main parameters and the weighting coefficient ($\mathit{Cp}$) calculated using

2

$$\text{CP}=\frac{\text{VP}}{\text{VP}1+\text{VP}2+\cdots +\text{VPK}}$$

And the sum $\mathit{Cp}$ of all the parameters of the matrix must be equal to 1.

The Consistency ratio (CR) is one of the most important components which is required to be determined when using AHP [34]. When the consistent ratio is below 0.1, then the said matrix is considered to have an acceptable consistency. The CR is calculated by multiplying each column of the matrix by the priority vector corresponding there to determine the overall priority [D];

• Divide each global priority by the corresponding priority vector to determine the rational priority [E] (Table 4);

• Determine the maximum eigen value (Xmax) by the following equation:

  3

  $$\text{Xmax}=\frac{\text{E}}{\text{K}}$$

Table 4. AHP normalization matrix

Flood inducing parameters	R	E	DR	D	SL	LULC	Soil	(Vp)a	(Cp)b
R	0.34	0.41	0.31	0.33	0.36	0.18	0.20	2.75	0.31
E	0.17	0.21	0.31	0.33	0.14	0.18	0.15	1.90	0.21
DR	0.17	0.10	0.16	0.20	0.14	0.18	0.25	1.56	0.17
D	0.07	0.04	0.05	0.07	0.22	0.18	0.10	0.81	0.10
SL	0.07	0.10	0.08	0.02	0.07	0.18	0.10	0.72	0.09
LULC	0.11	0.07	0.05	0.02	0.02	0.06	0.15	0.53	0.07
Soil	0.08	0.07	0.03	0.03	0.04	0.02	0.05	0.39	0.05
SUM	1.00	1.00	1.00	1.00	1.00	1.00	1.00		1.00

^aEigen vector (Vp), ^bWeighting coefficient (Cp)

Calculate the consistency index expressed as:

4

$$\text{CI}=\frac{\text{Xmax}-\text{K}}{\text{K}-1}$$

The Consistency ratio (CR)

5

$$\text{CR}=\frac{\text{CI}}{\text{RI}}$$

The (Tables 5 and 6) shows the standard Random Inconsistency values and Consistency ratio according to [35].

Table 5. Saaty random inconsistence values

N*	2	3	4	5	6	7	8	9
RI	0	0.52	0.9	1.12	1.24	1.32	1.41	1.45

N* is the number of parameters and RI represent the inconsistency values

Table 6. Computation of consistency ratio

Normalized matrix (row sum)		cp	D = A*cp	E = D/cp	Xmax	CI	CR
R	3.16	0.50	4.81	9.58
E	0.82	0.14	1.22	8.80
DR	1.11	0.16	1.57	9.94	9.30	0.20	0.15
D	0.88	0.10	0.96	10.10
SL	0.53	0.05	0.50	9.89
LULC	0.32	0.03	0.33	10.76
Soil	0.19	0.03	0.24	9.01
Sum	A = 7.00			68.09

A is the total summation of the values in the column of the “normalized matrix SUM’’

A consistency ratio of 0.10 was obtained, which could be said that the consistency within the parameters is acceptable with a slight increment.

Mann–Kendall and Sen’s slope estimate

In determining the variability and long-term monotonic trends in rainfall, the Mann–Kendal and Sen’s Slope estimate has been found to yield accurate results [36–38]. Analysis of rainfall patterns is of great interest because it makes the global community to be well-informed about the climate change. The Mann–Kendall test is preferably used because the test is not affected by the actual distribution of the data. In other words, it is less susceptible to outliers because it is dependent on the ranks of the observation rather than the actual values [36]. According to [39], the Mann–Kendall test has been used in many ways to quantify the significance of trends in hydro-meteorological time series. An upward trend in the estimate indicates that there is a positive trend whereas a downward trend in the estimate is evidence of a negative trend [38, 40]. Total annual rainfall data for a period of 31 years from 1990 to 2020 were used to determine the trend in rainfall for that period.

Processing of datasets

Thematic layer classification was conducted using ArcGIS® version 10.4 to prepare essential maps for flood-prone area assessment. The process involved generating elevation and land use/landcover maps from SRTM and World Cover datasets, respectively. These maps were clipped to the study area and reclassified into relevant classes. Additionally, soil type data and annual rainfall data from Climate Hazards Group were imported, clipped, and reprojected to create pertinent layers for analysis. Annual rainfall maps for four different years were produced through classification for further assessment. A distance from river map was generated by adjusting the coordinate reference system of stream network datasets to match that of the study area, followed by calculating Euclidean distances from the source using the Spatial Analyst toolbox. The model was validated and tuned before classification. Additionally, a slope map was produced from the DEM using the Arc-Hydro Surface tool and classified. Stream flow/discharge data was aggregated from daily to monthly and yearly levels for 1990 and 2000, resulting in maps to analyze discharge patterns over time.

