Decadal spatio-temporal dynamics of drought in semi-arid farming regions of Zimbabwe between 1990 and 2020: a case of Mberengwa and Zvishavane districts

Drought severity and frequency are increasing in tropical regions and its occurrence is not uniform in space and time. The study assessed the spatio-temporal dynamics of agrometeorological drought in Mberengwa and Zvishavane districts between 1990 and 2020. An empirical research design was adopted in this study. GIS and remote sensing techniques were used to determine and analyze agricultural drought conditions and the standardized precipitation generator was used to compute SPI for meteorological drought monitoring and analysis. Microsoft Excel 2011 was adopted for the analysis of drought coverage and performing the Mann–Kendall trend test for precipitation trend analysis. Results indicated that the area covered by drought declined between the 1990 and 2000 and 2001–2010 decades before increasing during the 2011–2020 decade. Results indicated significant spatial dynamics of drought severity and frequency in Mberengwa and Zvishavane districts and the majority of wards experienced increased frequency of severe droughts during the 2011–2020 decade. It was highlighted that meteorological drought was not evenly distributed across all decades. Dry conditions in the first months (October and November) of the rainy season have been indicated during the 1990–2000 decade, followed by wetter conditions during the 2001–2010 decade and comparatively drier conditions in the same months during the 2011–2020 decade. This indicated the late onset of the rainfall season during the 1990–2000 and 2011–2020 decades compared to the 2001–2010 decade. It was also shown that rainfall cessation was earlier during the 2001–2010 and 2011–2020 decades and late during the 1990–2000 decade. The study concludes that drought has increased both in severity and frequency in Mberengwa and Zvishavane districts; hence, there is a need for more support to drought resilience-building initiatives in these areas. The government of Zimbabwe is advised to set up climate research centers in all provinces of the country to improve the availability of climate change–related data which is useful when addressing the impacts of climate change.


Introduction
Due to climate change, drought incidence and severity have increased and myriads of social and economic sectors are being affected. Definitions have been advanced and shaped by systems and contexts in which drought impacts are perceived and felt. Drought manifests with a varying frequency almost in all parts of the world and in all types of economic systems of both developed and the developing worlds (Wilhite 1992;Frischen et al. 2020). Drought was classified by Wilhite and Glantz (1985) as a meteorological, agricultural, socioeconomic, and hydrological drought. Wilhite (2000) further stressed that these droughts are linked in some way. Despite the fact that temperature, wind speed, evapotranspiration, and soil moisture have a role in the incidence and progression of drought, the fundamental driver of every drought event is a lack of precipitation (Heim 2002;Smakhtin and Schipper 2008;Kumar et al. 2009;Vicente-Serrano et al. 2010;Mishra and Singh 2011). Drought is a severe climatic hazard and an extreme meteorological phenomenon that results from a lack of precipitation, resulting in a water scarcity for certain applications (Szinell et al. 1998). Precipitation deficit over a short period of time can be 1 3 a meteorological drought but not agricultural or hydrological drought. When this deficit prolongs, it leads to soil moisture deficit (agricultural drought) and another period beyond may result in hydrological drought (Wilhite 2000).
The twenty-first century began with a series of droughts that affected the whole globe. Severe to exceptional intensity droughts covered 7-16% and extreme droughts affected 2-6% of the world's land area (Kogan et al. 2013). Several droughts affected key agricultural regions in the northern hemisphere from 2000 to 2012, notably from 2010 to 2012. The area affected by severe drought grew from 8% in 2014 to 14% by the end of 2015 (Greenhalgh 2016). By the end of 2015, 30% of the world's land was experiencing drought, with 14% experiencing severe or extreme drought (Osborn et al. 2016). Between 2011 and 2012, the USA was hit by severe droughts. The combined drought index for Europe in the 2019/2020 season showed drought conditions over a few core regions including Ireland and the UK, Belgium, the Netherlands and Germany, the northwest Balkans, and Scandinavia. Some of the affected areas were in the peripherals located in central Italy, eastern France, Poland, Belarus, and Greece. However, parts of Romania, Moldovia, and northwest Ukraine showed signs of recovery compared to previous years. Western and central Europe also suffered severe drought from 2018 to 2020 (Barbosa 2020).
Literature on droughts in Asia shows that several droughts affected northern China which also experienced one of the most severe droughts that led to the drying up of the Yellow river in 1997 (Zhang et al. 2009). This was the worst drought in Pakistan and Afghanistan in 50 years (Barlow et al. 2002;2016;Zhang and Zhou 2015). An extreme drought hit Sichuan and the Yangtze River basin in 2006-2007, causing a drop in rice, potato, and bean production. In 2008-2009, the drought also hit Yunnan and Southwestern China, causing agricultural losses (Barriopedro et al. 2012;Yang et al. 2014). North China was also hit by a severe drought that lasted from the autumn of 2009 to the spring of 2010. (Yang et al. 2012). Since the 1970s, aridity has worsened over Northern China, and the occurrence of extreme droughts in the Central part of Northern China, North East China, and the eastern part of North West China has also increased.
