A catchment-area model

Inscribing the landscape

Conventional views suggest that the Early Bronze Age I–III periods (c. 3600–2400 BC) saw the appearance of the first urban communities in the southern Levant. During this period we see the emergence of larger communities, fortified sites, and large public buildings — though not at the same scale as the regions of the northern Levant or southern Mesopotamia.

One reading of the evidence treats these communities as small-scale versions of the third-millennium city-states in southern Mesopotamia and northern Syria.(Ben-Tor, 1992 Ben-Tor, A. (1992). The Archaeology of Ancient Israel. (Translated by R. Greenberg.) Yale University Press. ) An alternative view holds that the communities of the southern Levant in this period are no more than "corporate villages" — but that the developments we see still suggest significant modification of the landscape; that for the first time communities inscribe themselves on the landscape.(Chesson, 2000 Chesson, M. S. (2000). "Ceramics and Daily Life in the EBA Household." In Ceramics and Change in the Early Bronze Age of the Southern Levant, 365–379. ) (Philip, 2001 Philip, G. (2001). "The Early Bronze I–III Ages." In The Archaeology of Jordan, 163–232. )

One site, one question

This site focuses on a single case study: Tell esh Shuna North, a Chalcolithic / Early Bronze Age I settlement about 11 km south of Lake Tiberias on the eastern edge of the Jordan Valley floor. James Mellaart called it "one of the most important early sites in the entire Jordan Valley."

The archaeological deposits at Shuna preserve a "rare and important sequence" of continuous Chalcolithic-to-EBA occupation, with detailed palaeobotanical and faunal evidence. The same density of environmental evidence — einkorn, emmer, barley, lentils, sheep, goat, and the increase in cattle over time — is why the original thesis selected Shuna as its primary case study for landscape-catchment modelling. That density of direct evidence is what makes a quantitative catchment model possible. Without faunal + botanical assemblages, the population × diet × area arithmetic at the heart of the model has nothing to ground against.

The terrain

Shuna sits at the foot of sharp eastern terraces. The valley floor immediately west is alluvial flat, deposited by the slow retreat of Lake Lisan over ~10,000 years; the terrain east rises rapidly to over 1000 m elevation. Within ~6 km of the settlement the elevation spans roughly −265 m to +220 m.

This is the landscape an inhabitant of Shuna would have had to walk over to reach a field, a pasture, or a neighbouring site. You can toggle the map's 3D view to inspect the steep eastern terraces. The agricultural catchment depends not on Euclidean distance but on cost — the time-and-effort cost of moving across this terrain.

Walking time, not metres

The catchment engine builds a walking-time cost surface seeded at the settlement. Every cell in the DEM is assigned a cost — in seconds of walking time — based on Tobler's hiking function applied to the slope of every edge in the 8-connected grid. Tobler, W. (1993). Three Presentations on Geographical Analysis and Modeling. NCGIA Technical Report 93-1. The velocity model is asymmetric: gentle downhill is the fastest direction. Dijkstra's algorithm propagates accumulated cost outward from Shuna.

You can see the result on the map — the colourmap shows walking time from the settlement. Cells closer in terms of effort are lighter; cells beyond ~11 hours of walking are treated as unreachable. The valley-floor reach is broader than the uphill reach toward the terraces, even though both are the same number of metres away.

How much land for 50 hectares?

Given a target hectarage to cultivate, the model finds the smallest cost-distance ring around Shuna whose suitable land (within an elevation band, in this scenario) adds up to that target. A binary search over thresholds settles in 9 iterations to within ±2.5% of the target.

The blue overlay on the map is the converged catchment for a 50 Ha target. Achieved area: 49.14 Ha. Threshold: ~22 minutes of walking time. Unlike the desktop pipeline, the web explorer is fully interactive: you can click to relocate the settlement seed, draw travel barriers (like rivers or lakes) directly on the landscape, and toggle a clean Sentinel-2 satellite view to study routing.

Run the model yourself → or read the methodology for the algorithm and the data provenance for what the DEM represents.

Beyond Shuna — the regional layer

Shuna was one site among many. Chapter 6 of the thesis steps out from the single-settlement case study to consider it in conjunction with 125 other Early Bronze I sites in the southern Levant. The same engine that finds Shuna's catchment underpins a regional view that compares territorial allocation methods between neighbouring settlements.

Eight named EB I sites near Shuna are seeded as picker options on the Regional explorer. Thiessen polygons (equal-weight tessellation), Renfrew & Level's XTent model (site-size weighted), and walking-time path networks computed with Tobler hiking costs over the DEM — all the territorial-allocation methods § 6.3 and § 6.4 compare — are live, with way-point optimisation following the thesis's § 6.4 recipe.

Open the Regional explorer → Pick a few sites, hit Compute, and switch between methods to see how the territorial answers differ.