WRI Aqueduct Water Risk Skill

Queries the World Resources Institute (WRI) Aqueduct 4.0 water risk dataset via ArcGIS REST FeatureServer. Returns water stress, drought, flood, and groundwater risk indicators for any country, state/province, or watershed.

No API key required. WRI Aqueduct data is publicly accessible through ArcGIS REST services.

API Structure

Base URL:

https://services.arcgis.com/LG9Yn2oFqZi5PnO5/arcgis/rest/services/

Primary Service — Aqueduct 4.0 Annual:

https://services.arcgis.com/LG9Yn2oFqZi5PnO5/arcgis/rest/services/Aqueduct_Annual/FeatureServer/0/query

This is an ArcGIS FeatureServer endpoint. Use POST requests with form-encoded parameters for queries with complex WHERE clauses containing apostrophes.

Query Parameters

Parameter	Description	Example
`where`	SQL WHERE clause using field names	`country_un='United States of America' AND name_1='West Virginia'`
`outFields`	Comma-separated field names to return	`name_0,name_1,bws_cat,bws_label`
`f`	Response format	`json`
`resultRecordCount`	Max features to return (server max: 1000)	`200`
`returnGeometry`	Include geometry in response	`false` (set false for data-only queries)
`orderByFields`	Sort field	`bws_cat DESC`
`resultOffset`	Pagination offset	`0`, `200`, `400`

Key Field Names — Aqueduct 4.0

Location Fields

Field	Description	Example Values
`name_0`	Country name	United States of America
`name_1`	Province/state name	West Virginia, Texas
`name_2`	Sub-province/county name (where available)	Monongalia County
`aq_name`	Aqueduct watershed name	Upper Ohio, Cheat River
`pfaf_id`	Pfafstetter watershed code	72100000

Water Risk Indicators

Field	Full Name	Scale	Interpretation
`bws_cat`	Baseline Water Stress (category)	0-4, -1	Primary scarcity indicator
`bws_label`	Baseline Water Stress (label)	text	Low/Low-Med/Med-High/High/Ext High
`bws_score`	Baseline Water Stress (score)	0-5 continuous	Higher = more stressed
`bwd_cat`	Baseline Water Depletion (category)	0-4, -1	Groundwater + surface depletion
`bwd_label`	Baseline Water Depletion (label)	text
`iav_cat`	Interannual Variability (category)	0-4, -1	Year-to-year flow fluctuation
`iav_label`	Interannual Variability (label)	text
`sev_cat`	Seasonal Variability (category)	0-4, -1	Within-year flow fluctuation
`sev_label`	Seasonal Variability (label)	text
`drr_cat`	Drought Risk (category)	0-4, -1	PDSI-based drought frequency and severity
`drr_label`	Drought Risk (label)	text
`rfr_cat`	Riverine Flood Risk (category)	0-4, -1	Flood inundation return period
`rfr_label`	Riverine Flood Risk (label)	text
`gtd_cat`	Groundwater Table Decline (category)	0-4, -1	Aquifer depletion trend (GRACE)
`gtd_label`	Groundwater Table Decline (label)	text
`cep_cat`	Coastal Eutrophication Potential	0-4, -1	Nutrient loading risk
`ucw_cat`	Untreated Connected Wastewater	0-4, -1	Wastewater treatment coverage

Risk Category Scale (All Indicators)

Category	Label	Percentile	Meaning
-1	Arid and Low Water Use	—	Arid region; water withdrawal so low relative to availability that stress ratio is undefined; does NOT mean safe
0	Low	< 10th	Low risk
1	Low-Medium	10-37th	Moderate risk
2	Medium-High	37-67th	Elevated risk
3	High	67-90th	High risk
4	Extremely High	> 90th	Highest risk globally

