Walatsi sub-catchment, bounded by 34˚10'0''E and 34˚35'0''E longitudes and 0˚25'0''N and 0˚40'0''N latitudes, is approximately 423 km² with an altitude range of 1180 m to 1532 m above sea level. It comprises ten micro-catchments and hosts commercial centers (small urban) along main roads (tarmac) and other roads (mostly murram), as shown in Figure 2.

The area experiences bimodal rain (Mar. to May, Sept. to Nov.) and dry (Dec. to Feb., Jun. to Aug.) seasons. Rain, an essential input to the overall water supply, greatly impacts human activity, natural vegetation, surface runoff, and groundwater recharge. Annual rainfall ranges from 1091 to 1840 mm, and the average monthly total is from 47 mm to 235 mm. Annual mean maximum temperature ranges from 32.34˚C to 38.15˚C, and yearly monthly evaporation ranges from 2056 to 2303 mm.

Its geology includes Mumias Granites and the Kavirondian system [10] . Soils formed on granites dominate the study area. It is enclosed by agroecological zones Upper Midland (UM) to Lower Midland (LM), suitable for agriculture, infrastructure and settlement, forest and natural vegetation, and other commercial activities [11] . It derives its economic activities mainly from agriculture, sand harvesting, and trade.

The sub-catchment has a population of 268,089 people with a density of 400 to 700 persons/km² and a growth rate of 2.3% per annum [12] . Most people live in rural areas where rivers and other available water sources strongly influence their food security and social well-being. A small percentage of households and institutions are connected to piped water served by various water distribution schemes.

Surface water is drawn from the Walatsi River system, with tributaries drawing water from micro-catchments in valleys where springs originate. Groundwater from aquifers is drawn through deep and shallow wells and scoop holes. Water tanks and pans constitute storage facilities.

Data sources included key informant interviews, geographic coordinates of water points, direct observation of geographic features, streamflow measurements, GIS analyses, and published sources, as summarized in Table 1.

Quantum Geographical Information System¹ (QGIS) software delineated catchments and its geostatistical tools for GIS analyses and mapping various sub-catchment water demands, supplies, and land cover. Microsoft Excel was deployed in data and statistical analyses, and WEAP modeling software utilized GIS and Microsoft Excel analyses to achieve the study goal.

Surface water comprised water in the river system (stretching over a 144-kilometer network) made of wetlands, springs, tributaries, and the main river (Figure 3); groundwater covered water from scoop holes, shallow wells, and deep wells (boreholes) that was drawn from Mumias granites and Kavirondian aquifers [10] on which the study area sits (Figure 3); and storage included harvested rainwater stored in water tanks and water pans.

Derived from the FDC and Figure 5, Figure 6 shows apportionable surface water after reserve allocation. The results of the study experiment in Figure 4 conducted in early October of 1.07 m³/s fell short of 0.04 m³/s of the average monthly flow estimates, attributable to the period of onset of a rainy season. The results were key in affirming the flow estimates. Q95, by allocation standards [14] , represented 0.13 m³/s, Q80 0.30 m³/s, and Q50 0.65 m³/s. Figure 6 results as an input in the WEAP model established monthly averages ranging from 0.46 to 6.78 Mm³ per year in a normal year. The monthly regular flow pattern in Figure 5 depicts the bimodal rainy and dry seasons experienced according to Jaetzold and Schmidt [11] and NASA [21] . Allocable groundwater was estimated at 0.02 Mm³, and water in storage (utilizable in dry seasons) was estimated at 0.004 Mm³ average monthly supplies, respectively, in a normal year. Total average monthly supplies ranged from 0.48 to 6.80 Mm³ in a normal year, and total annual supplies in a regular year were estimated at 31.63 Mm³ and for a leap year at 31.65 Mm³ for the entire study period. The established apportionable water supply in the sub-catchment is summarized in Figure 8 and Table 3. Water loss estimates through evaporation were insignificant and, therefore, negligible. Surface water is the predominant and main source.