The elevation layer was reclassified into 3 classes based on susceptibility shown in Fig. 4 (right). Areas with high elevation (211–460 m) were assigned a scale of 1 (low susceptibility), a scale of 2 (moderate susceptibility) was assigned to areas with elevation values of (180–211 m). A scale of 3 (high susceptibility) was also assigned to areas with elevation values of (0–180 m).To rank the percentage of each prone area covered, the area was estimated as 45% for low prone zones (1999 km²), 35% for moderate prone zones (1575 km²) and 20% for high prone zones (907 km²).

Fig. 4 [Images not available. See PDF.]

Maps showing elevation of the study area (left) and reclassified elevation (right)

Annual total precipitation (PRCPT)

GIS technology was used to create maps representing precipitation levels (PRCPT) for the years 1990, 2000, 2010, and 2020. Different factors were analyzed and categorized based on their susceptibility to flooding. Rainfall data was divided into three classes: high, moderate, and low susceptibility to floods, with corresponding scales of 3, 2, and 1. These susceptibility levels are summarized in (Table 7).

Table 7. Area covered by each flood prone zone of the 4 PRCPT

Annual total 2PRCPT	PRCPT values (mm)	Prone level	Area covered (Km2)	% area
1990	721–774	Low	1339	30
	774–808	Moderate	1703	38
	808–903	High	1408	32
2000	793–882	Low	1803	40
	882–938	Moderate	1609	36
	938–1043	High	1067	24
2010	1001–1065	Low	2496	55
	1065–1116	Moderate	1006	22
	1116–1192	High	996	22
2020	1016–1064	Low	1399	31
	1064–1102	Moderate	1993	44
	1102–1169	High	1088	24

From Fig. 5, and on the PRCPT 1990, it can be seen that South-western portion of the area recorded the highest annual total rainfall, whereas the eastern portion was moderate. The PRCPT (2000) map had the highest rainfall at the southern portion of the basin whereas the northern portion had moderate to low rainfall recorded as can be seen from below (Fig. 5). Table 8 above shows the proportion of the of the area coverage for all classes. In the PRCPT (2010) map, the middle to eastern portion of the area had a moderate to low annual total rainfall whereas the western part of the basin recorded the highest rainfall. The area and percentage coverage of each zone is recorded in Table 7. In PRCPT (2020) map, the rainfall was distributed all over the basin with the middle portion dominated with low values of rainfall. The western part of the basin also recorded high values of rainfall, including some portion of the eastern side of the basin.

Fig. 5 [Images not available. See PDF.]

Maps showing the reclassified Annual Total Rainfall (PRCPT)

Table 8. Case studies, layers involved, districts and the flood prone communities

Case studies	Data used	Affected districts and flood-prone zones
Case 1	PRCPT 1990, Annual total discharge (1990), LULC, DR, E, SL, S,	High Prone Zone: Talensi, Bolgatanga, Bawku West, South Western part of Tempane Low prone zone: Garu, Tempane, Binduri, Pusiga, Bawku Moderate zone: Bongo, Nabdam, Northern and mid-portions of Bawku West
Case 2	PRCPT 2000, Annual total discharge(2000),LULC, DR, E, SL, S,	High zone: Southern section of Talensi, Bawku West, Tempane Moderate zone: Middle portion of Talensi, Bawku West and Tempane Low zone: Bongo, Pusiga, Nabdam, Binduri, Bawku, Northern portion of Bawku West
Case 3	PRCPT 2010, total annual discharge (2000), LULC, DR, E, SL, S,	High zone: (Talensi, Bolgatanga, small southern portion of Tempane Moderate zone: Parts of Talensi, Bolgatanga, Bongo, Southern parts of Bawku West and Tempane, mid-sections of Bawku West Low zone (Bawku, Pusiga, Garu, parts of Tempane, parts of Bawku West)
Case 4	PRCPT 2020, total annual discharge (2000), LULC, DR, E, SL, S,	High zone: Parts of Talensi and Bolgatanga, middle portion of Garu, southern-Tempane) Moderate zone: Portion of Garu, Tempane, Binduri, Bawku, middle portion of Bawku west, part of Nabdam, portion of Talensi and Bolgatanga) Low zone: Dominated by Nabdam, Bawku west and eastern parts of Pusiga

L = slope, DR = distance from river, LULC = land use/landcover, E elevation and S is soil, PRCPT Annual total rainfall

The slope layer was categorized into three susceptibility levels based on its inclination degree. Slopes ranging from 7.8° to 105.1° were classified as having low susceptibility (scale 1), while those between 3.7° and 7.8° were deemed to have moderate susceptibility (scale 2). Areas with slopes between 0° and 3.7° were considered to have high susceptibility (scale 3) as shown in Fig. 6

Fig. 6 [Images not available. See PDF.]

Maps of reclassified slope (left), buffered distances from rivers

The distance from the river layer was divided into three classes of flood susceptibility. Areas close to the river (0–1068 m) were labeled as highly susceptible (scale 3), those at a moderate distance (1069–2359 m) were classified as moderately susceptible (scale 2), and areas farther away (2360–11,351 m) were considered to have low susceptibility (scale 1) as presented in Fig. 7 right.

Fig. 7 [Images not available. See PDF.]

