Dianne❤️

Geomatic Engineer|GIS Analyst| Cartographer|
Leader.

[Yohan @yohaniddawela]

A new geospatial foundation model can now estimate how poor your neighbourhood is, track how fast that's changing, and do it without anyone filling out a single survey form.

A team of Stanford researchers recently published Tempov, a foundation model trained on three million pairs of Landsat images spanning two decades. It takes raw satellite imagery and predicts asset wealth at the village level, across entire continents, updated in near real-time.

The benchmark numbers are worth sitting with. In Malawi and Mozambique, the model explains 87% and 74% of the variation in household wealth from satellite imagery alone. That's from six spectral bands. No census forms. No field enumerators. No mobile phone metadata.

The harder problem is tracking change, not just level. Most existing models are trained to predict a static snapshot. When you ask them to predict how wealth shifted between 2008 and 2018 in the same locations, performance collapses to near-random. Tempov holds at R² = 0.69 for Malawi and 0.46 for Mozambique on that same change-tracking task.

What makes the difference is how the model was pretrained. The researchers constructed bitemporal image pairs that maximise seasonal variance, then forced the model to learn representations that are stable across seasons but sensitive to genuine long-run economic shifts. The learned embeddings spontaneously delineate road networks, urban structure, and agricultural patterns from natural background, without ever being told to.

The scarcity problem is where it gets interesting for development economics. The standard tools for measuring poverty rely on the Demographic and Health Surveys. DHS data is spatially sparse and resurveyed infrequently. The correlation between asset wealth in Malawi's earlier and later censuses is only 0.42. In Mozambique it's actually negative: -0.71.

Tempov gets around this with a two-stage adaptation. Train on historical census data, then fine-tune to the target year using only 5% of contemporary survey points. With that 5% adjustment, it outperforms geospatial foundation models that were given 100% of available survey data. Combining a strong historical prior with minimal contemporary calibration can substitute for the full survey investment.

The researchers then deployed it continent-wide. Five models trained under cross-validation on all 34 African countries with recent DHS surveys, averaged into a single ensemble, producing 6 km × 6 km wealth maps for the entire African continent in 2015 and 2025. Roughly 80% of measured wealth inequality across the continent is within countries, not between them.

The decadal change map shows wealth gains concentrated in West and East Africa and substantial declines across parts of Southern and Central Africa. Country-level factors explain only about a third of the variation in wealth change. Local temperature trends and nearby conflict events predict the changes better than institutional-quality proxies do.

For the applied economics side: the model achieves competitive performance with 10% of available survey samples where baseline foundation models need 100%. That's not a modest efficiency gain. That's a different cost structure for poverty measurement entirely.

DHS survey rounds are already under funding pressure. The World Bank's Living Standards Measurement Surveys have become increasingly irregular. The status quo is a slow degradation in the quality and frequency of ground-truth data on living standards in the places that most need monitoring.

What Tempov suggests is that the role of household surveys may be shifting from the primary measurement instrument to the calibration anchor. You don't stop running surveys. You run fewer, target them better, and use them to tune a model that fills in the rest from orbit.

The code and weights are open-source. The continent-wide wealth maps are public. The methodology is reproducible by a national statistics office with a laptop and a moderate AWS bill.

The hard part was always getting data out of places that couldn't afford to collect it. That constraint just got significantly looser.

Link to paper: https://t.co/lNbIJp3mn6

[Yohan @yohaniddawela]

The reason satellites keep missing Africa's informal settlements isn't resolution. It's that colour imagery can't tell a slum from an ordinary dense neighbourhood, and radar can.

Getting them onto a map matters more than it sounds. Informal settlements are where a large share of urban Africa actually lives, and if a city can't see them, it can't plan water, sanitation, roads or disaster response for them. The obvious tool is satellite imagery, but the ordinary kind measures colour and brightness, and here that betrays you. A tightly packed formal neighbourhood of small concrete houses reflects light in much the same way as an informal settlement of tin-roofed shacks. To the sensor the two blur together, so the settlement gets folded into the surrounding city and vanishes from the count.

A team from Sapienza University of Rome and the University of Pavia found a way to pull them back apart, and the trick is to stop relying on multispectral images and start listening to structure. Alongside the usual optical imagery they brought in radar from the Sentinel-1 satellites, which doesn't care what colour a roof is. Radar bounces off shape and material, off the angles and clutter of how buildings sit together, and informal settlements have a very particular texture from above: irregular, densely packed, metal-roofed, jumbled at odd angles rather than laid out in neat blocks.

