Capstone: Site Suitability & a Neighbourhood Plan

Every real GIS project, however grand, is five moves: acquire, prepare, analyse, plan, communicate. It looks linear on a diagram and never is in practice - you will reach the analysis and discover a layer is in the wrong CRS, and loop back. That looping is normal and healthy. The one thing you must not do is skip a move: an analysis on unprepared data, or a plan with no honest map, produces a result that is confidently wrong. Treat the diagram as a checklist you keep returning to, not a track you ride once.

Before any data, write the brief in one sentence and decide your criteria. For the anganwadi: it should sit near where young children live, be within an easy walk of homes, avoid flood-prone and steep ground, and not duplicate an existing centre's catchment. Those plain-English rules become your analysis - so choosing them honestly, up front, is the most important step of all.

Acquire. Everything here is free. Take population and built-up context from OpenStreetMap (via QuickOSM) and from Bhuvan's thematic layers; terrain from Bhuvan's CartoDEM for slope; a ward boundary from a state portal or digitised from a Survey of India Open Series sheet; and existing facilities as points from OSM or the local body. Note each source's date and licence as you go - the metadata habit from the last lesson starts now, not at the end.

Prepare. This is where projects are quietly won or lost. Reproject every layer into one projected, metre-based CRS - EPSG:32643 for this part of India - so distances and areas are in real metres; leaving one layer in EPSG:4326 is the classic capstone bug. Clip everything to the ward so you are not analysing the whole city. Clean the obvious errors, and join any tabular data - a household count, say - to its boundary. When this step is done properly the analysis is almost anticlimactic; when it is skipped, the analysis lies.

The heart of the project is suitability analysis - the idea Ian McHarg made famous by stacking transparent map layers on a light table long before computers, and which GIS now does with arithmetic. Each criterion becomes a layer scored on a common scale (say 1 = poor to 5 = ideal): a slope layer where gentle ground scores high, a distance-to-homes layer where near scores high, a flood layer where safe scores high, a distance-to-existing-centres layer where far scores high. Then you combine them - the weighted overlay (or weighted sum): multiply each scored layer by a weight reflecting its importance, with the weights adding to one, and sum the layers into a single suitability surface from low to high.

Where do the weights come from? You can set them by reasoned judgement, or derive them more formally with the Analytic Hierarchy Process (AHP), which turns pairwise comparisons ("is nearness to homes more important than avoiding slope, and by how much?") into consistent weights. Either way, write the weights down and justify them - the weighting is the value judgement in the whole analysis, and burying it is how a subjective choice masquerades as an objective map. The output is not "the answer"; it is a transparent surface that says, given these criteria and these weights, here is where suitability is highest.

Plan. A suitability surface is not yet a decision. Read it with judgement: identify the highest-scoring cluster that is also a plausible, available parcel, cross-check it against anything the analysis could not capture (land ownership, a road you know floods, a community preference), and name a preferred site with one or two alternatives. The GIS narrows the field and defends the shortlist; the planner still chooses.

Communicate. The finest analysis dies if the last map is unreadable. Build a proper layout sheet: a clear title, one uncluttered map showing the ward, the suitability surface and the chosen site, and the furniture a non-specialist needs - a legend, a north arrow and a scale bar - plus a caption stating the CRS, the data sources and their dates. The reader should grasp the recommendation in ten seconds and be able to check your method if they wish. That sheet, not the analysis behind it, is what a councillor, a client or a community will actually see - so it deserves as much care as everything upstream.

Step back and look at what the capstone exercised. You acquired open data under real licences (Module 2). You wrangled CRS, clipped and joined (Modules 0 and 3). You ran vector and raster overlay (Modules 4 and 5). You reasoned about weights and multi-criteria judgement, and you communicated with cartographic care (Modules 9 and 10). If you saved the analysis as a model or script (11.2), documented it to standard (11.3), and could have stored it in a spatial database (11.1), you did professional-grade GIS end to end. The tools were free; the discipline was yours.