Ensemble – Vector Explorer
Explore Eclipse summary data and score ensemble realizations directly from local SMSPEC/UNSMRY files. Ensemble data stays on the workstation.
Summary vectors
Choose a type, then a parameter and the relevant region number, group or well. These choices use metadata only.
Browse all vectors · advanced
| Vector | Keyword | Name | Number | Unit |
|---|
Selected vector
Ensemble analysis
Vector
Show data preview
| Index | Time | Value |
|---|
Vector
Histogram
Show realization statistics
| Realization | Timesteps | End date | Completeness | Status | Final value |
|---|
History-match QC
Select entities and QC prefixes that have matching history vectors (for example WOPR + WOPRH). The score logic mirrors the uploaded notebook: normalized RMSE → percentile score → category score → weighted total score.
Regional choices show summary family + number: FIPNUM is the standard set; FIP family REG, PRE, etc. are reported family tags. A tag may abbreviate a longer DATA name. Unknown forms keep the raw keyword and are not assigned to a FIP array.
Entities
Only entities with matching simulation/history pairs are listed. Select entities first; their available QC prefixes appear alongside. Selected items stay visible as chips.
QC prefixes
Only prefixes with pairs for the selected entities are offered. Counts refer to this entity selection. Previously selected prefixes without pairs stay marked until you remove them.
Weights for total score
Sliders set relative priorities; percentages show normalised shares. Preview only — run QC to apply.
Other: · fixed zero priority; not changed by Equal priorities.
Preview · select QC vectors
Relative priority values (0–1)
Percentages are rounded. Bar segments use the unrounded shares.
Acceptance criteria
Every selected QC vector must have a valid misfit ≤ the limit. Weights affect ranking only; zero priority does not skip acceptance testing.
Weights used in last completed QC
Total score by realization
Vector and category scores
Colour shows relative score. × marks a vector that fails the misfit limit; ? means unavailable. Every selected vector must pass for acceptance.
Colour scale: Relative score
QC vector
History position within ensemble
| Key | Category | History key | Coverage inside P10–P90 | Avg. history percentile |
|---|
Representative forecast cases
Select a forecast vector for the accepted QC realizations. Use a single-value summary or compare entire curves. Scalar mode identifies QC Reference, Low / P90, Base / P50, High / P10 and an Alternative. QC Reference is the highest-scoring QC case eligible for the selected forecast; it may differ from the overall QC leader. Curve mode finds the closest actual realization to each percentile curve; it does not rank final volumes.
Forecast entity includes Field, region family + number, Group and Well when present in SMSPEC. FIPNUM is the standard set; other families show their summary tag, for example FIP family REG. Tags may abbreviate longer DATA names. Unresolved region families are temporarily hidden from forecast choices. Choose Region, then search for a family, region number and vector. Forecast vectors do not need matching history; the accepted QC population stays unchanged.
Forecast eligibility
| Realization | QC score | Last valid date / time | Expected end | Forecast status | Reason |
|---|
Smart shortlist
| Case | Realization | QC score | Forecast value | Percentile target |
|---|
Representative cases
Explore selected cases across other vectors
FOIP + First can show initial oil in place. First/Last use the first/last finite report in the entire loaded series, including history; zero is valid. Mean gives each report equal weight. Sum is not a time integral. This comparison does not reselect cases or change QC.
Grey: whole ensemble with finite values. Green: QC-accepted cases with finite values. Coloured markers: existing shortlist cases, positioned on the whole-ensemble distribution. Several case labels may share the same REAL and marker position. These ranks are not probabilities of success or new P90/P50/P10 case labels.
| Shortlist case | Realization | Comparison value | Whole ensemble rank (≤) | QC accepted rank (≤) | First/Last sample date | Comparison status |
|---|
Unavailable comparison values
| Realization | Reason |
|---|
Forecast versus comparison value
X comes from the completed distribution above. Y uses the original shortlist forecast vector with the aggregation chosen here. Curve-based shortlists default to Last for Y; curve RMSE is not used as an axis value. This plot does not change the CDF, QC or shortlist.
Grey: other ensemble cases. Green: QC-accepted cases. Coloured overlays: existing shortlist cases. Background groups do not double-count a REAL; repeated shortlist roles may overlap. Hover for REAL, values, report dates and QC status. The axes show physical values, not cumulative percentages.
| Shortlist case | Realization | X value | Y value | X First/Last date | Y First/Last date | Paired status |
|---|
Unavailable X/Y pairs
| Realization | Reason |
|---|
Ranked score table
Realizations
Show realization details
| Realization | Status | Case | SMSPEC | UNSMRY |
|---|