They fed that radar signal in through three escalating layers. First the raw strength of the return. Then the texture, a measure of how rough and disordered the radar pattern is across a patch of ground. Then a physics-guided index tuned to the specific fingerprint of a slum, the weak, scattered returns you get from small chaotic structures. To train and test it they used expert-drawn labels from very high-resolution imagery, roughly 426,000 pixels' worth, across Nairobi and Eldoret in Kenya, with a further check in Kigali, Rwanda.

The payoff shows up exactly where the problem was. Judged on how well it finds informal settlements specifically, an optical-only map scored 0.44 in the dry season, roughly a coin toss dressed up as a classifier. Adding the radar layers lifted that to 0.67, and in the wet season from 0.52 to 0.68. Overall accuracy landed around 81.6% in the dry season and 80.7% in the wet, beating the standard reference product's 70.4%, and the mislabelling of informal areas as ordinary dense housing fell by about 7 percentage points.

The seasonal detail carries more weight than it first appears. Optical imagery drifts with the weather, greener in the wet season, dustier in the dry, which is part of why maps built from it wobble month to month. The radar texture barely moved between seasons, so the settlement stayed visible either way. Obviously, a map you can only trust half the year is not much of a map, and this is one step to fixing that.

link to full article: https://t.co/botTCUZlXK

[Yohan @yohaniddawela]

Disaster response is drowning in free satellite data that responders still can't turn into a decision when the ground is shaking.

This issue is the subject of a new study from researchers at IUSS Pavia, Italy's Civil Protection Department and ISPRA, and it lands on something nobody likes to admit. Europe runs Copernicus, that pours out a continuous stream of satellite data on land, water, air and emergencies. In theory it's a goldmine for the people who manage earthquakes and floods. In practice, much of it goes unused, because it was never built around the decisions those people have to make.

Picture the night a quake hits. A responder doesn't want a beautiful map of the whole region rendered for a journal. They want to know, within the hour, how far the damage runs and which roads still carry weight. The data to answer that is overhead, collected on schedule. The product on offer is too generic, arrives too slowly, or speaks the language of a scientist rather than someone deciding where to send the trucks. So the firehose keeps flowing and the thirst goes unmet.

The researchers did something unglamorous and useful about it. Instead of launching another sensor, they sat down with the national Civil Protection Department and built a structured catalogue of what it actually needs. Every hazard on the Italian list went in: earthquakes, volcanoes, the whole water-driven family of landslides, floods, sinking land and drought, plus wildfires and industrial accidents. For each one they wrote down the real questions and the technical requirements, then laid that against what Copernicus and its sister services currently deliver.

What fell out was a map of mismatches. Plenty of products exist that no responder has a use for. Plenty of questions responders ask every emergency have no product behind them at all. And a fair amount of genuinely good data sits one translation step away from being useful, framed for research when it could be framed for a decision under pressure.

The fix they argue for is a reversal of direction. Today most satellite services are designed forwards, from what is technically easy to produce, and responders are left to make do with whatever comes out. The team wants them designed backwards, starting from the choice an officer makes at three in the morning and working out what the satellite must deliver to inform it. Same instruments, often. Different brief.

The timing isn't an accident. Italy is rolling out its own home-grown satellite constellation, called IRIDE, and the next decade of European missions is still being shaped. Decide now what these systems should answer, and you get instruments that earn their cost when the ground is shaking. Decide later, and you get more impressive data that responders still can't slot into the moment that counts.

link to full article: https://t.co/kikcIVCvar

[Marko Denic @denicmarko]

I found this amazing tool that makes it easy to create grid and flex layouts.

Check it out!

[Dianne❤️ @khayenzeli_]

@Keshi72438439 @wanjohikibui The X and Y😂😂 very funny.

[Dianne❤️ @khayenzeli_]

This is quite impressive 👏

[Sholla Ard 🇰🇪 @sholard_mancity]

Do your Geography teacher teach u this ?
This madam was recorded explaining it to tourists
Shouldn't ministry of tourism make her a tour guide or recognize her ?

Gideon Moi auctioned igathe Abel mutua

[Africa GeoPortal @AfricaGeoPortal]

Mark your diary for this webinar on Monday May 6th. #heat #urbanheat #Africa.
Register now: https://t.co/3iN53NgMJe

[Dianne❤️ @khayenzeli_]

@SufAHMED6 @MzalendoAlenga 😂😂😂 we are waiting

[Dianne❤️ @khayenzeli_]

@AfricaFactsZone Kenya typing and deleting.😂

[Dianne❤️ @khayenzeli_]

@MzalendoAlenga Availability of resources 😂😂 open source data

[RCMRD GMES project @RCMRD_GMES]

RCMRD under @GMESAfrica program has developed a wetland change map and an animated spatial time series of Lake Olbolosat found in central Kenya, from #sentinel #radar #data showing the decreasing levels of water and change from 2018 to 2023.