In tropical Africa, the Congo River basin and most of Central Africa experienced a severe drought which lasted from 2000 to 2006. Remote sensing images showed severe vegetation stress over the Congo river basin and central Africa during this time period (Calow et al. 2010;Zhou et al. 2014). The period 2008-2010 saw another drought hitting most countries in East Africa especially countries within the horn of Africa. According to Zaitchik et al. (2012), the drought between 2008 and 2010 in the lower eastern part of the horn of Africa was characterized by prolonged rainfall deficits which actually pushed this drought to 2011 (Masih et al. 2014;Nicholson 2014). Southern Africa experienced drought during the 2018/2019 season and only northeastern South Africa and southern Mozambique experienced floods during that season. Worst drought conditions were experienced in North Namibia, South Angola, North Zimbabwe, the whole of Lesotho, and South Botswana during the 2018/2019 season (Masante et al. 2019). Negative anomalies of rainfall worsened over the whole of Zimbabwe, southern Mozambique, Zambia, and northeast Botswana during the 2019/2020 season. Given these spatial and temporal dynamics of drought in different parts of the globe, it can be noted that past decades have been experiencing severe droughts which threatened the livelihoods of many people especially in developing countries thus affecting the achievement of sustainable development goals 1 and 2. This is because the majority of the population in developing countries are in rural areas where they depend on rain-fed agriculture hence frequent droughts compromise the livelihood sources of these people. However, mitigating the impacts of drought require information on historical droughts, their trajectories, severity, and spatio-temporal distribution so as to come up with an informed decision to counter associated impacts. Southern Zimbabwe is largely arid-semi-arid where precipitation is so erratic (450-500 mm) and temperatures are high (averaging 28 °C) (Manatsa et al. 2020). These characteristics already make this area drought prone hence the need to study drought spatial and temporal distribution at the district level for informed planning for drought resilience building. Mberengwa and Zvishavane districts are part of this area; however, their selection was based on grounds that they have similar climatic conditions but are dominated by soils of different water retention capacities (Mupepi and Matsa 2021). Therefore, the need to confirm the perceived similar climatic conditions and whether different soils have an influence on drought severity motivated the selection of both districts for this study. Against this background, this article assesses decadal spatio-temporal dynamics of drought in the Mberengwa and Zvishavane districts of Zimbabwe between 1990 and 2020 with the main thrust of the study being to (1) determine the temporal coverage and severity of drought in Mberengwa and Zvishavane districts, (2) analyze the spatial severity of drought in Mberengwa and Zvishavane wards between 1990 and 2020, and (3) to assess the seasonal distribution of meteorological drought in Mberengwa and Zvishavane districts between 1990 and 2020.

Study area
The study was conducted in the Midlands province's adjacent districts of Zvishavane and Mberengwa, at an elevation of 1039 m above sea level. The total areas of Mberengwa and Zvishavane are 5066 km 2 and 2476 m 2 , respectively (Zimbabwe national population statistics 2012). According to FAO (2016), Zimbabwe's total area is 390,760 km 2 , making Mberengwa and Zvishavane districts 1.93% of the total size of Zimbabwe. Zvishavane is located 97 km west of Masvingo on the main Bulawayo-Masvingo route (20° 20′S; 30° 02′E) (Fig. 1), whilst Mberengwa is located approximately 24 km south of Zvishavane (20° 29′ 0″ S, 29° 55′ 0″ E). Both districts are in agro-ecological zones 4 and 5a, with an annual rainfall of 450-650 mm and less than 650 mm, respectively, and an annual average temperature of about 28 °C (Manatsa et al. 2020). High temperatures in these areas increase evapotranspiration, increasing the likelihood of agricultural droughts. Seasonal moisture deficits and severe mid-season dry spells are common in these regions due to the unpredictable nature of precipitation, resulting in increased agro-meteorological drought occurrences.
Soils in Mberengwa and Zvishavane districts are Sandy Ustalfs which require consistent moisture to maintain agricultural production; however, the unpredictable and sparse rainfall puts crop and animal production at risk, as moisture required to support crops and pastures for livestock is rarely available. The geology of the area is linked to the Great Dyke, which consists of layered mafic intrusions linked to economically important metals like platinum, chromium, nickel, vanadium, copper, titanium, iron, and tin, providing communities in these areas with alternative livelihood sources through mining activities (Badlock et al. 1991).
Zvishavane has a population of 52,734 people, with 27,815 women and 24,919 men (Zimbabwe Central Statistical Office (CSO) 2012 report), whilst Mberengwa has a population of 185,757 people, with males accounting for 86,764 and females accounting for 98,993. In Zvishavane and Mberengwa, mining is one of the main sources of income (Zimbabwe Poverty Assessment Study Survey Summary 2003). Despite the challenges posed by changing climatic conditions, rural residents in both districts continue to grow cash crops such as cotton, sunflowers, staple cereals maize and sorghum, and vegetables, as well as livestock for survival.
Agriculture is vulnerable in Mberengwa and Zvishavane districts due to erratic or unreliable precipitation, as communities rely on rainfall for crop and livestock production. As a result of frequent crop failures and livestock deaths, primarily due to water scarcity, the majority of people are at risk of food insecurity and starvation. Large deposits of minerals such as platinum, gold, and asbestos, which are found in Zvishavane, and gold, tantalite, and emeralds, which are abundant in Mberengwa, however, provide an alternative source of income through employment in mining companies as well as participation in small-scale and artisanal mining activities.

