Evaluating extreme climatic indices in a mountainous region using satellite data: a framework to advance climate resilience and sustainable development

A valuable tool for improving sustainable development is historical weather and climate data (HWCD), which, when properly utilized, can enhance the lives of vulnerable communities susceptible to climate change impacts (CCIs) (Dinku et al. 2018). Batwa communities, in the highlands of Kigezi-southwestern Uganda, are particularly vulnerable to climatic threats such as flash floods (FFs), soil erosion, and disease incidence (van Bavel et al. 2020). The frequency of these climate hazards has intensified of late, posing more danger to this Indigenous group (Satyal et al. 2021). The fact that over 70% of the Batwa households live in perpetual deficiency of basic needs such as food, housing, clothing, and education raises concern (Scarpa et al. 2021), especially in the face of a changing climate. The hilly, solitary, and remote terrains in which they live further exacerbate the challenge (Luliro et al. 2022). Harnessing HWCD is essential for these indigenous communities since it facilitates climate-sensitive early-warning system design, planning, research, and the identification of climate risks such as catastrophic weather events. Additionally, the HWCD provides a baseline for future climate projections as well as establishing the climate change (CC) trajectory (Ngoma et al. 2021).

Considering statistics indicating that Africa is the continent most exposed to CCIs (IPCC 2023), harnessing the HWCD becomes even more essential to investigate changes in climate indices (CIs) for early-warning purposes. This should receive the utmost interest since the majority of African economies depend on climate-sensitive industries like agriculture for development. About 80% of agriculture worldwide is rain-fed, according to Dinku (2019), and in Africa, this percentage exceeds 95%. In Africa, agriculture contributes significantly to GDP and provides a living for a larger proportion of the continent's rural population. The application of HWCD in location-specific local contexts would be crucial for African nations to meet their nationally established development plans and the 2030 Agenda SDGs (Dorward et al. 2020). This is due to the fact that successful CC adaptation techniques are frequently context-specific, taking into account local sociocultural, sectoral, and geographic variables that differ from place to place.

Exploitation of the HWCD in Africa is inadequate because the majority of National Meteorological and Hydrological Services (NMHS) gather the data with poor quality checks, thus reducing its relevance to the local community (Dinku 2019). Faniriantsoa & Dinku (2022) reported that the primary challenge to climate data availability and quality in Africa is attributed to a sparse and uneven distribution network of weather stations. The coverage is typically lower in rural areas, where livelihoods are most vulnerable to the adverse impacts of CC. Other problems included low-quality station observations with many missing data points and the inaccessibility of the NMHS's available data. Dinku (2019) claims that insufficient dissemination tools and capacity, scarce resources, and other factors make data accessibility in Uganda a major challenge. In order to overcome these challenges, the present investigation employed rainfall and temperature datasets from CHIRPS and NASAPW, respectively. The selection was based on the datasets' free accessibility, high resolution, and global coverage, which provide a more dependable and contextualized outlook at the local level (Dhanesh et al. 2020; Halimi et al. 2023; Tan et al. 2023; Du et al. 2024).

The study focused on Kabale, Kanungu, and Kisoro (KAKAKI) districts of southwestern Uganda (see Figure 1). The geographical latitudes and longitudes of the study area are −1.5° S to −0.3° S and 29.4° E to 30.3° E, respectively. The area is inhabited by a significantly greater number of Batwa communities. The study area is bounded by Ntungamo to the east, the Democratic Republic of Congo to the west, Rukungiri to the north, and Rwanda to the south. According to Satyal et al. (2021), Batwa settlements are scattered throughout the hilly, solitary, and remote parts of these districts, comprising 10–20 households. These areas are frequently susceptible to climatic threats such as FFs, soil erosion, and disease incidence (van Bavel et al. 2020). A study by Scarpa et al. (2021) revealed that about 70% deficiency in basic living standards prevails among the Batwa people. The study location is situated in the tropical Savannah climate zone, with a bimodal (MAM and SOND) rainy season each year. The major crops produced under the rain-fed farming system in the area include Irish potatoes, bananas, coffee, sweet potatoes, beans, and cereals (Scarpa et al. 2021).