LULC map, 2021(left) and Reclassified LULC map (right)

Land use/landcover map and reclassified map

The land use/landcover layer was reclassified into three classes of susceptibility. Permanent Water bodies were given a value of 3 which has high level of susceptibility; a value of 2 (moderate) was assigned to buildup areas, areas of croplands, grasslands, bare lands and shrublands. Areas of tree cover were assigned a value of 1 (low susceptibility) to flooding (Fig. 7 right).

Soil type reclassification

The soil type layer was reclassified into 3 susceptibility scales, soil with moderately high runoff potential (< 50% sand and 20–40% clay) was assigned a susceptibility value 1 (low susceptibility), value of 2 (moderate susceptibility) for soil with high runoff potential unless drained (< 50% sand and 20–40% clay). A susceptibility value of 3 (high susceptibility) was assigned to soil with high runoff potential unless drained (< 50% sand and < 40% clay) as presented in Fig. 8.

Fig. 8 [Images not available. See PDF.]

Reclassified soil map

The streamflow data which is a daily discharge point measurement was used to prepare the streamflow map of the area. The streamflow covers the entire study area so there was no reclassification. It is assumed all the area contributes to the flow and for that reason it was considered as one class.

Results

Case study analysis

The 4 years Annual total rainfall (PRCPT) was used as the basis for conducting case studies to produce the flood prone zonation maps due to the changing rainfall patterns. The streamflow or discharge for 1990, 2000, was also used together with the remaining flood induced parameters. (i.e., slope, elevation, distance from river, and soil). Case study, layers involved, districts and the flood prone communities are shown in Table 8 whilst the flood-prone maps generated for Cases 1, 2, 3 and 4 are depicted in Fig. 9A–D, respectively.

Fig. 9 [Images not available. See PDF.]

Flood prone maps; (1990-A, 2000-B, 2010-C, and 2020-D)

The area coverage of each flood-prone zone for the various case studies (1, 2, 3 and 4) are presented in (Table 9), together with the list of communities for each zone in every case study. The average flood prone zones under all case studies for the Low, Moderate and High zones are (Area/% area) are 965 km²/23%, 2827 km²/66% and 502 km²/12% respectively.

Table 9. Area occupied by each flood prone zone for the case studies

Case study	Prone zones	Area in (km2)	% Area	Flood-prone communities
Case study 1 (1990)	Low	864	21	Widnaba, kokore, Zaari, Kpikpira Nantinga, Gagbari, Garu, Tempane, Kaadi, Pafwah
	Moderate	2820	68	Sinebaga, Kugri, Tilli, Zebilla, Bawku, Pusiga, Kobori, Kasumaba
	High	497	12	Balure, Nungu, Kejetia, Bari, kantia, Tolla, Yale, Biung, Tongo, Shiega
Case study 2 (2000)	Low	1126	25	Widnaba, Dubila, Kongo, Duusi, Sakote, Dakio, Tilli, Kobori, Kasumaba, Gbega, Guoshe, zebilla, Binduri, Bawku, Pusiga
	Moderate	82,805	63	Yale,Balungu, Tengzu, Tongo, Kugri, Zaari, Winkongo, Sawaliga, Shia, Kaadi, Garu, Gagbiri, Kpikpira Nantinga Kugri
	High	525	12	Balure, Nungu, Kejetia, Sinebaga Tolla, Pwalugu, Shiega
Case study 3 (2010)	Low	864	21	Tilli, Kusanaba, zebilla, Pafwah, Kugri, Pusiga, kobori, zaari, Garu, Tempane, Kpikpira Natinga, Binduri, Bawku
	Moderate	28,208	68	Tolla, Bingo, Yale, Kongo, Guohe, Kunkoa, Dakio, Sinebaga, kaadi, Gagbiri, Zongoyiri
	High	497	12	Winkongo, Nungu, Pwalugu, Balure, Yazore, Kejetia, Tongo, Shirigu, Yinduri, Shiega
Case study 4 (2020)	Low	1005	23	Widnaba, Zebilla, Kokore, Tilli, kugri, Pusiga, Kusanaba, Zongoyiri
	Moderate	2863	66	Bingo, Biung, kpatia, Bawku, Binduri, kaadi, Kobori, Shia, Sinebaga, Zaari, Gagbiri, Kpikpira Nantinga, Tempane, Garu, Tolla
	High	483	11	Pwalugu, Tongo, Nungu, Kejetia, Shiega, Gorogo, Zarre, Tengzug, Bari, Balure, Balungu, Winkongo

Scenario analysis

Scenario analysis was performed to obtain how each of the changing parameters, PRCPT, LULC and discharge contribute to flooding and the extent in terms of area. The various scenarios, parameters involved and the weights assigned to each parameter are presented in Table 10.

Table 10. Scenarios, parameter, weights

Scenario analysis	Parameters	Weight	% Weight
Scenario 1	PRCPT	0.5	50
	LULC	0.25	25
	Discharge	0.25	25
Scenario 2	PRCPT	0.25	25
	LULC	08.5	50
	Discharge	0.25	25
Scenario 3	PRCPT	0.25	25
	LULC	0.25	25
	Discharge	0.5	50

The maps for the scenario analysis are presented in Fig. 10A–C.

