SharpGIS

How SharpGIS finds the best land

Crop rules and ready-made data of 14 November 2026. Every report says which versions made it.

In short: SharpGIS scores every small square of land on several factors (rain, soil, slope, land cover, roads…), combines the scores, and colours each square from not suitable to highly suitable. It is a well-known method (multi-criteria analysis, as in the FAO land evaluation framework), using public data. It is a good first look, not a final answer: check the best places on the ground.

1. Cover the area with squares

The study area (a district, a province, all of Zambia, or your own farm boundary) is covered with square cells, measured in metres. Small areas get small cells (tens of metres); large ones get cells up to about a kilometre, so the analysis stays quick.

2. Score each factor from 1 to 4

For every cell, each factor gets a score: 4 best, 3 good, 2 fair, 1 poor. A score of 0 rules the cell out whatever else it scores (for example very steep land, open water, towns or national parks). The scores for each crop are in the report, for example "Rainfall: 650 to 1,000 mm a year is best".

3. Combine the scores

The combined score is then grouped: under 1.75 not suitable, 1.75 to 2.5 moderately suitable, 2.5 to 3.25 suitable, 3.25 and over highly suitable.

4. Check how sure the answer is

Weights are judgement, not fact. So SharpGIS changes each weight by a quarter, down and then up, re-runs the combination, and reports how much of the land keeps the same class every time. Over 90% is very sure; under 75% means the answer depends on the weights, so read it with care. The report also names the factor the result is most sensitive to.

5. Explain every place

Click any place on a result map to see each factor's actual value there (for example "Rainfall 820 mm a year: best; Slope 9°: ruled out"), and what held it back. The report also ranks districts, or your own farm blocks, by how much suitable land they have.

The ready-made data

FactorSourceDetailKnown limits
Rainfall, and how much it variesCHIRPS v3.0 (Climate Hazards Center, UC Santa Barbara)5 km cells; average of 1991–2020A 30-year average: one season can be very different. Satellite estimates are less certain in hilly areas.
Daytime heatCHTclim (Climate Hazards Center)5 km; average daily high, November to MarchRegional; frost pockets and valleys can differ.
Soil texture and pHSoilGrids 2.0 (ISRIC)1 km; topsoil 0–30 cmModelled from soil samples and satellite data, not measured on your field. A soil test is much more reliable; upload your own soil map if you have one.
Slope and elevationCopernicus DEM GLO-30Slope worked out at 30 m, then averagedSmall gullies and terraces don't show.
Land coverESA WorldCover 2021Most common class in each cell (from 10 m)From 2021: land cleared or planted since then shows as it was.
Roads and permanent waterOpenStreetMap, via Overture MapsDistance to trunk, primary, secondary and tertiary roads; rivers, canals, lakes and dams not marked seasonalNew or small roads and boreholes may be missing.
National parksOpenStreetMap, via Overture MapsAlways left outGame management areas and other reserves are not left out: farming is allowed in many of them.

The crop rules

Each crop (maize, groundnuts, soybean, cassava, sunflower, cotton, irrigated vegetables) has starting scores and weights based on general crop requirements (as in the FAO crop database). They have not yet been checked by Zambian agronomists; that review is planned. You can ask the assistant to change them for your purpose, for example "near water" or "keep the forest".

What SharpGIS does not consider

Using your own data

You can use your own farm boundary as the study area and your own soil map (texture names or pH). The report warns you when your data covers only part of the area (the rest is left out), when some of its values weren't recognised (they get a middle score), and when the map's cells are much finer than some of the data. Every report ends with a short fingerprint: the same fingerprint means the same inputs and settings, so results can be compared and redone.