The traditional methods for measuring rainfall at the local level are rain gauges and weather radars (Shahid et al. 2021). However, because of their limited distribution throughout the research area, it is difficult to record the spatio-temporal variability in precipitation, particularly in the Kigezi region's mountainous areas. The employed daily precipitation data was obtained from CHIRPS, which is accessible at https://app.climateengine.org/climateEngine. The preference for CHIRPS data is attributable to its high resolution (0.05° × 0.05°) and its integration of satellite imagery with ground station data, offering a more comprehensive raster and reliability (Basheer & Elagib 2019; Dhanesh et al. 2020; Du et al. 2024). A study by Shahid et al. (2021) affirms that CHIRPS precipitation data reliably complements ground station data in addition to providing continuous precipitation monitoring over large temporal and spatial scales. Moreover, CHIRPS provides free daily datasets available for extended periods dating back to 1981. To download the precipitation data (PRD), the area of interest was initially zoomed in and a polygon drawn to represent it on the map. Additionally, the CHIRPS 4.8 km dataset was selected under the ‘climate and hydrology’ section, and the ‘variable’ section was modified to ‘precipitation.’ The units were set to ‘milimetres,’ while the ‘statistics’ section was set to ‘total’ to indicate cumulative precipitation. The required data's time period was adjusted to range from January 1, 1983 to December 31, 2023. A graph was generated by clicking on ‘Get Time Series,’ and after clicking on the ‘download button’ in the top right corner of the graph, the PRD was downloaded as a CSV file (Obura et al. 2024c). These steps were repeated for each of the three KAKAKI districts containing the Batwa. Eventually, an average of the three districts was obtained to serve as the representative rainfall dataset.

The NASAPW was exploited to retrieve the DT_Min. and DT_Max. datasets at 2 m from https://power.larc.nasa.gov/data-access-viewer/. The NASAPW is a comprehensive, freely accessible, and dependable satellite data source for temperature with high resolution and global coverage, which enables studies to be conducted even on a local scale. NASA implements stringent quality control procedures to guarantee the accuracy of its data. Furthermore, numerous studies have reported on the validity and reliability of NASAPW TED (Aboelkhair et al. 2019; Marzouk 2021; Carrara et al. 2023; Halimi et al. 2023; Tan et al. 2023). The user community ‘Agroclimatology’ was chosen after opening the ‘data access viewer’ on the NASAPW in order to download temperature data (TED). Each of the three KAKAKI districts' lat/long coordinates was input, and the temporal average was set to ‘daily.’ The output file format was set to CSV, and the timeframe was set from January 1, 1983, to December 31, 2023. The minimum and maximum temperatures (MMTs) were nominated for download when the ‘Temperatures/Thermal IR Flux’ section under the ‘Select Parameters’ section was opened. After pressing the ‘Submit’ button, daily native resolution NASA Power (NASAP) CERES/MERRA-2 data was downloaded in a CSV format (Obura et al. 2024c). The MMTs were represented by averaging the three district datasets.

The consistency and quality of the daily input data were checked before the indices were computed. A crucial requirement for a robust analysis of climate time series is homogeneity. This study adopted free and user-friendly RH-tests software to perform the data quality control diagnostics (DQCD). Based on the penalised maximal t-test (PMT) or f-test (PMF), the programme can identify and correct multiple change points in a time series, as demonstrated by J. Du et al. (2020) and validated by Liu et al. (2017). Reference stations are necessary for PMT when conducting a homogeneity study, but PMF can be employed as an absolute approach, which means it can be used in isolation or in the absence of neighbouring stations for comparison. Therefore, the PMF method was used in this investigation.