Fig. 10 [Images not available. See PDF.]

Map showing the 3-scenario analysis

The area covered by each flood-prone zone from each of the scenarios conducted are presented in Table 11.

Table 11. Area covered by each flood-prone zone of the scenario analysis

Scenarios	Flood prone zones	Area in (km2)	% area
1	Low	1800	40
	Moderate	2678	60
	High	1	0
2	Low	268	1
	Moderate	42,452	99
	High	1	0
3	Low	–	–
	Moderate	1814	40
	High	2665	60

Ranking of flood generating parameters showed that river discharge, Precipitation and Land use/Landcover were the main causative factors with discharge being the highest ranked.

The Mann–Kandall test is applicable when the trend is most probably monotomic and there are no seasonal or other cycles present in the data. The Z value describes the trend as a positive (upward) trend and a negative trend (downward). It is used for estimating the trend of time series data. When the number (n) of data used in the estimate is less than 10, the S value is used and when the number of time series data is 10 or more, the Z value is used. In the data estimate, the number of time series data is 31, so Z was used where Z = − 0.07 indicating a downward trend. The significance level of 0.1 means that there is a 10% probability that we made a mistake when rejecting the presence of a trend. Sen’s slope estimates Q for the true slope of a linear trend, i.e., change per unit period (in this case a year) is also negative (− 1.56). Test B estimate the constant B with the equation f(year) = Q*(year-firstYear) + B for a linear trend. B is positive which ranges between a minimum of 1.21 to a maximum of 9.76.

Discussion

The purpose of this study was to identify and evaluate a section of the White Volta Basin’s flood risk zones. The study uncovers a range of findings that have important ramifications for managing flood risk in that area of the White Volta Basin.

From the weights assigned to the various parameters, the results showed that rainfall has the highest weight (0.31), followed by elevation (0.21), distance from river (0.17), discharge (0.10), slope (0.09), LULC, (0.07), soil (0.05) respectively (see Table 4). The consistent ratio (CR) value obtained was 0.10 which falls within the range of (CR ≤ 0.10), and this indicates that the weighting of the parameters was consistent. In the case study 1 (Fig. 9: 1990-A), the high flood prone locations were focused at the south-western part of the basin. This strongly correlates with the PRCPT (1990) which was very high at that location. It strongly means that the amount of rainfall had a great influence on the generation of floods at the area. The Talensi District and Bawku West Municipal were those found in the high flood prone zones.

In case study 2 (Fig. 9: 2000-B), the southern portion of the basin was found to be within the high flood prone zone, the south portion of Talensi, Bawku West and Garu District were within the high flood prone zone. The middle and north portion of the basin was found to be within the moderate and low prone zones which includes Bongo District, Bawku Municipality, Bolgatanga Municipality, Binduri and the Nabdam District. This also correspond to the rainfall data of 2000 where the highest annual rainfall recorded was within the zone of high flood prone and the lowest annual rainfall situated within the moderate and low flood prone zones.

In case study 3 (Fig. 9: 2010-C), the high flood prone zone was located in the western part of the basin and a small area in the southern part. Talensi District, Bolgatanga Municipality, was found to occupy the high flood prone zone. The middle and eastern portion were found to occupy moderate and low prone zones. Moreover, the western portion of the area which include Talensi District, portion of Bolgatanga Municipality, some area of Garu District also was found within the moderate and low prone zones.

According to NADMO, flood-prone co mmunities include Tolla, Pwalugu, Yinduri, Santeng, Biung, Nungu, Daare, Bapella, Winkongo, Balungu and Vung areas (NADMO,Talensi). Meanwhile, from the case studies, it was observed that some co mmunities such as Nungu, Balure, pwalugu, Kejetia, Winkongo, Tongo, Yinduri, Shiega, Balungu were those located in high flood prone areas as shown in Table 9. Biung, Zarre Yinduri were also found located within flood prone zones, while co mmunities including Tempane, Garu, Binduri, Widnaba, kokori, Pusiga, Kobori, Kugri, Bawku, Tilli were also found within low flood prone zones. In all, Talensi District, Bawku West, and some parts of the Bolgatanga Municipality were highly prone to floods, whereas Bawku Municipality, Garu, Tempane, Binduri, Pusiga, Bongo Districts are within moderate to low prone zones.

When scenario analysis were conducted, there was no significant area highly prone to flooding. Only a small proportion, less than 1 km² of the entire area, was found to be highly prone after assigning a higher weight to total annual rainfall (PRCPT), as shown in Fig. 10A. Similarly, a higher weight assigned to LULC also resulted in only a small portion of the area (less than 1 km²) being highly prone to flooding, as seen in Fig. 10B.