Methodology
This study employed the empirical research design which is based on the use of verifiable evidence in order to arrive at research outcomes. Empirical research design was adopted because it is based on evidence obtained through scientific data collection methods (Borry et al. 2006). In this study, scientific measurements of drought in both Mberengwa and Zvishavane districts using Geographic information system (GIS) and remote sensing techniques were employed. These constituted the computation of the vegetation condition index (VCI) and temperature condition index (TCI) using ArcMap 10.5 software and the computation of the standardized precipitation index (SPI) using the SPI generator software. These mathematically computed indices allowed for the determination of agricultural and meteorological droughts in the Mberengwa and Zvishavane districts between 1990 and 2020. GIS and remote sensing were found ideal for assessing surface water dynamics in Mberengwa and Zvishavane due to their ability to provide a synoptic spatial and temporal visualization of drought.
Landsat images with a spatial resolution of 30 m (Table 1) were acquired from the United States Geological Survey (USGS) website for the period 1990 to 2020. Landsat 4-5 images were downloaded for the period 1990 to 2014, and Landsat 8 images were downloaded for the period 2015 to 2020.
The research maintained consistency in using Landsat images for the whole study period in order to maintain mapping uniformity using images of the same spatial resolution. Landsat 4-5 images were used for the period 1990 to 2014 because it was the only Landsat satellite platform that was perfectly functional during this period and Landsat 8 images were acquired for the period 2015 to 2020 because this satellite was launched to succeed Landsat 4-5 thus available to provide images of improved quality for this period (USGS 2015). These Landsat images were used for the computation of NDVI, VCI, and TCI for the characterization of agricultural drought. Images for the month of March were used for drought monitoring because maximum chlorophyll in vegetation and croplands is experienced in March when peak growth is attained (Yang et al. 2015;Alemayehu et al. 2017); hence, deviations from normal vegetation conditions during this month would be a reliable depiction of drought due to moisture deficit even during the previous months.
The removal of the no data value region and conduction of geometric and radiometric corrections to remove distortions and radiometric noise constituted the pre-processing stage (data cleaning). Mberengwa and Zvishavane district shapefiles were created and used to extract study areas from Landsat images. Image mosaicking was done for Landsat 4/5 images since these images could not cover the whole study area in a single swath.
Many researchers (Di et al. 1994) who studied drought employed NDVI to monitor vegetation status that shows whether there is drought or not. The use of NDVI is based on differences in spectral responses of healthy and unhealthy vegetation. Healthy vegetation reflects less in the visible electromagnetic spectrum due to chlorophyll absorption in this region whilst it reflects more in the near-infrared spectrum due to internal reflectance of the mesophyll spongy tissue of healthy green leaf. In contrast, stressed vegetation reflects more in the visible electromagnetic spectrum and less in the near-infrared spectrum (Campbell 1987). These differences in spectral response by both drought-stressed and healthy vegetation assists in determining drought and its spatial extent. NDVI can be calculated as the ratio of red and the NIR bands of a sensor system and is represented by the following equation: where RED and NIR are reflectances in the red and nearinfrared bands respectively. Due to difficulties in accurately interpreting NDVI values in heterogeneous terrains (Kogan 1987) as the case in both Zvishavane and Mberengwa districts, and the weakness that NDVI does not clearly show the weather impact separately, VCI as suggested by Kogan (1997), was adopted in this research. The calculated NDVI was used to compute VCI in this research.
The vegetation condition index was first suggested by Kogan and Sullivan (1993) to indicate the status of vegetation cover as a function of maximum and minimum NDVI experienced in a specific area over a specific period of time. The vegetation condition presented by VCI was in the form of a percentage with values ranging from 0 to 100%. Values from 50 to 100% indicate optimum to above normal conditions whilst values close to 0% signify extremely dry conditions. These values assisted in determining agrometeorological drought severity in different areas of Mberengwa and Zvishavane districts as they showed how much vegetation has responded to climate (Fig. 2). Vegetation condition index was computed as: where NDVI j is the current year NDVI, NDVI min is the multiyear minimum NDVI, and NDVI max is the multi-year maximum NDVI.
(1) NDVI = NIR − RED NIR + RED (2) VCI = NDVI j − NDVI min NDVI max − NDVI min × 100 According to Thenkabail et al. (2004), the temperature condition index was suggested by Kogan (1997) to reflect vegetation response to temperature. When using TCI, higher temperature corresponds to severe drought conditions, moderately high temperatures correspond to mild drought, and optimum temperatures correspond to normal conditions. In this research, TCI was used to confirm drought conditions detected using VCI and SPI. TCI was computed as: where BT is the brightness temperature, BT max is multi-year maximum brightness temperature, BT min is the multi-year minimum brightness temperature, and BT j is the current year brightness temperature. Maximum and minimum BT was calculated based on long-term records of remote sensing images for particular periods. Low TCI values indicate hot weather conditions and high values show milder conditions.