Example API Calls

Query West Virginia Water Risk Data

curl -s "https://services.arcgis.com/LG9Yn2oFqZi5PnO5/arcgis/rest/services/Aqueduct_Annual/FeatureServer/0/query" \
  --data-urlencode "where=country_un='United States of America' AND name_1='West Virginia'" \
  --data-urlencode "outFields=name_0,name_1,aq_name,bws_cat,bws_label,bwd_cat,bwd_label,drr_cat,drr_label,rfr_cat,rfr_label,gtd_cat,gtd_label,iav_cat,iav_label" \
  --data-urlencode "returnGeometry=false" \
  --data-urlencode "f=json" \
  --data-urlencode "resultRecordCount=50"

Query Texas (Permian Basin context), sorted by stress

curl -s "https://services.arcgis.com/LG9Yn2oFqZi5PnO5/arcgis/rest/services/Aqueduct_Annual/FeatureServer/0/query" \
  --data-urlencode "where=country_un='United States of America' AND name_1='Texas'" \
  --data-urlencode "outFields=name_0,name_1,aq_name,bws_cat,bws_label,bwd_cat,bwd_label,drr_cat,drr_label,gtd_cat,gtd_label" \
  --data-urlencode "returnGeometry=false" \
  --data-urlencode "orderByFields=bws_cat DESC" \
  --data-urlencode "f=json" \
  --data-urlencode "resultRecordCount=100"

Query Extremely High Stress Watersheds in U.S.

curl -s "https://services.arcgis.com/LG9Yn2oFqZi5PnO5/arcgis/rest/services/Aqueduct_Annual/FeatureServer/0/query" \
  --data-urlencode "where=country_un='United States of America' AND bws_cat=4" \
  --data-urlencode "outFields=name_0,name_1,aq_name,bws_cat,bws_label" \
  --data-urlencode "returnGeometry=false" \
  --data-urlencode "f=json" \
  --data-urlencode "resultRecordCount=200"

Parse JSON Response (bash + jq)

curl -s "..." | jq '[.features[].attributes |
  {name_1, aq_name, bws_label, drr_label, rfr_label, gtd_label}]'

Workflow

Step 1 — Identify Location

Map the user's location query to Aqueduct field values:

Country name: Use country_un with official Aqueduct spellings (see table in references/api_reference.md)
State/Province: Use name_1 exactly as stored
Watershed basin: Use aq_name if user specifies a basin name
County level: Not directly available — query at state level and identify watershed(s) covering the target county by name or geography

Key country name mappings:

User says	Aqueduct country_un
USA / US / United States	United States of America
UK / Britain	United Kingdom
Russia	Russian Federation
Iran	Iran (Islamic Republic of)
Venezuela	Venezuela (Bolivarian Republic of)

Step 2 — Query ArcGIS FeatureServer

Execute the query with appropriate WHERE clause. Start with state/province level. If the response has many features, sort by bws_cat DESC and report the highest-stress watersheds first. If the response is empty, check country name spelling — try without the state filter.

Retrieve at minimum: bws_cat, bws_label, drr_cat, drr_label, rfr_cat, rfr_label, gtd_cat, gtd_label, iav_cat for each watershed.

Step 3 — Aggregate Results

For a state query returning multiple watersheds:

Report the range (min to max category) for each indicator
Identify the specific high-stress watersheds by name
Note if results are heterogeneous (e.g., "western Texas watersheds are Extremely High stress while eastern Texas watersheds are Medium-High")

Step 4 — Apply Petroleum Operations Implications

Use the standard implications table below to interpret each risk category.

Step 5 — Produce Output

Format as a risk score table followed by narrative summary.

Output Format

Risk Score Table

## Water Risk Assessment — [Location]
Source: WRI Aqueduct 4.0 Annual Data

| Indicator | Category | Label | O&G Operations Implication |
|-----------|----------|-------|---------------------------|
| Baseline Water Stress | [0-4] | [label] | [see implications table] |
| Baseline Water Depletion | [0-4] | [label] | [see implications table] |
| Drought Risk | [0-4] | [label] | [see implications table] |
| Riverine Flood Risk | [0-4] | [label] | [see implications table] |
| Groundwater Decline | [0-4] | [label] | [see implications table] |
| Interannual Variability | [0-4] | [label] | [see implications table] |