Furthermore, after assigning greater importance to the yearly discharge, it was noted that a significant portion of the basin (2665 km²), roughly 60%, displayed a high level of susceptibility to flooding, as depicted in Fig. 10C. This serves as clear evidence that flooding in the White Volta Basin is heavily impacted by discharge, thereby confirming NADMO's claim that the combination of heavy rainfall and the opening of the Bagre dam in Burkina Faso are the primary causes of flooding in the area. It should be noted that both heavy rainfall and water from the Bagre dam contribute to the discharge. Given the weight assigned to discharge, several co mmunities located downstream of the basin are at an increased risk of experiencing flooding, i.e., Biung, Tolla, Bingo, Balure, Pwalugu, Zongoyiri, Sinebaga, Kugri, Kajetia, Balungu, Tongo, Shia, Yale, Kpikpira, Nantinga, Zaare, Garu, Tempane, Tengzu, Kpatia, Gorogo. In terms of flood generation, discharge contributes the most, followed by rainfall and LULC, respectively.

The moderate flood-prone zone occupied the largest area (2827 km²), representing 66% of the total area of the entire basin. This was followed by the low prone zone (956 km²), accounting for 23%, and the high prone zone (502 km²), representing the smallest area at 12%. This indicates that the basin has a substantial area that is moderately prone to flooding. Areas downstream of the White Volta River are highly prone to floods, as shown in Fig. 10: 1990-A, Fig. 10: 2000-B, Fig. 10: 2010-C, and Fig. 10: 2020-D.This aligns with the work of [41] which indicated that the most significant exposure factors that contribute to flood risk, includes proximity to the Bagre Dam and location up-or downstream of the catchment.

From the Mann-Kendal test results, though the downward trend was less significant, meaning that the decrease in rainfall within the period is very minimal. The decreasing trend in the data means that rainfall availability in the coming years is likely to decrease further. The decline in precipitation, aligns with previous work by [38] which found a decreasing significant trend in rainfall when they studied rainfall variability and changes in Ghana. However, there is no agreement in the literature regarding precipitation trends; some studies report decreases [42, 43], while others note increases or stabilized precipitation patterns [44–46].

Although this study indicates a decreasing trend in precipitation, a literature search reveals that from 1999 to 2019, the downstream area of the White Volta area experienced seven unprecedented floods [47]. Proving therefore that other factors, rather than rainfall, are more likely to be responsible for the flooding that occurs in the majority of the study area.

The negative decreasing trend of precipitation also support the claim by NADMO that rainfall alone, has not caused any severe flooding incident in the White Volta Basin unless it is coupled with spillage from Bagre Dam. The water-holding capacity of the river system both upstream and downstream of the Bagre Dam becomes saturated when precipitation simultaneously occur in Ghana and the neighboring Burkina Faso. As a result, in the event of a spill, the system at downstream cannot hold the water, which results in flooding of immediate downstream areas. The decreasing trend from8 the Mann–Kendall also means that agricultural produce is on the verge of been affected, food in2security is expected to rise in the coming days.

Conclusion

The study’s objective was to conduct an appraisal of the flood risk zones of the White Volta Basin. The specific objectives were as follows: to identify flood-prone locations through mapping, to determine the causative factors of flooding and rank their contributions to floods, and to estimate the extent of flood-prone zones. In this regard, Geographic Information Systems (GIS) and Analytical Hierarchy Process (AHP) are recognized as powerful tools for identifying flood-prone areas, which can help communities become aware of areas that may endanger their properties, lives, and resources during flood incidents.

Annual total rainfall (PRCPT), slope, elevation, soil type, land use/land cover (LULC), streamflow, and distance from the river were the main flood-induced parameters used for generating the maps. The AHP weighting gave a Consistency Ratio (CR) value of 0.10, making the weighting of the parameters consistent and acceptable. The study examined four case studies using annual total rainfall data for the years 1990, 2000, 2010, and 2020, along with other flood-induced parameters. Additionally, three different scenario analyses were conducted using PRCPT, LULC, and discharge as variables. The AHP weighting resulted in total annual rainfall having the highest weight (0.5), (0.31), followed by elevation (0.21), distance from the river (0.17), discharge (0.10), slope (0.09), LULC (0.07), and soil type (0.05), respectively.

The key findings of the study showed in case study 1 that, the Talensi District, the Bawku West Municipal, and some portions of the Bolgatanga Municipality were found in the high flood-prone zone. Communities such as Balure, Nungu, Yale, Biung, Tongo, Tolla, Kajetia, Bari, and Shiega were also in this zone. The Nabdam District, the northern and middle portions of Bawku West, dominated the moderate prone zones, while Garu and Tempane were part of the districts within the low flood-prone zones. Additionally, it was also observed that areas receiving sufficient annual rainfall were more highly prone to floods when compared to areas with low rainfall. This is the confirmation of the obvious, however significant differences occur as a result of land cover, soil and slope attributes.

In all the case studies conducted, the flood-prone zone with the largest area coverage was the moderate prone zone, followed by the low prone zones, indicating that a significant portion of the basin is moderately prone to floods. The scenario analysis also revealed that land use/land cover (LULC) and PRCPT alone do not have any significant influence on flood risks. Discharge or streamflow had the greatest impact on the flood vulnerability of areas within the basin. Discharge from the basin makes almost all areas downstream of the river highly susceptible to floods. This could be attributed to the fact that a high volume of water is often released by the operators of the Bagre dam in Burkina Faso.

