PhD research · first full run · dry seasons 2017, 2021, 2025
Kampala Heat Atlas
How Kampala grew, where it is hottest, who lives in the heat, and what trees and rooftops could do about it. Six analyses on one 30 m grid covering the city and its growth frontier.
0.10–0.55° N · 32.35–32.85° E · UTM 36N · Landsat 8/9, Sentinel-2, Open Buildings 2.5D, Overture, ERA5-Land, FABDEM
Growth+38%built-up land, 658 → 909 km² (2017–2025)
Heat236 km²in the hottest class in 2025 (220 km² in 2017)
Form+6.3 °Cpacked low-rise blocks vs the city land mean (other dense +5.3 °C)
Trees−1.4 °Caverage cooling from 50 km² of new canopy, for about 554,000 people
Solar33 TWha year from 2.0 M roofs (22 GWp technical potential)
Module 1 · Urban growth
The city added a third more built-up land in eight years
Built-up land rose from 658 to 909 km² between 2017 and 2025. Growth runs east along the Jinja Road and north into Wakiso.
The core densified: sparse fabric turned dense across about 112 km².
Classifier agreement with its training labels is 82–83 % (kappa 0.78–0.80) on held-out samples.
Area of each land-cover class, three epochs.
Before citing. Agreement with training labels is not map accuracy. 350 validation points still need to be labelled by eye. Tree-cover gains probably include seasonal cropland confusion.
Module 2 · Thermal classification
Compact neighbourhoods run 5 to 7 °C above the city
The hottest class covers 236 km² in 2025. About 132 km² are hot in all three epochs: these are the persistent hotspots.
Compact mid- and low-rise zones are the hottest Local Climate Zones. Large low-rise sheds and warehouses warm fastest. Dense tree cover sits about 2 °C below the mean.
Surface temperature covers 100 % of the area, against about 60 % for the standard product, and agrees with it at r = 0.81–0.95.
Surface heat by Local Climate Zone, 2017 and 2025.
Module 3 · Settlements
Blocks that packed in fastest warmed most
Small, tightly packed low-rise blocks sit +6.3 °C above the city land mean, about 1 °C hotter than other dense fabric (+5.3 °C).
Blocks that densified into packed low-rise fabric between 2017 and 2025 warmed +1.0 °C relative to the city. Blocks already dense in 2017 warmed +0.4 °C.
Checked against the literature. Against the 58 informal settlements mapped by the slum dwellers' federation (SDI/NSDFU/ACTogether, 2014), georeferenced from their map book: established informal settlements are as hot as other dense fabric (+5.2 to +5.9 °C vs +5.4 to +5.5 °C) and warmed less since 2017 (+0.35 to +0.54 vs +0.62 to +0.66 °C; ranges over ±250 m georeferencing shifts). The fast warming belongs to blocks that densified, not to informality as such. The morphology layer covers 21–29 % of the federation's settlements.
Module 4 · Where to plant
47 km² where heat, people and bare ground meet
Each extra 10 % of canopy goes with 0.39 °C cooler surfaces in built-up cells.
The very-high priority class holds about 319,000 residents and almost no trees today (0.3 % canopy).
Planting half the open land in high and very-high cells adds 50 km² of canopy and cools those cells by about 1.4 °C.
Residents and plantable land per priority class.
Before citing. The cooling model explains about a third of the variation in surface temperature (R² 0.32). Surface cooling is larger than air-temperature cooling.
Module 0 · The cube
Sun overhead, breeze from the lake
At 0.3° N the noon sun never drops below 66°. Flat or 5–10° panels are best; tilt changes yield by under 1.5 %.
East and west walls take about 2.1 times the direct sun of north and south walls. Long axes should run east–west.
From 09:00 the Lake Victoria breeze builds from the south, peaking at 15:00 (74 % of hours from S or SE). Long south facades catch it; north–south gaps let it through.
Wind direction through the day, ERA5-Land 2015–2024.
Central Kampala, 5.4 × 5.4 km: FABDEM terrain plus Open Buildings heights, coloured by 2025 surface heat relative to the window.
Module 5 · Rooftop solar
Roof space is not the constraint
2.0 million roofs could host about 22 GWp and generate 33 TWh a year, at 1,516 kWh per kWp.
The largest tenth of roofs alone exceeds Uganda's 2024 distribution sales of 4.7 TWh (Umeme).
The real limits are daytime demand, grid absorption, storage and finance. That is the question for the solar chapter.