The CIs required to comprehend the climatology of the study location were generated utilising ClimPACT v3.1.6 in this investigation. ClimPACT3 is an open-source tool programmed by the Expert Team on Sector-specific Climate Indices (ET-SCI) for the World Meteorological Organisation (WMO). ClimPACT v3.1.6 was chosen because it allows for customization based on local conditions, enhancing the relevance of findings for a specific region (Murara & Mendonça 2019; Dubey et al. 2022). It is also incorporated with diverse internationally recognized CIs for determining precipitation extremes, heat waves, and cold waves, making it a reliable tool for long-term climate data analysis (Nakaegawa & Murazaki 2022). Furthermore, ClimPACT integrates with various climate data formats, promoting wide applicability, supporting research, and climate adaptation planning (WMO 2016). It employs daily parameters against monthly datasets because several significant pieces of information are concealed in monthly climate data, which may be crucial to multiple fields, including agriculture, health, and water resources. Employing daily data to produce CIs provides a comprehensive understanding of multiple variables, including the periodicity and intensity of extreme rainfall and measurements of extremely wet, dry, hot, or cold periods that influence society's economy.

Furthermore, the Mann–Kendall test at 0.05 and Sen's slope estimator were adopted for trend analysis (Agarwal et al. 2021; van der Walt & Fitchett 2021; Şen & Şişman 2023). Ten (10) indices (see Table 1) implemented in this investigation were nominated based on their significance to health (H), agriculture and food security (AFS), as well as water resources and hydrology (WRH). By capturing significant trends and anomalies, CIs helps in summarizing complex climate data to inform decision-making in sectors such as agriculture, health, and water management. In addition, the Standardized Precipitation Evapotranspiration Index (SPEI) was selected in contrast to the Standardized Precipitation Index (SPI) due to its ability to amalgamate the multi-temporal nature of the SPI with its computational simplicity and sensitivity to changes in evaporative demand ignited by temperature trends and fluctuations (Vicente-Serrano et al. 2010). The WMO approved the SPEI as the reference drought index for more effective determination of drought sensitivity and climate risk management of drought hazards on sensitive systems (Mishra & Aadhar 2021).

The HWCD in CSV format for the study location was imported and analysed in R-Instat 0.7.16.50. The data was checked, reshaped, and arranged using the ‘prepare’ tab. Then, the ‘climatic’ tab was applied to prepare climatic summaries and determine the onset and cessation of the rains and the duration of each season. Furthermore, the statistical analysis was performed for the bimodal rainy seasons using R-Instat. The criterion to establish the rainy season's onset was set to three consecutive days with total precipitation (TPR) ≥ 20 mm and the absence of nine consecutive dry days (CDDs) in the first 21 days. Onset dates were established for the MAM and SOND seasons. The last day of the season to receive the PR > 10 mm was designated as the the season’ cessation. Finally, the output from R-Instat was exported to an Excel spreadsheet for additional modification, where plots for seasonal TPR and the extent of each season were generated. R-Instat was selected for this study because of its easy-to-use interface, which makes complicated statistical processes simpler. It offers quality control tools and also accomodates huge climatic datasets. Advanced statistical techniques like regression, time series analysis, and climatic trend evaluations are integrated into the software, simplifying analysis. R-Instat is an open-source and free software (Fundi et al. 2024), making it easily accessible for use in this study.

A significantly positive SPEI could be valuable for agricultural productivity by increasing soil moisture and water availability, favouring crop growth. This has a positive implication on SDG 2 (Zero Hunger) and especially contributes to achieving SDGTs 2.1, 2.2, and 2.3. In the WRH sector, a positive SPEI trend could improve water supplies but also calls for the improved infrastructural need to manage the excess water. This has direct implications for the efforts to achieve SDGTs 6.1, 6.2, and 6.3. Healthwise, wetter than usual conditions pose significant health risks, including the emergence of water- and vector-borne diseases, communicable diseases, and psychological disorders due to the evacuation of people from flood-vulnerable zones. A study by van Bavel et al. (2020) claims that increased moisture, which supports larger mosquito populations and other insects that spread diseases like dengue fever and the West Nile virus, is a contributing factor in the proliferation of vector-borne diseases in wetter environments. This calls for concerted and prioritized efforts from relevant stakeholders since a laxity directly hampers the achievement of target 3.3 of SDG 3 (Healthy lives and well-being for all).