The standardized precipitation index was developed by Mckee et al. (1993) for the characterization of drought. SPI is the number of standard deviations that the observed value would experience from the long-term mean of a normally distributed random variable. This index was adopted to determine deviations of precipitation from the normal expected from 1990 to 2020 in the Mberengwa and Zvishavane districts. SPI was used to quantify precipitation deficits over the studied period thus determining meteorological drought. Six-month SPI (SPI-6) was adopted in this research to incorporate all months within the rainfall season (October-March) in Zimbabwe. This SPI was computed using 1990-2020 monthly precipitation records obtained from the meteorological services department. The standardized precipitation index (SPI) was computed using the SPI generator software: where Pi is the seasonal precipitation, Pm is the seasonal mean, and SD is the standard deviation of the long-term record.
The study produced average vegetation condition and temperature condition index maps for 1990-2000, 2001-2010, and 2011-2020 decades and seasonal SPI-6 (for October-March) and 1990-2020. Therefore, each decade had 2 maps showing the average TCI and VCI (Fig. 2). Average TCI and VCI per decade were determined by (4) SPI = Pi − Pm SD finding mean cell statistics for 10-year TCI and VCI raster images within each decade using the cell statistics function of Arc Map 10.5 software. Therefore, TCI and VCI for all years within each of the study decades were averaged to get the average TCI and VCI for each decade.
The spatial and temporal severity of drought in both Zvishavane and Mberengwa districts was analyzed in ArcMap 10.5 software using a raster calculator and spatial analyst tools to produce VCI and TCI thematic maps showing the distribution of drought in the Zvishavane and Mberengwa districts. The raster calculator and cell statistics tools were used to compute TCI and VCI for characterizing drought conditions. The cell statistics function was used to average TCI and VCI to determine average decadal drought conditions for presentation in the form of thematic maps. The Mann-Kendall trend test was used to determine trends in precipitation distribution in the Mberengwa and Zvishavane districts. Microsoft Excel 2013 was used to analyze drought coverage statistics and creation of graphs for the presentation of decadal drought area coverage.

Temporal coverage and severity of drought in the Mberengwa and Zvishavane districts between 1990 and 2020
Based on VCI and TCI indices, drought conditions were categorized into extreme-severe, moderate, and mildno drought classes using the Kogan (2002) guidelines (Table 2). Thermal stress (indicated by TCI) and water deficit stress (indicated by VCI) were used to detect drought conditions (Kogan 2002;Winkler et al. 2017 in the Mberengwa and Zvishavane districts. Average drought conditions per each decade have been determined through finding the mean VCI and TCI values of each year's image pixels over the 10-year period in each decade so as to come up with average drought conditions experienced by each image pixel during each decade (Liang et al. 2017). This allowed for the analysis of average drought conditions for areas in both Mberengwa and Zvishavane districts during each of the three studied decades (1990-2000, 2001-2010, and 2011-2020).
For the purpose of analyzing spatial coverage and severity of drought between 1990 and 2020, the area under each drought category was calculated and compared across decades.

Spatio-temporal coverage of drought in Mberengwa and Zvishavane districts based on VCI
Based on average vegetation condition index values, the area under mild-no drought conditions dominated in both Mberengwa and Zvishavane districts during the 1990 to 2000 decade as it constituted 47.2% in Mberengwa district and 35% in Zvishavane district. This was followed by the area under moderate drought conditions which constituted 31.6% of Mberengwa and 33.1% of Zvishavane districts respectively. The area under severe drought conditions during the same decade was 21.2% in Mberengwa and 31% in Zvishavane district. This signifies that greater parts of Mberengwa and Zvishavane districts experienced drought during the 1990-2000 decade despite that considerable proportions of these districts experienced mild to no drought conditions. When comparing the two districts, Mberengwa had a larger proportion of its area under mild-no drought conditions whereas Zvishavane had a larger proportion of its area under drought. The dominance of soils with less clay content in Zvishavane than in Mberengwa (Mupepi and Matsa 2021) can be one of the main reasons why Zvishavane is comparatively drier than the Mberengwa district. These kinds of soils according to Balasubramanian (2017) have poor moisture retention capacity; hence, they quickly dry up when exposed to high temperature and less precipitation over a long period of time (Fig. 3). Average vegetation condition index (VCI) values during the 2001-2010 decade confirmed that the area that experienced mild-no drought conditions was the largest in Mberengwa constituting 49.4% followed by the area under moderate drought condition (46.5%) and area under severeextreme drought condition (4.1%). In Zvishavane, the area under moderate drought conditions dominated, constituting 53.4% of the total area, followed by the area under moderate drought (41%) and severe-extreme drought conditions (5.5%). In Mberengwa district, despite the dominance of areas that experienced mild-no drought conditions, the total area under drought (both moderate and severeextreme) constituted the greater part of the district. However, in Zvishavane, overall, the proportion of area under drought conditions was larger than that of Mberengwa, a scenario likely to be driven by the dominance of sandy soils in Zvishavane which are poor in terms of moisture storage thus exposing them to more agricultural drought  (Balasubramanian 2017). In general, the area under severe to extreme drought was the smallest in both districts but the combined area under drought conditions in these districts was larger than that which experienced mild-no drought conditions. During the 2011-2020 decade, the average VCI values indicated that both Mberengwa and Zvishavane districts had a larger proportion of their area under severeextreme drought conditions (40% in Mberengwa district and 41% in Zvishavane district). In both districts, the area under moderate drought was the second largest, constituting 35% in Mberengwa and 34.6% in Zvishavane districts. The area that experienced mild-no drought conditions was the smallest in both districts that is 25.1% in Mberengwa and 24.4% in Zvishavane. Overall, the area under drought conditions in this decade was the largest with severe-extreme drought conditions dominating in both districts. conditions. In Zvishavane, the area that experienced severeextreme drought conditions was the second largest constituting 33.5% of the total area whilst 25.5% of the area experienced mild-no drought conditions. This shows that the largest proportion of both Mberengwa and Zvishavane districts experienced drought conditions whilst smaller areas experienced mild-no drought conditions. When comparing the two districts, Zvishavane had more of its area under drought compared to Mberengwa though both of them were dominated by drought conditions (Fig. 4).