Standard Implications by Indicator and Category

Baseline Water Stress (bws):

0-1 (Low/Low-Medium): Freshwater sourcing feasible; regulatory burden moderate; surface water permits available
2 (Medium-High): Monitor permitted withdrawals carefully; report water use; consider produced water reuse as partial alternative
3-4 (High/Extremely High): Surface and groundwater sourcing severely constrained; produced water reuse strongly preferred or required; water procurement plans needed before permit approval

Baseline Water Depletion (bwd):

0-1: Groundwater replenishment adequate; aquifer availability stable
2: Aquifer drawdown concerns; model seasonal and long-term impacts before new groundwater withdrawals
3-4: Aquifer at risk of permanent depletion; avoid new groundwater wells for completion fluid sourcing; subsidence risk

Drought Risk (drr):

0-1: Freshwater availability reliable; drought unlikely to interrupt operations
2: Plan for periodic shortages (1-in-5 to 1-in-10 year events); maintain water storage buffer
3-4: Chronic shortage risk; operations dependent on surface water should have contingency plans including produced water reuse or trucked-in water

Riverine Flood Risk (rfr):

0-1: Minimal flood hazard to surface facilities; standard drainage design adequate
2: Locate pits and tanks above 25-year flood stage; check FEMA flood maps for pad siting
3-4: High infrastructure exposure; pad siting, secondary containment design, and emergency response plans must explicitly account for 100-year flood inundation; saltwater containment failure during floods is a major regulatory liability

Groundwater Table Decline (gtd):

0-1: Aquifer stable; long-term groundwater availability reasonable
2: Declining trend; produced water disposal via UIC Class II wells preferred over new freshwater withdrawals
3-4: Active aquifer depletion (GRACE satellite data confirms); competing use conflicts likely; water disposal to saline zones may face increasing regulatory scrutiny

Interannual Variability (iav):

0-1: Consistent water availability year to year; minimal planning burden
2: Variability manageable with modest on-site storage or diversified supply
3-4: High year-to-year swings; water sourcing agreements and reuse infrastructure are critical for operational continuity

Narrative Summary Structure

Location context and which O&G producing formation this covers
Key finding — the highest-category indicator and its significance
Operations relevance — which specific operations are most affected
Produced water reuse value — higher stress = higher economic and regulatory value of produced water reuse over freshwater sourcing
Regulatory signal — high stress regions attract more water withdrawal scrutiny from state regulators
Comparison to major O&G basins for context

Example Output

## Water Risk Assessment — West Virginia (Appalachian Basin / Marcellus)
Source: WRI Aqueduct 4.0 Annual Data

| Indicator | Category | Label | O&G Implication |
|-----------|----------|-------|-----------------|
| Baseline Water Stress | 0-1 | Low to Low-Medium | Freshwater sourcing feasible; Ohio/Monongahela systems available |
| Baseline Water Depletion | 0 | Low | Aquifer replenishment adequate |
| Drought Risk | 1 | Low-Medium | Occasional dry periods; not chronic |
| Riverine Flood Risk | 2-3 | Medium-High to High | Pad siting and pit location need flood staging assessment |
| Groundwater Decline | 0 | Low | Appalachian aquifers not under extraction pressure |
| Interannual Variability | 1 | Low-Medium | Reasonably consistent year-to-year precipitation |

**Summary:** West Virginia sits in the water-abundant Appalachian region.
Baseline water stress is low compared to major U.S. O&G basins — the
Permian Basin in West Texas typically scores Extremely High, Bakken
in North Dakota typically Medium-High. The primary water risk for Marcellus
operations in WV is riverine flood exposure along the Monongahela and
Kanawha corridors, affecting pad siting and produced water pit design.
Freshwater sourcing for hydraulic fracturing is generally feasible with
appropriate WVDEP permits. Produced water reuse is economically driven by
disposal cost reduction rather than water scarcity here. [Certainty: MEDIUM —
Aqueduct data at watershed scale; site conditions may differ]