The MAM's onset window ranged from March 1 to March 31; its cessation window spanned from May 1 to June 30. The SOND season's onset window stretched from September 1 to September 30, and its cessation window was from November 15 to December 15. The MAM and SOND season commencement and cessation dates for the most recent 20 years of research are presented in Table 3, which was generated using R-Instat software.

From 1983 to 2023, the extent of the MAM season fluctuated more than the SOND season, with ranges between 40–110 days and 73–105 days, respectively (see Figure 5(c)). The shortest MAM season (40 days) occurred in 1984, whereas that for SOND (73 days) was experienced in 2020. The year 1995 recorded the longest MAM season (110 days) from March 1 to June 19, while 2006 experienced the longest SOND season (105 days), with September 1 as the onset and December 15 as the cessation of the season. The increasing MAM seasonal variability poses a risk for climatic hazards like FFs (Ainuddin et al. 2017), which severely impact infrastructure, livelihoods, and food security. These consequently impede sustainable development. Furthermore, the results also depicted more PR inconsistency in the MAM season than in the SOND season, fluctuating between 308–655 mm and 378–623 mm, respectively. The area experiences MAM and SOND seasons' average near-normal rainfall of 404 and 531 mm, respectively. Despite this, the MAM rains were more intense, with an intensity of 135 mm/month compared to SOND's 132 mm/month. Although the MAM season for the year 1984 was the shortest (40 days), it received the highest intensity of rainfall of 253 mm/month. This can trigger climatic hazards for communities. The year 2017 recorded the lowest MAM PR totals (308 mm), whereas the year 2018 received the greatest MAM rains (655 mm). The SOND season's minimum PR totals (378 mm) occurred in 1993, while the largest amounts (623 mm) were received in 2010. Heavy PR over shorter periods exacerbate land degradation (Tofu et al. 2022), destroying terrestrial ecosystems and thus hampering the efforts to achieve SDG 15 (Life on Land) especially targets 15.1–15.6 and 15.8. These findings necessitate the urgency for sustainable land management practices to be sought to preserve ecosystems and promote sustainable agriculture in the face of a changing climate.

The current study utilized the CHIRPS and NASAPW remote sensing data to investigate fluctuations in extreme CIs. The ClimPACT3 and R-Instat models were applied in the assessment process. The findings show that the region generally receives wetter than normal weather, with PR values from 1,006 to 1,489 mm. There was a significant (p-value < 0.05) increase in the ASDPR > the 95th percentile. Conversely, the DT_Min. and DT_Max. varied from 13.1 to 14.5 °C and 24.0 to 26.4 °C, respectively. These observations are particularly imperative for the agricultural, health, and water resources sectors. Furthermore, CISs would aid in the achievement of SDG 1, SDG 2, and SDG 13 targets by enhancing food security and climate resilience in Batwa farmer communities. For the reason of making well-informed decisions about water management, the participants in the hydrological and water resources sectors need information on temperature extremes, rainfall variability, and SPEI trends. In order to prepare for an epidemic of food-borne, water-borne, vector-borne, or air-borne diseases, which could help achieve SDG 3, SDG 6, and SDG 11 targets, these outputs are also critical to the health sector. In summary, even though the CHIRPS and NASAPW time series datasets have been modified to improve homogeneity, it is vital to keep in mind that certain parts of these records may still be non-homogeneous when evaluating index changes. To increase reliability, future research should think about using precise station observation data.

M.E. contributed to methodology, formal analysis, writing – original draft; W.J. contributed to data curation, writing – review and editing; and D.O. contributed to conceptualization, visualization, writing – original draft; writing – review and editing.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.