Live demo · illustrative lender portfolio

Climate & nature screen, asset by asset.

Physical quantities from open datasets — never a composite score. Pick a hazard, flip between today and 2050, and open any asset. The disclosure extract below rolls it all into what a bank reports. Or just ask.

Ask the portfolio · AI agent

Open-weights model · free tier · every figure from code · Model card & evals →

Try an example:

16
assets · 7 countries
1
> 0.5 m flood at 1-in-100, 2050
16
more hot days by 2050
9
within 10 km of protected nature
12
with a value the data can’t resolve

Today → 2050, heat: 12 of 16 assets with data move to a more severe band.

Days a year above 30 °C mean, 2050 (model median)

< 10
10–30
30–90
90–180
> 180
No data / not resolved

Thailand · industrial

Industrial estate, Ayutthaya

Industrial estates here were under water for weeks in the 2011 Chao Phraya floods

14.350, 100.580

At a glance · 2050

River flood · 2050
0.72 m
Coastal flood · 2050
0.00 m
Heat · 2050
252 days
Water stress · 2050
Extremely high

River flood

1-in-100-year depth 0.20 m today → 0.72 m in 2050 (RCP8.5)

Range across 5 climate models: 0.60–0.91 m

Undefended depth in the asset's ~1 km cell; flood defences are not modelled. Future is the median of 5 CMIP5 GCMs (min and max given).

WRI Aqueduct Floods v2 (via OS-Climate physrisk) · CC BY 4.0 (WRI open data default)

Coastal flood

1-in-100-year depth 0.00 m today → 0.00 m in 2050 (RCP8.5)

Without land subsidence: 0.00 m

No modelled coastal inundation in this cell or within about 2 km at any return period.

WRI Aqueduct Floods v2 (via OS-Climate physrisk) · CC BY 4.0 (WRI open data default)

Heat

Days a year above 30 °C daily mean: 108 (92–115) today → 252 (220–303) in 2050 (SSP5-8.5). Median and range across 6 climate models.

Per-model values

Model	Today	2050
ACCESS-CM2	112	303
CMCC-ESM2	115	228
CNRM-CM6-1	104	287
MIROC6	92	227
MPI-ESM1-2-LR	113	220
NorESM2-MM	102	276

Point value from the 0.25° (~28 km) cell containing the asset; median, min and max across the CMIP6 GCMs available.

OS-Climate chronic heat indicators v2: days above a mean-temperature threshold · NEX-GDDP-CMIP6 is CC0 1.0

Water stress

Extremely high (0.81) in 1999 → Extremely high (1.15) in 2050 (SSP5-8.5)

Sub-basin value read from the ~9 km cell containing the asset.

WRI Aqueduct 4.0 baseline and future water stress (via OS-Climate physrisk) · CC BY 4.0

Vegetation · Sentinel-2

Median NDVI 0.24 within 500 m, from 5 cloud-masked scenes.

Median over 7406 land pixels (10 m) in the 500 m buffer; per-pixel median of 5 scenes on tile MGRS-47PPR, 2026-01-05 to 2026-04-05.

Sentinel-2 Collection 1 Level-2A surface reflectance (Earth Search) · Copernicus Sentinel data legal notice (free, full and open access)

Protected nature · TNFD Locate

Nearest: Wat Tan En Non-Hunting Area, 18.7 km away.

Nearest nature protected area in OSM; counted as nature on protect_class=4. Excluded as non-nature: 1 OSM area(s) in range.

OpenStreetMap nature protected areas (Overpass API) · Open Database Licence (ODbL) 1.0

Caveats for this asset

Riverine flood: the five GCMs disagree on 2050 RP100 depth (0.60 to 0.91 m).
Protected area: Historic City of Ayutthaya at 1.3 km is excluded as not a nature area: social or cultural protection (protect_class=22).

The portfolio

Sixteen assets, four hazards, two periods.

Dots use the map’s colour bands for 2050. Hollow ring: lowest band. Grey: no data or not resolvable at this resolution.