Error Handling

Condition	Action
ArcGIS returns 200 with empty features array	Country_un name mismatch most likely; try country-only WHERE clause; check alternate name spellings in references/api_reference.md
HTTP 400 Bad Request	Malformed WHERE clause; check for unescaped apostrophes (use doubled single quote: O''Brien), verify field name spelling, ensure strings are quoted
HTTP 499 or 500	ArcGIS service temporarily unavailable; retry after 30 seconds; suggest WRI Aqueduct web tool https://www.wri.org/aqueduct as fallback
More than 100 features returned	Summarize by range and modal category; identify top-5 highest-stress watersheds by aq_name
Category = -1 returned	Arid and Low Water Use classification; explain this does NOT mean low risk — it means so arid that stress ratio is undefined; note extreme absolute scarcity
User asks for county-level data	Aqueduct data is at HydroSHEDS watershed scale; identify overlapping watersheds by aq_name; note that county boundaries do not align with watershed boundaries

Regional Context for U.S. O&G Basins

Pre-loaded context to interpret results against major producing regions:

Basin	State(s)	Typical BWS Category	Typical GTD	Operations Context
Permian (Delaware/Midland)	TX, NM	4 (Extremely High)	3-4	Severe scarcity; PW reuse economically compelling and increasingly required
Bakken	ND, MT	2-3 (Med-High/High)	1-2	Semi-arid; seasonal constraints on surface water; moderate PW reuse activity
Marcellus	WV, PA	0-1 (Low/Low-Med)	0-1	Water-abundant; flood risk is primary concern not scarcity
Utica	OH, WV, PA	0-1 (Low/Low-Med)	0-1	Similar to Marcellus; Great Lakes basin protection regs relevant in OH
Eagle Ford	TX (South)	3 (High)	2-3	High stress; significant PW volumes support reuse economics
Haynesville	LA, TX	1-2 (Low-Med)	0-1	Lower stress; Gulf Coast flooding is primary surface risk
DJ Basin	CO, WY	2-3 (Med-High/High)	2	Front Range water competition; strict state water use regulations
Anadarko	OK	2 (Med-High)	1-2	Moderate; induced seismicity near disposal wells adds complexity

Caveats and Data Limitations

Watershed scale, not point scale. Aqueduct data represents hydrological watershed units (HydroSHEDS level 6, averaging ~50,000 km2). Site-specific conditions within a watershed can vary substantially.
Historical baseline. Aqueduct 4.0 uses long-term averages (1979-2019). It does not reflect real-time drought. For current conditions, use USGS WaterWatch or the U.S. Drought Monitor.
Demand-side uncertainty. Baseline Water Stress is calculated as total withdrawals divided by total renewable supply. Agricultural withdrawal data (USGS, FAO) has significant uncertainty and may understate or overstate actual consumption in specific watersheds.
GRACE groundwater resolution. Groundwater Table Decline uses GRACE satellite gravity anomaly data at ~300 km resolution — lower spatial detail than other indicators.
Produced water not separately accounted. Aqueduct does not model produced water reuse, which can significantly offset freshwater demand in active O&G basins. Stress scores therefore overstate true freshwater demand where reuse programs are active.
Quality vs. quantity. Aqueduct covers water quantity risk. For water quality context near O&G operations, use pnge:usgs-waterdata (streamflow and water quality gauges) and pnge:epa-enviro (NPDES, UIC permits).

Implementation Notes

Use curl with POST form encoding (--data-urlencode) for queries with apostrophes in country/state names — GET URLs will fail on apostrophes
Parse response: response.features[].attributes contains the field data
ArcGIS server max recordCount is typically 1000; paginate with resultOffset if needed for national-scale queries
Service metadata (layer definition, field aliases): append ?f=json to the FeatureServer/0 URL (without /query)
See references/api_reference.md for all available Aqueduct service layers (monthly, food-water-energy, flood module), complete field list, and all country name spellings used by the Aqueduct database

wri-aqueduct