Generative Snow Architecture

Introduction

A Family Archive, Read Against the Climate

Inspired by nostalgic family photographs of skiing, this experimental Bachelor's thesis critically examines the future of winter sports in an era of climate change. Focusing on the Axamer Lizum resort near Innsbruck, our demographic and historical analysis revealed a massive, continuous surge in international tourism.

Yet our climate mapping exposed a contradictory reality: snow levels are drastically declining. To accommodate the crowds despite the lack of snow, conventional ski infrastructure increasingly excavates the mountain, leaving permanent, devastating scars on the alpine ecosystem.

Family ski archive, early-to-mid 20th century, Tyrol

Family archive — early winter tourism in Tyrol, the starting point of this research.

Ski infrastructure development over decades, Axamer Lizum and surroundings

Decades of infrastructure growth — from gondolas to parking lots, tracing the resort's expansion.

Climate Mapping

Reading the Mountain Through Satellite Data

To move from anecdote to evidence, we analyzed five years of satellite imagery (2019–2023) across a 50×50 km area around Axamer Lizum — comparing snow index (NDSI), vegetation index (NDVI), and water index (NDWI) year over year, alongside raw satellite and infrared bands.

NDSI, NDVI, NDWI satellite comparison 2019-2023

NDSI / NDVI / NDWI comparison, 2019–2023 — snow, vegetation, and water indices over the same 50×50 km region.

Satellite and infrared imagery comparison 2019-2023

Raw satellite and infrared imagery, 2019–2023 — the visible decline in snow cover across five winters.

Rings of activity life in the mountain, zoned diagram

Rings of activity life in the mountain — mapping top mountain, ski zones, valley, and settlement as concentric layers of human use.

Conventional ski infrastructure increasingly excavates the mountain — leaving permanent, devastating scars on the alpine ecosystem.

Concept

A Dual-Layered, Zero-Surface-Impact Architecture

In response, our project proposes a dual-layered, zero-surface-impact architecture — one strategy for when there is snow, and another for when there isn't.

Layer 01 — Winter

Generative Snow Structures

The first layer addresses the winter months: by analyzing the microscopic formation of snowflakes, we trained custom AI models and generative algorithms to design ephemeral surface structures. When it snows, the falling snow naturally builds these temporary, sculptural forms, which melt away entirely in the spring without a trace.

Generative structure variations, voxel growth study

Generative structure viewed from multiple orientations

Generative growth variations (left) and the same structure read from top, north, west, east, and south (right) — testing how each form behaves under directional snow load.

Extended set of generative voxel structure variations, eight isometric studies

A broader set of generative outputs from the same algorithm — each variation grown from a different parameter set, testing density, porosity, and structural rhythm.

Layer 02 — Snowless Periods

An Underground Network

The second layer addresses the increasingly frequent snowless periods. Recognizing that the mountain surface needs to be protected, our design shifts the permanent human footprint underground.

A network of subterranean tunnels provides a hidden infrastructure where visitors can still navigate and experience the mountain when there is no snow. By moving activity below the surface, the alpine landscape above is left untouched, allowing the fragile ecosystem to recover while still sustaining the region's vital tourism.

Process

From Archive to Algorithm

01 Collection and reading of family ski archives to trace the social history of Axamer Lizum.
02 Demographic and historical analysis revealing the continuous surge in international tourism.
03 Five-year satellite climate mapping (NDSI, NDVI, NDWI, infrared) exposing the decline in snow cover.
04 Mapping the "rings of activity" — how human use is layered across the mountain, from peak to valley.
05 Microscopic analysis of snowflake formation, translated into training data for custom AI models.
06 Generation and testing of ephemeral structure variations across orientations and snow-load scenarios.
07 Design of a complementary underground tunnel network for snowless periods — zero impact on the surface.

Alternative layout study, generative snow architecture, Axamer Lizum

Created with AI.