BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.earthmonitor.org//A98WPB
BEGIN:VTIMEZONE
TZID:Europe/Amsterdam
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-global-workshop-2026-A98WPB@pretalx.earthmonitor.org
DTSTART;TZID=Europe/Amsterdam:20261008T190000
DTEND;TZID=Europe/Amsterdam:20261008T191500
DESCRIPTION:Time series and spatial modeling are commonly used to generate 
 cloud- and gap-free satellite imagery. Most existing approaches reconstruc
 t the entire dataset using advanced models\, which requires high computati
 onal resources and time. In this study\, we introduce a new\, computationa
 lly efficient pipeline to reconstruct monthly Landsat data without gaps or
  clouds. The pipeline includes four levels of gap filling. In the first st
 ep\, we apply a clean mask to biweekly Landsat data and create a 7-image w
 eighted window spanning the current and neighbouring months. For each band
  and month across the 28-year period\, we generate 25th and 75th percentil
 e thresholds and calculate a weighted median\, giving 50% weight to the cu
 rrent month and 25% to neighboring months\, using only values within the 2
 5th–75th percentiles. In the second step\, remaining gaps are filled usi
 ng an annual land cover classification derived from the GLAD dataset and L
 andsat data from up to ten previous years\, restricted to pixels in the sa
 me land cover class. The third step fills small gaps of up to 2×2 pixels 
 using a 4×4 averaging kernel. These steps fill approximately 40–60% of 
 land pixels depending on tile location. Finally\, a pretrained temporal mo
 del is applied to fill the remaining gaps. We tested this pipeline on a CP
 U server with 96 threads and 1 TB RAM. Each tile can be processed in under
  2000 seconds. Parallelization across tiles and bands enables global proce
 ssing in under six weeks\, significantly reducing the computational time c
 ompared to full dataset reconstruction\, which would take approximately si
 x months. The resulting dataset provides clean\, gap- and cloud-free month
 ly Landsat imagery suitable for a variety of research applications. Limita
 tions remain\, mostly related to input/output operations\, and future work
  could apply embedding models to reduce dataset size and produce abstract 
 representations for faster access.
DTSTAMP:20260624T084419Z
LOCATION:Room 18
SUMMARY:A Multi-Layer Gap-Filling Pipeline for Continuous Monthly Landsat D
 ata (1997–2025) - Sajed
URL:https://pretalx.earthmonitor.org/global-workshop-2026/talk/A98WPB/
END:VEVENT
END:VCALENDAR