NaqsheraAPPLIED AI SYSTEMS

TIME-SERIES RESEARCH / EUV FORECASTING

Turn the recent solar past into a 60-day view ahead.

This applied research explores how a model can use the most recent 30 or 60 days of solar activity and EUV observations to estimate what the following 60 days may look like.

RECENT OBSERVATIONS30–60 DAY CONTEXT60-DAY OUTLOOK
Discuss a forecasting workflow ↗
SELECTED RESEARCH OUTPUTDAY 60
A research chart comparing observed EUV irradiance with several model forecasts over a 60-day horizon.
Selected 60-day comparison from the research output. Model comparisons and values are specific to this experiment and are shown as evidence of the forecasting workflow—not as a universal operational forecast.

Forecasting is not a single number. It is a structured view of uncertainty over time.

TIME SERIESMULTI-HORIZONDATA-SPECIFIC EVALUATION

THE RESEARCH QUESTION

How much of the recent past is useful for the next 60 days?

Rather than relying on a fixed, one-size-fits-all lookback, the study compares model setups that see different recent observation windows before they generate the same forward horizon.

01

Recent solar activity

The workflow starts with a recent sequence of observed solar behaviour, preserving the order and changing shape of the signal.

SEQUENTIAL CONTEXT
02

EUV conditions

EUV values are treated as a related time-varying signal, allowing the research to examine how the series move across a shared horizon.

RELATED SIGNALS
03

A forward window

The output is a 60-day estimated sequence, not a promise of a single exact future value on one selected day.

FORWARD OUTLOOK

WHAT THE MODEL IS ASKED TO LEARN

A moving window, a future sequence and an honest measure of fit.

The research evaluates forecast behaviour against separately held evaluation data. That separation matters: it helps distinguish a model that recognises a pattern from one that only remembers its training examples.

INPUTRecent observed window30 OR 60 DAYS
TRAININGHistorical sequences available to the modelLEARNING SET
EVALUATIONSeparate, unseen sequence for comparisonHELD OUT
OUTPUTForward 60-day estimated trajectoryRESEARCH OUTPUT
WHAT THIS SHOWS

A disciplined forecasting research contribution.

The work covers model runs, experimental analysis, error analysis and technical reporting for a time-series research effort.

WHAT IT DOES NOT CLAIM

Not a universal solar forecast.

Any result belongs to its specific experiment, data period and evaluation design. The academic project’s concept, direction and ownership remain with its supervising lead.

START WITH THE DECISION HORIZON

Need a more useful view of what comes next?

Talk to Naqshera ↗