Author contributions

Author Contribution Statement 1—Conceived and designed the experiments; S. A., R. K. & A. B. 2—Performed the models; R. K. & S. A. 3—Analyzed and interpreted the data; R. K., S. A., G.O., J. A. & A. B. 4—Contributed reagents, materials, analysis tools or data; R. K., S. A., G. O., J. A. & A. B. 5—Wrote the paper; R. K., S. A., J. A. & A. B.

Data availability

Data sets generated during the current study are available from the corresponding author upon reasonable request. Input data on rainfall was obtained from; https://doi.org/10.1016/j.dib.2023.109115 (Ampofo, S., Annor, T., Aryee, J. N., Agyekum, J., & Amekudzi, L. K. (2023). Gridded daily rainfall data for Ghana for the period 1960–2015: approach and validation process. Data in Brief, 48, 109115. https://doi.org/10.1016/j.dib.2023.109115) and river discharge from the Water Resources Commission of Ghana. Other datasets used are publicly available and can be accessed from various online databases.

Declarations

Competing interests

The authors declare no competing interests.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Najibi, N; Divineni, N. Recent trends in the frequency and duration of global floods 2018; 10.5194/esd-9-757-2018Esd Copernicus Org;

2. Abass, K. Rising incidence of urban floods: understanding the causes for flood risk reduction in Kumasi, Ghana. GeoJournal; 2022; 87, 2 pp. 1367-1384. [DOI: https://dx.doi.org/10.1007/s10708-020-10319-9]

3. Guha-Sapir D, Hoyois PH, Below WP. R. Annual disaster statistical review 2016: the numbers and trends. Brussels: CRED; 2016. http://emdat.be/sites/default/files/adsr_2016.pdf

4. Merz, B; Blöschl, G; Vorogushyn, S; Dottori, F; Aerts, JCJH; Bates, P; Bertola, M; Kemter, M; Kreibich, H; Lall, U; Macdonald, E. Causes, impacts and patterns of disastrous river floods. Nat Rev Earth Environ; 2021; 2, 9 pp. 592-6092. [DOI: https://dx.doi.org/10.1038/s43017-021-00195-3]

5. Liu, J; Feng, S; Gu, X; Zhang, Y; Beck, HE; Zhang, J; Yan, S. Global changes in floods and their drivers. J Hydrol; 2022; 614, [DOI: https://dx.doi.org/10.1016/J.JHYDROL.2022.128553]

6. Venegas-Cordero, N; Cherrat, C; Kundzewicz, ZW; Singh, J; Piniewski, M. Model-based assessment of flood generation mechanisms over Poland: the roles of precipitation, snowmelt, and soil moisture excess. Sci Total Environ; 2023; [DOI: https://dx.doi.org/10.1016/j.scitotenv.2023.164626]

7. Blöschl, G; Gaál, L; Hall, J; Kiss, A; Komma, J; Nester, T; Parajka, J; Perdigão, RAP; Plavcová, L; Rogger, M; Salinas, JL; Viglione, A. Increasing river floods : fiction or reality?. WIREs Water; 2015; [DOI: https://dx.doi.org/10.1002/wat2.1079]

8. Ansah, SO; Ahiataku, MA; Yorke, CK; Otu-Larbi, F; Yahaya, B; Lamptey, PNL; Tanu, M. Meteorological analysis of floods in Ghana. Adv Meteorol; 2020; [DOI: https://dx.doi.org/10.1155/2020/4230627]

9. Haddad, EA; Teixeira, E. Economic impacts of natural disasters in megacities: the case of floods in São Paulo, Brazil. Habitat Int; 2015; [DOI: https://dx.doi.org/10.1016/j.habitatint.2014.06.023]

10. Arya, AK; Singh, AP. Multi criteria analysis for flood hazard mapping using GIS techniques: a case study of Ghaghara River basin in Uttar Pradesh, India. Arab J Geosci; 2021; [DOI: https://dx.doi.org/10.1007/s12517-021-06971-1]

11. Asumadu-Sarkodie, S; Owusu Phebe, A; Rufangura, P. Impact analysis of flood in Accra, Ghana. Adv Appl Sci Res; 2015; 6, 9 pp. 53-78.

12. Ghanaiantimes. Akosombo dam spillage: districts hit by floods.. farmlands, houses, and other properties affected in Greater Accra, Eastern, and Volta Regions. https://ghanaiantimes.com.gh/. Accessed 12 Dec 2023.

13. UNICEF Ghana. The Akosombo Dam Spillage. https://www.unicef.org/ghana/blog/akosombo-dam-spillage. Accessed 12 Dec 2023.