Spatio-temporal coverage of drought in Mberengwa and Zvishavane districts based on TCI
During the 2001-2010 decade, the average temperature condition index (TCI) indicated that more than half of both the Mberengwa (52%) and Zvishavane (54.9%) areas experienced moderate drought conditions. This was followed by areas that experienced mild-no drought conditions during the decade which accounted for 34.6% in Mberengwa and 28.9% in the Zvishavane district. Only 13.5% and 16.2% of the area experienced severe-extreme drought conditions in Mberengwa and Zvishavane districts respectively. Both districts were dominated by moderate drought with a considerable area under mild-no drought and very small proportions that experienced severe-extreme drought between 2001 and 2010. Overall, both districts were dominated by drought but Zvishavane had more of its area under drought conditions compared to Mberengwa district.
During the 2011-2020 decade, the average temperature condition index (TCI) showed that 48.1% of the Mberengwa area experienced severe-extreme drought conditions compared to 41.6% in Zvishavane. For the Zvishavane district, it was indicated that 44.3% of the area experienced moderate drought whereas in Mberengwa, 38.3% of the area experienced the same drought conditions. This was followed by only 13.6% and 14.1% of the Mberengwa and Zvishavane areas respectively which experienced mild-no drought conditions during this decade. Overall, the average temperature condition index indicated that Mberengwa experienced more severe drought conditions than Zvishavane during this decade though the margin of difference was slight. These findings indicate fluctuations in drought severity across the three study decades (1990-2000, 2001-2010, and 2011-2020). Both VCI and TCI drought indices showed that drought declined during the 2000-2010 decade before a phenomenal surge during the 2011-2020 decade. When compared to the 2001-2010 decade, the 1990-2000 decade experienced more drought. This can be attributed to the most severe drought of all time that affected Mberengwa and Zvishavane during the 1991/1992 season (Sachikonye 1992), followed by the 1993/1994 and the 1997/1998 season droughts which are among the most severe droughts in Zimbabwe. These droughts were due to the El Nino phase of the ElNino Southern Oscillation (ENSO) cycle that affects drought occurrence in Zimbabwe and other southern African countries (Tadross et al. 2005;Manatsa et al. 2008

Decadal spatial dynamics of drought in Mberengwa and Zvishavane districts between 1990 and 2020
Despite some variations in the area covered by mild-no drought, moderate drought, and severe-extreme drought as indicated by the vegetation condition index (VCI) and temperature condition index (TCI), the general pattern of drought conditions was the same in both districts. Both indices indicated a decline in areas covered by severe-extreme drought, an increase in areas under moderate drought, and an increase in areas under mildno drought conditions between the 1990-2000 and 2001-2010 decades. Both indices also indicated an increase in areas under severe-extreme drought conditions and a decline in areas under moderate and mild-no drought conditions between the 2001-2010 and 2011-2020 decades. As far as average drought conditions per decade are concerned, results indicated changes in spatial severity of drought conditions in both Mberengwa and Zvishavane districts over the past three decades. However, some areas indicated to have been constantly affected by specific drought conditions (Fig. 5).
During the 1990-2000 decade, average TCI and VCI agreed in indicating that wards 36, 1, 13, and 34 to the west, southwestern parts of ward 32, most of ward 2 to the north, wards 25, 24, and parts of ward 26 to the south, and parts of wards 21 and 22 to the southeast of Mberengwa experienced severe-extreme drought conditions. In Zvishavane, wards 6, 3, and northern parts of ward 5 to the west, wards 2, 4, and 9 to the north, and ward 15 to the southeast of the district experienced severe-extreme drought conditions. Both indices indicated that some parts of central (including wards 7, 8, 17) and eastern (including wards 5, 6, and 19) Mberengwa experienced moderate drought conditions during the 1990-2000 decade (Fig. 5). For Zvishavane, it was highlighted that wards 5 and 11 to the west and ward 18 to the southeast experienced moderate drought conditions. Both VCI and TCI indicated mildno drought conditions in parts of wards 2, 6, 20, and 30 in Mberengwa district. In Zvishavane, they agreed that the eastern part of ward 3, most parts of ward 7, eastern parts of ward 17, and western parts of ward 18 experienced mild-no drought conditions during the 1990-2000 decade.