Download all results (Excel)

Portfolio assets with hazard bands for the selected period
Country	River flood	Coastal flood	Heat	Water stress
Singapore	—	0.00 m	81 days	Low
Singapore	—	0.00 m	80 days	Low
Indonesia	0.11 m	—	128 days	High
Indonesia	0.00 m	0.00 m	60 days	Low
Indonesia	0.00 m	0.00 m	43 days	Low
Indonesia	—	—	59 days	Extremely high
Malaysia	—	0.21 m	56 days	Low-medium
Malaysia	—	0.00 m	7 days	Low
Thailand	—	0.00 m	267 days	Extremely high
Thailand	0.72 m	0.00 m	252 days	Extremely high
Vietnam	—	0.18 m	179 days	Extremely high
Vietnam	—	—	188 days	Low-medium
Philippines	—	0.00 m	80 days	High
Australia	0.21 m	0.00 m	2 days	High
Australia	—	0.00 m	1 days	Low-medium
Australia	—	0.00 m	2 days	Extremely high

From screening to disclosure

What the bank actually publishes.

Banks and insurers in Singapore and Australia now disclose under ISSB-based rules (SGX/ACRA, AASB S2), with MAS and APRA expecting the same analysis inside risk management — and TNFD for nature. The asset results above roll up into those lines, with every definition written down, because an auditor will ask.

Climate- and nature-related disclosures · draft extract

Illustrative

Metrics — physical risk (IFRS S2 ¶29(c))

As at the reporting date, SGD 255 million (27%) of the illustrative lending portfolio of SGD 945 million was assessed as vulnerable to climate-related physical risks by 2050 under a high-emissions scenario (RCP8.5 for flood; SSP5-8.5 for heat). An asset is vulnerable where its modelled 1-in-100-year flood depth exceeds 0.5 m (1 asset) or where it experiences 90 or more days a year with a daily mean temperature above 30 °C (5 assets).

Vulnerable exposure is concentrated in Thailand and Vietnam, driven mainly by extreme heat; river and coastal flood exposure is assessed with undefended, screening-grade layers and will be refined with licensed models before use in credit decisions.

Nature. 9 of 16 assets (SGD 595 million) lie within 10 km of a nature-protected area and are prioritised for the TNFD LEAP Evaluate phase. 8 assets (SGD 440 million) sit in sub-basins with high or extremely high baseline water stress.

Metric	SGD m	% of book
Amount and percentage of assets vulnerable to climate physical risksIFRS S2 ¶29(c) · AASB S2 · SGX	255	27%
Interface with ecologically sensitive locationsTNFD LEAP L4 · MAS ENRM	595	63%
Activities in areas of water scarcityTNFD C3.0 · IFRS S2 physical risk	440	47%

Generated from portfolio.json by a fixed template — no language model writes these numbers. Exposure amounts are invented for illustration; no real borrowers. Paragraph references: IFRS S2 (2023, amended Dec 2025); TNFD Recommendations (Oct 2026).

How each line is defined

Thresholds are a stated policy for the demo; a real institution sets and defends its own.

IFRS S2 ¶29(c) · AASB S2 · SGX
Computed
Amount and percentage of assets vulnerable to climate physical risks
1-in-100 flood depth > 0.5 m (river or coastal) or ≥ 90 hot days a year, in 2050 under a high-emissions scenario
Production weights by actual exposure and collateral, applies damage curves, and runs at least two scenarios (one orderly, one high-emissions).
IFRS S2 ¶22 (climate resilience)
Partly computed
Scenario analysis of physical risk
Baseline versus 2050 under RCP8.5 (flood) and SSP5-8.5 (heat, water)
Add an orderly / Paris-aligned scenario, more horizons, and NGFS transition pathways for the transition-risk side.
TNFD LEAP L4 · MAS ENRM
Partly computed
Interface with ecologically sensitive locations
Within 10 km of a mapped nature-protected area (OpenStreetMap)
Production uses licensed IBAT layers (WDPA, Key Biodiversity Areas) and ecosystem integrity, not OpenStreetMap alone.
TNFD C3.0 · IFRS S2 physical risk
Computed
Activities in areas of water scarcity
WRI baseline water stress “High” or “Extremely high”
C3.0 also needs the borrower’s own water withdrawal and consumption volumes.
IFRS S2 ¶29(a) · PCAF
Not in the demo yet
Financed emissions (Scope 3 Category 15)
Borrower emissions × attribution factor, with PCAF data-quality score
Needs borrower financials and emissions data; next on the roadmap (PCAF engine with data-quality scoring).