14. Bempah, SA; Olav, A. The role of social perception in disaster risk reduction: beliefs, perception, and attitudes regarding flood disasters in communities along the Volta River, Ghana. Int J Disaster Risk Reduct; 2017; 23, pp. 104-108. [DOI: https://dx.doi.org/10.1016/j.ijdrr.2017.04.009]

15. Asare-Kyei, D; Forkuor, G; Venus, V. Modeling flood hazard zones at the sub-district level with the rational model integrated with GIS and remote sensing approaches. Water (Switzerland); 2015; 7, 7 pp. 3531-3564. [DOI: https://dx.doi.org/10.3390/w7073531]

16. Atubiga, JA; Atubiga, AB; Nyade, LT; Donkor, E. Evaluating the perennial flooding on the White Volta River and the Bagre Dam Spillage on agricultural activities in the Sudan Savanna in the Upper East Region, Ghana. Int J Multidiscip Res Anal; 2023; 06, 03 pp. 934-946. [DOI: https://dx.doi.org/10.47191/ijmra/v6-i3-11]

17. Kheradmand, S; Seidou, O; Konte, D; Barmou Batoure, MB. Evaluation of adaptation options to flood risk in a probabilistic framework. J Hydrol Reg Stud; 2018; 19, pp. 1-16. [DOI: https://dx.doi.org/10.1016/j.ejrh.2018.07.001]

18. Balgah, RA; Ngwa, KA; Buchenrieder, GR; Kimengsi, JN; Balgah, RA; Azibo Ngwa, K; Buchenrieder, GR; Kimengsi, JN. Impacts of floods on agriculture-dependent livelihoods in sub-Saharan Africa: an assessment from multiple geo-ecological zones. Mdpi Com; 2023; [DOI: https://dx.doi.org/10.3390/land12020334]

19. Amoako, C; Boamah, F. The three-dimensional causes of flooding in Accra, Ghana. Int J Urban Sustain Dev; 2014; 7, 1 pp. 109-129. [DOI: https://dx.doi.org/10.1080/19463138.2014.984720]

20. Valeo, C; Rasmussen, P. Topographic influences on flood frequency analyses. Can Water Resour J; 2013; 25, 4 pp. 387-406. [DOI: https://dx.doi.org/10.4296/cwrj2504387]

21. Braimah, MM; Abdul-Rahaman, I; Oppong-Sekyere, D; Momori, PH; Abdul-Mohammed, A; Dordah, GA. A study into the causes of floods and its socio-economic effects on the people of Sawaba in the Bolgatanga Municipality, Upper East, Ghana. Int J Pure Appl Biosci; 2014; 2, 1 pp. 189-195.

22. Mensah, H; Ahadzie, DK. Causes, impacts and coping strategies of floods in Ghana: a systematic review. SN Appl Sci; 2020; [DOI: https://dx.doi.org/10.1007/s42452-020-2548-z]

23. Jackson LP, Devadason CA. Climate change, flooding and mental health. A report prepared for the Secretariat of the Rockefeller Foundation Economic Council on Planetary Health at the Oxford Martin School; 2019. https://www.inet.ox.ac.uk/publications/climate-change-flooding-and-mental-health

24. Gashaw, W; Legesse, D. Flood hazard and risk assessment using GIS and remote sensing in Fogera Woreda, Northwest Ethiopia. Nile River Basin; 2011; 6, 9 pp. 179-206. [DOI: https://dx.doi.org/10.1007/978-94-007-0689-7_9]

25. Li, C; Dash, J; Asamoah, M; Sheffield, J; Dzodzomenyo, M; Gebrechorkos, SH; Anghileri, D; Wright, J. Increased flooded area and exposure in the White Volta river basin in Western Africa, identified from multi-source remote sensing data. Sci Rep; 2022; 12, 1 pp. 1-13.[COI: 1:CAS:528:DC%2BB38XmsFaisbc%3D] [DOI: https://dx.doi.org/10.1038/s41598-022-07720-4]

26. Syam, MA; Heryanto,; Balfas, MD. Mapping of landslide susceptibility using analytical hierarchy process in Sukamaju Area, Tenggarong Seberang, Regency of Kutai Kartanegara. IOP Conf Ser Earth Environ Sci; 2019; [DOI: https://dx.doi.org/10.1088/1755-1315/279/1/012002]

27. Yin, J; Yu, D; Wilby, R. Modeling the impact of land subsidence on urban pluvial flooding: a case study of downtown Shanghai, China. Sci Total Environ; 2016; 544, pp. 744-753.[COI: 1:CAS:528:DC%2BC2MXitVSgsL%2FM] [DOI: https://dx.doi.org/10.1016/j.scitotenv.2015.11.159]

28. Hijar, G; Bonilla, C; Munayco, CV; Gutierrez, EL; Ramos, W. El niño phenomenon and natural disasters: public health interventions for disaster preparedness and response. Revista Peruana de Medicina Experimental y Salud Publica; 2016; 33, 2 pp. 300-310. [DOI: https://dx.doi.org/10.17843/rpmesp.2016.332.2205]

29. Akpoti, K; Groen, TA; Dossou-Yovo, E; Kabo-Bah, A; Zwart, SJ. Climate change-induced reduction in agricultural land suitability of West-Africa’s inland valley landscapes. Agric Syst; 2022; 200, [DOI: https://dx.doi.org/10.1016/j.agsy.2022.103429]

30. Issahaku, AR; Campion, BB; Edziyie, R. Rainfall and temperature changes and variability in the Upper East Region of Ghana. Earth Space Sci; 2016; 3, 8 pp. 284-294. [DOI: https://dx.doi.org/10.1002/2016EA000161]