Since VCI is more based on vegetation response to moisture deficit (Kogan and Sullivan 1993;97) and TCI is based on vegetation response to thermal stress (Thenkabail et al. 2004), some differences were noted in the severity of drought conditions in some parts of both Zvishavane and Mberengwa districts. During the 1990-2000 decade, TCI indicated that most wards to the southeast of Mberengwa were in mild-no drought conditions in as far as thermal stress is concerned whereas VCI indicated moderate drought conditions based on the moisture deficit response of vegetation. In wards 30, 31, and 29 to the southwest, ward 4 to the northeast, and ward 35 to the west of Mberengwa, there was higher thermal stress than moisture deficit as TCI indicated severe-extreme drought whereas VCI indicated moderate drought conditions. Some areas experienced more moisture deficit than thermal stress as the case with wards 14 and 1 of Zvishavane where VCI indicated moderate drought conditions with TCI indicating mild-no drought conditions. Average VCI and TCI for the 2001-2010 decade indicated that all wards to the south, southeast, southwest, and central parts of Mberengwa except wards 11, 12, and Southern parts of ward 35 experienced moderate drought conditions. Both indices agreed on the dominance of moderate drought conditions in most of the southern, southwestern, western, northern, northwestern, and central parts of the Zvishavane district. Both TCI and VCI indicated that most wards to the western part of Mberengwa including wards 1, 36, 13, 34, parts of ward 11 and 37, ward 2 and parts of ward 3 to the north, and wards 31 and 32 to the south west experienced severe extreme drought conditions during the 2001-2010 decade. In Zvishavane, greater parts of ward 14 to the east, wards 16, 15, 17, and 18 to the southeast experienced severe to extreme drought conditions. These indices also confirmed that parts of wards 4 and 6 to the northeast, wards 9, 17, and 35 at the central part, wards 14, parts of ward 11, 30, and 34 to the southwest of Mberengwa experienced mild-no drought conditions during the 2001-2010 decade. In Zvishavane, wards 1 and 2 to the north, some parts of ward 15, and parts of ward 19 to the south experienced mild-no drought conditions as indicated by both indices.
Despite both indices agreeing on the spatial distribution of drought in Mberengwa and Zvishavane districts, some areas experienced more thermal stress than moisture deficit stress and others suffered more moisture deficit stress than thermal stress. Parts of wards 14, 30, and 34 southwest of Mberengwa district and parts of ward 9 to the south of Zvishavane, wards 7, 12, and 10 at the center, and wards 4, 3, 2, and 1 to the north of Zvishavane experienced moderate moisture deficit stress as shown by VCI. In Mberengwa district, ward 5 proved to have experienced more thermal stress than moisture deficit as TCI indicated severe-extreme drought whereas VCI indicated mild-no drought conditions in most parts of this ward.
During the 2011-2020 decade, both VCI and TCI values indicated that most wards to the west (1, 36, 13, 34, 11, and 37) and east (6, 19, 20, and 21), wards 24 and 22 to the south, and ward 2 to the north of Mberengwa district experienced severe-extreme drought conditions. For Zvishavane, all western wards except wards 12, 11, and 5 and all central and eastern wards experienced severe-extreme drought conditions. Wards to the southern parts of the district and ward 18 to the southeast of the district also experienced severeextreme drought conditions. Both indices agreed that some parts of ward 36 to the east of Mberengwa, parts of ward 2 to the north, and parts of ward 17 at the center experienced moderate drought conditions. In Zvishavane, it was indicated that some parts of ward 19 to the south, parts of ward 4 to the north, and parts of wards 17 and 14 to the southeast and east respectively experienced moderate drought during the 2011-2020 decade.
However, wards 14, 33, 30, 29, and 32 to the southwest, ward 35 to the east, ward 3 to the north, and parts of wards 22 and 28 to the southeast of Mberengwa, wards 1 and 2 of Zvishavane to the north and most parts of ward 17 to the southeast of Zvishavane experienced less thermal stress than moisture deficit as TCI and VCI indicated mild-no drought conditions and severe to extreme drought conditions respectively. Conversely, wards 25 and 26 to the south, ward 4 to the northeast of Mberengwa, ward 19 to the south, and parts of 5 of Zvishavane experienced more thermal stress than moisture deficit stress as TCI indicated severe to extreme drought conditions whilst VCI indicated moderate drought conditions in these areas.
Findings revealed that none of the wards in both Mberengwa and Zvishavane districts frequently experienced moderate drought conditions during all study decades, an indication of dynamics of droughts biased more towards severe-extreme drought conditions. Wards 5, 6, 7  (2001( , 2005( -2006( , 2007( -2011( ) than the 1990( -2000( and 2011( -2020( decades (NOAA 2021. These were confirmed by the US National Oceanographic and Atmospheric Administration, and usually, these conditions are associated with drought events in Southern Africa including Zimbabwe (Manatsa et al. 2008). Generally, the findings signify that there were some changes in drought conditions of most wards during the study period as indicated by shifts from one drought condition to another in most areas of both Mberengwa and Zvishavane districts. Spatial and temporal variability in ward-level drought severity in Mberengwa and Zvishavane has been confirmed by precipitation records obtained from AGRITEX which indicated variations in monthly precipitation from ward to ward thus indicating localized precipitation variability as one of the key factors that determine spatial and temporal variability in drought at a local scale (Manjowe et al. 2018). These results show that drought is now highly variable in terms of place and time within different areas (wards) of the same districts.