There was no evidence of strategies in the region’s development plans to increase the number of water supply sources; therefore, the study assumed the established water amounts as dependable monthly and annually for allocation purposes throughout the study period.

Six types of demand identified were domestic, institutional, livestock, poultry, industrial, and irrigation. The demand was also confirmed by the land use analysis in Table 4 and Figure 9. According to Onyango et al. [22] , due to pressure from increasing population and occasioned demands in the region, land coverage for built-up areas is increasing while that for forests is decreasing, this was in agreement with the outcome of demand analyses.

The distribution of demand magnitudes in the sub-catchment is influenced by the spreading of household densities, small urban areas, health and learning institutions, and water distribution schemes, as shown in Figure 10. For instance, high household densities translate to high water needs, and access to piped water through various schemes means high water usage. The presence of small urban areas and institutions also attracts more water demand.

The study adopted two scenarios for demand and water balance simulations in the model: BUS and the PDS. BUS considered the current established demand with predefined growth rates and projections. In contrast, PDS looked into the new developments (demand) to be implemented as outlined in the development strategies alongside the BUS demand to give an overall picture of what the water use situation would be in the sub-catchment.

The outcome in Figure 11 showed growth in annual water demand from 6.45 Mm³ in 2022 to 7.36 Mm³ in 2030 in BUS and 8.23 Mm³ in PDS. PDS results in Figure 12 indicated a growth of 13.49% by 2030 from the BUS demand and by 27.59% from 6.45 Mm³ in 2022. The spike in PDS is attributed to the implementation of the planned projects (irrigation and domestic distribution schemes between 2023 and 2026). The overall average monthly demand in BUS ranged from 0.53 to 0.59 Mm³ and from 0.58 to 0.63 Mm³ in PDS for the entire study period.

The results in both scenarios showed an overall monthly water balance that ranged on average from −0.05 to 6.21 in BUS and −0.10 to 6.16 Mm³ in PDS (Figure 13), with February recording a negative balance. The negative water balance implies a deficit in water supply linked to competition for scarce water resource, eventually leading to conflicts [6] [23] . Further inquiry into the negative balance revealed an annual increasing deficit trend to 2030 in February, whose estimation is vital in planning various interventions to normalize the water balance. The annual deficit trend (in February) ranged from −0.02 (2022) to −0.09 (2030) in BUS and −0.02 (2022) to −0.15 (2030) Mm³ in PDS (Table 5).

Comparatively, the annual deficit in PDS grew by −67% in 2030 from BUS. Further, Table 5 demonstrates the deficit annual trend of PDS from BUS, with spike years in negative water balance in 2023 and 2026 when the envisaged development strategies were implemented; other years reflected expected trend deficits emanating from various applied demand growth rates. The model output (Figure 14) further showed an annual drop of the water balance from 25.18 in 2022 to 24.27 Mm³ in 2030 for BUS and 25.18 in 2022 to 23.40 Mm³ in 2030 for PDS; this represented a −0.94% drop in 2030 in BUS and −7.07% in 2030 in PDS, respectively (Figure 15).

The correlation between water balance and demand results was −0.999 in BUS and −0.9999 in PDS. The trendlines in Figure 16 illustrate the relationship in PDS. The results established a solid negative correlation in that there is an equivalent decrease in water balance with an increase in water demand in the sub-catchment, crucial information for strategies to manage water resources sustainably.

The negative water balance represents the volume of demand that goes unmet by the supply. All the demand types were treated with the same priority level, and the resultant unmet volumes were distributed, as shown in Figure 17. From the results, the percentage of unmet demand grew (steadily) annually from 0.24% (2022) to 1.16% (2030) in BUS, while in PDS, it rose (significantly) from 0.24% to 1.85%. The results indicate that the growth in demand led to a negative water balance. The solid negative correlation supported the unmet demand gap,

which meant exploitation of the reserve, detrimental to its dependents like the aquatic ecosystem [24] .