31. Ampadu, B; Sackey, I; Cudjoe, E. Rainfall distribution in the Upper East Region of Ghana, 1976–2016. Ghana J Sci Technol Dev; 2020; 6, 2 pp. 45-59. [DOI: https://dx.doi.org/10.47881/168.967x]

32. Hajar, YAA. Using Analytical Hierarchy Process (AHP) to build suppliers’ selection model. Int J Acad Res Bus Soc Sci; 2017; [DOI: https://dx.doi.org/10.6007/ijarbss/v6-i12/2552]

33. Ampofo, S; Issifu, JS; Kusibu, MM; Mohammed, AS; Adiali, F. Selection of the final solid waste disposal site in the Bolgatanga municipality of Ghana using analytical hierarchy process (AHP) and multi-criteria evaluation(MCE). Heliyon; 2023; [DOI: https://dx.doi.org/10.1016/j.heliyon.2023.e18558]

34. Mokhtari, E; Mezali, F; Abdelkebir, B; Engel, B. Flood risk assessment using analytical hierarchy process: a case study from the Cheliff-Ghrib watershed, Algeria. J Water Clim Change; 2023; 14, 3 pp. 694-711. [DOI: https://dx.doi.org/10.2166/wcc.2023.316]

35. Saaty, TL; Vargas, LG. Hierarchical analysis of behavior in competition: prediction in chess. Behav Sci; 1980; 25, pp. 180-191. [DOI: https://dx.doi.org/10.1002/bs.3830250303]

36. Aditya, F; Gusmayanti, E; Sudrajat, J. Rainfall trend analysis using Mann-Kendall and Sen’s slope estimator test in West Kalimantan. IOP Conf Ser Earth Environ Sci; 2021; [DOI: https://dx.doi.org/10.1088/1755-1315/893/1/012006]

37. Aswad, FK; Yousif, AA; Ibrahim, SA. Trend analysis using Mann-Kendall and Sen’s slope estimator test for annual and monthly rainfall for Sinjar District, Iraq. J Duhok Univ; 2021; 23, 2 pp. 501-508. [DOI: https://dx.doi.org/10.26682/csjuod.2020.23.2.41]

38. Ampofo, S; Annor, T; Aryee, JNA; Atiah, WA; Amekudzi, LK. Long-term spatio-temporal variability and change in rainfall over Ghana (1960–2015). Sci Afr; 2023; 19, [DOI: https://dx.doi.org/10.1016/j.sciaf.2023.e01588]

39. Gocic, M; Trajkovic, S. Analysis of changes in meteorological variables using Mann-Kendall and Sen’s slope estimator statistical tests in Serbia. Glob Planet Change; 2013; 100, pp. 172-182. [DOI: https://dx.doi.org/10.1016/j.gloplacha.2012.10.014]

40. Hussien, SA; Mustafa, BY; Medhat, FK. Trend analysis of annual and monthly rainfall in Erbil City, Kurdistan Region. Iraq. Polytech J; 2019; 9, 2 pp. 30-36. [DOI: https://dx.doi.org/10.25156/ptj.v9n2y2019.pp30-36]

41. Smits, WK; Attoh, EM; Ludwig, F. Flood risk assessment and adaptation under changing climate for the agricultural system in the Ghanaian White Volta Basin. Clim Change; 2024; [DOI: https://dx.doi.org/10.1007/s10584-024-03694-6]

42. Dickinson, KL; Monaghan, AJ; Rivera, IJ; Hu, L; Kanyomse, E; Alirigia, R et al. Changing weather and climate in Northern Ghana: comparison of local perceptions with meteorological and land cover data. Reg Environ Change; 2017; 17, 3 pp. 915-928. [DOI: https://dx.doi.org/10.1007/s10113-016-1082-4]

43. Yaro, JA. The perception of and adaptation to climate variability/change in Ghana by small-scale and commercial farmers. Reg Environ Change; 2013; 13, 6 pp. 1259-1272. [DOI: https://dx.doi.org/10.1007/s10113-013-0443-5]

44. Haarsma, RJ; Selten, FM; Weber, SL; Kliphuis, M. Sahel rainfall variability and response to greenhouse warming. Geophys Res Lett; 2005; 32, 17 pp. 1-4. [DOI: https://dx.doi.org/10.1029/2005GL023232]

45. He, Y; Manful, D; Warren, R; Forstenhäusler, N; Osborn, TJ; Price, J et al. Quantification of impacts between 1.5 and 4 °C of global warming on flooding risks in six countries. Clim Change; 2022; [DOI: https://dx.doi.org/10.1007/s10584-021-03289-5]

46. Klutse, NAB; Owusu, K; Nkrumah, F; Anang, OA. Projected rainfall changes and their implications for rainfed agriculture in northern Ghana. Weather; 2021; 76, 10 pp. 340-347. [DOI: https://dx.doi.org/10.1002/wea.4015]

47. FloodList. Floods recorded in Ghana. FloodList; 2022. https://floodlist.com/?s=Ghana&submit. Accessed 4 Mar 2023.