Seasonal distribution of meteorological drought between 1990 and 2020 (based on the standardized precipitation index)
Drought categorization was done based on McKee et al. (1993) interpretation values (Table 3). The standardized precipitation index for 6 months (SPI-6) (October to March) was computed to indicate the temporal distribution of meteorological drought during cropping seasons from 1990 to 2020.
Standardized precipitation index (SPI) values for Mberengwa and Zvishavane districts during the 1990-2000 decade indicated significant temporal dynamics of meteorological drought during the 1990-2000 decade. During this decade, the month of October experienced moderately dry conditions on average, both in Mberengwa and Zvishavane districts (Table 4). However, extremely dry conditions were experienced during the 1991/1992 and 1993/1994 seasons in both districts.
Mberengwa district also experienced extreme drought during October of the 1996/1997 season. For Zvishavane, October of the 1994/1995 season was in moderately wet conditions whilst in Mberengwa, moderately wet and very wet conditions were detected during 1999/2000 and 1994/1995 respectively. The remaining years had near-normal drought conditions during the same month.
For the month of November, no year experienced extremely dry conditions but moderately dry conditions to wet conditions were experienced in all years. In Zvishavane 1992Zvishavane /1993Zvishavane and 1996Zvishavane /1997Zvishavane seasons experienced extremely wet conditions whilst 1995Zvishavane /1996Zvishavane and 1998Zvishavane /1999 seasons had very wet conditions during the month of November. For Mberengwa, season 1998/1999 experienced very wet conditions whilst seasons 1994/1995 and 1999/2000 had moderately wet conditions. On average, during this decade, both districts were in near-normal but wetter conditions during the month of November.
As for the month of December, on average, both districts experienced near-normal but drier-than-normal conditions as indicated by negative SPI values (0.59 for Zvishavane and 0.37 for Mberengwa). However, in Zvishavane, December of seasons 1991December of seasons /1992December of seasons , 1994December of seasons /1995December of seasons , and 1997December of seasons /1998 was extremely dry whilst in Mberengwa, these same conditions were experienced during the 1997/1998 season with season 1991/1992 comparatively better as it experienced severely dry conditions. In Zvishavane, the month of December had very wet conditions during the 1992/1993 and 1996/1997 seasons whereas moderately wet conditions were experienced during the 1995/1996 season. In Mberengwa district, only seasons 1992Mberengwa district, only seasons /1993Mberengwa district, only seasons , 1996Mberengwa district, only seasons /1997Mberengwa district, only seasons , and 1998Mberengwa district, only seasons /1999 experienced a moderately wet month of December.
As for the month of March, during the 1990-2000 decade, both Zvishavane (− 0.09) and Mberengwa (− 0.24) districts experienced near normal but drier drought conditions though Zvishavane was comparatively wetter. Both districts experienced extremely dry March during the 1992/1993 season. Zvishavane experienced severely dry March during the 1997/1998 season, moderately dry conditions during the 1993/1994 season, and very wet conditions between the 1994/1995 and 1996/1997 seasons. However, Mberengwa had another extremely dry March during the 1993/1994 season, and a very wet month of March during the 1994March during the /1995March during the , 1996March during the /1997March during the , and 1999March during the /2000 seasons.
During the 2001-2010 decade, SPI values indicated variations in drought conditions in both Mberengwa and Zvishavane districts (Table 5) For the month of November, on average, both Zvishavane and Mberengwa districts experienced nearnormal but wetter conditions as shown by 0.16 and 0.12 SPI values respectively (Table 5) For Mberengwa, seasons 2001/0202, 2004/2005, 2005/2006, 2006/2007, and 2007  During the 2011-2020 decade, on average, Zvishavane experienced moderately dry October with an SPI value of − 1.08 whereas Mberengwa district experienced nearnormal but drier conditions as shown by an SPI value of − 0.78 (Table 6). Driest conditions during the month of October were experienced during seasons 2018/2019 and 2019/2020 seasons in Zvishavane whilst wettest conditions during the same month were experienced during the 2012/2013 season when the SPI value indicated moderately wet conditions. In Mberengwa, the driest conditions in the same month were experienced during seasons 2014/2015, 2017/2018, and 2019/20 when SPI values showed extremely dry conditions whereas the wettest conditions during this month were experienced during season 2012/2013.
On average, the decade 2011-2020 experienced nearnormal but drier November conditions in both Zvishavane (− 0.70) and Mberengwa (− 0.01) districts (Table 6) Manatsa et al. (2008) and Frischen et al. (2020) who omitted the 2010/2011 season among drought years in Zimbabwe. This is because these scholars were more focused on the agriculturally significant drought that affected the whole rainfall season not just precipitation shortage within specific months.
Zvishavane had near normal but drier conditions (− 0.16) on average, during the 2011-2020 decade whilst Mberengwa experienced moderately dry conditions (− 1.45) on average. In Zvishavane, the driest conditions in the month of March were experienced during the 2010/2011 season (extremely dry conditions) whilst the wettest conditions were experienced during the 2015/2016 season when SPI showed extremely wet conditions. For Mberengwa, the driest conditions in the month of March were experienced during the 2010/2011, 2011/2012, 2012/2013, 2017/2018, and 2019/2020 seasons. This made the decade 2011-2020 the driest as far as the month is concerned. This indicates early rainfall season cessation in Mberengwa and Zvishavane districts, a scenario also observed by Tadross et al. (2005) in the whole of southern Africa and Zimbabwe as a whole. Overall, seasonal drought conditions from 1990 to 2020 showed increasing frequency as shown by a decline in the frequency of extremely wet and very wet conditions at the expense of extremely dry to moderately dry conditions.
The month of March in both districts proved to be dry though the 1990-2000 decade was comparatively better (near normal but drier than normal conditions). When comparing the two districts regarding drought conditions for the month of March, Mberengwa was drier than Zvishavane throughout the study period. This was due to lower precipitation in Zvishavane compared to Mberengwa as indicated by precipitation records from the meteorological service department. Low precipitation during the month of March in Mberengwa might have been due to the weakening of the ITCZ more than westerly cloud bands during dry years in Zimbabwe, a scenario that induces more severe drought conditions especially for mid-summer in the far south and southwestern Zimbabwe (Manjowe et al. 2018). Therefore precipitation gradient in these cases lowers towards Mberengwa which is south of Zvishavane. However, higher precipitation and comparatively less drought conditions in Zvishavane during October and November can be attributed to the movement of the ITCZ from north to south as the summer begins (Manjowe et al. 2018); hence, Zvishavane received precipitation-inducing conditions first than Mberengwa which experiences these conditions latter.
Results of the Mann-Kendall trend test showed that the p-values for all months are greater than the significance level alpha = 0.05, which indicates no statistically significant trend in precipitation distribution between 1990 and 2020. This indicates that precipitation distribution during months within the cropping season in both Mberengwa and Zvishavane districts did not follow a well-defined statistically positive or negative trend but fluctuated over the study period.

Conclusion
The study assessed spatio-temporal dynamics of agrometeorological drought in Mberengwa and Zvishavane districts between 1990 and 2020. Both VCI and TCI drought indices showed the same drought trajectory between 1990 and 2020. It was shown that areas covered by drought conditions declined between the 1990 and 2000 and 2001-2010 decades before increasing during the 2011-2020 decade. Findings indicated that the 2011-2020 decade experienced more droughts followed by the 1990-2020 decade whilst the 2001-2010 decade was comparatively less droughty. This was an indication of an increase in drought frequency in both Mberengwa and Zvishavane districts between 2011 and 2020 after a decline between 2001 and 2010.
Results indicated that none of the wards in Mberengwa and Zvishavane districts has been constantly under moderate or no drought conditions which shows that all wards experienced mostly severe-extreme drought conditions. However, the majority of wards experienced moderate drought conditions only during the 2001-2010 decade. Almost all wards have been in severe to extreme drought conditions more frequently during the 2011-2020 decade. Findings therefore indicate significant spatial dynamics of drought conditions in both the Mberengwa and Zvishavane district wards.
It was highlighted that meteorological drought was not evenly distributed across all decades. Dry conditions in the first months (October and November) of the rainy season have been indicated during the 1990-2000 decade, followed by wetter conditions during the 2001-2010 decade and comparatively drier conditions during the same months during the 2011-2020 decade. This indicated the late onset of the rainfall season during the 1990-2000 and 2011-2020 decades compared to the 2001-2010 decade. It was also shown that rainfall cessation was earlier during the 2001-2010 and 2011-2020 decades and late during the 1990-2000 decade. Therefore, drought was heterogeneously distributed both spatially and temporally in Zvishavane and Mberengwa districts.

Recommendation
In light of the findings from this study, the following recommendations are suggested: The Meteorological Services Department of Zimbabwe is encouraged to improve the spatial distribution and number of rain gauges in all districts of Zimbabwe so as to improve spatial analysis of meteorological drought in Zimbabwe. The government of Zimbabwe is advised to set up climate research centers in all provinces of Zimbabwe with a mandate to conduct climate change and associated impact researches. This will improve the availability of climate change-related data which is useful when addressing the impacts of climate change. The Meteorological Services Department of Zimbabwe is also advised to consider remote sensing methods of precipitation measurement which will improve meteorological data availability as well as reduce gaps in meteorological data like missing precipitation records. The Government of Zimbabwe is encouraged to support scientific monitoring of drought in the country through ensuring the availability of software and highquality datasets that allow for more detailed mapping of drought. Drought researchers are advised to adopt both meteorological and remote sensing methods of drought mapping which ensures solid mapping of drought since these methods complement each other in drought mapping. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Code availability Not applicable.

Declarations
Ethics approval Approval was granted by the Midlands State University to carry out the research as well as to publish under its name. All sources were properly cited to avoid plagiarism.