SLIDE: OUTPUT CHECKING – NDVI (Future Tense)

Presentation Script:

“After generating the cloud-free Sentinel-2 composite, output checking will be conducted to verify its
spectral accuracy.

For vegetation assessment, the Normalized Difference Vegetation Index or NDVI will be generated using
QGIS. NDVI will utilize the Near-Infrared band, which is Band 8, and the Red band, which is Band 4.

High NDVI values will indicate dense and healthy vegetation, while low or negative values will
correspond to bare soil, built-up areas, or non-vegetated surfaces.

The NDVI results will allow visual and quantitative confirmation that vegetation signals in the composite
are properly preserved.”

SLIDE: OUTPUT CHECKING – NDWI (Future Tense)

Presentation Script:

“In addition to NDVI, the Normalized Difference Water Index or NDWI will be computed to identify water
bodies within the area of interest.

NDWI will be calculated using the Green band, Band 3, and the Near-Infrared band, Band 8.

Positive NDWI values will indicate the presence of water, while negative values will represent vegetation
or dry land areas.

The NDWI output will help confirm that the cloud-free composite imagery retains correct spectral
behavior after cloud and shadow masking.”

SLIDE: RESULTS AND DISCUSSION – CLOUD REMOVAL & COMPOSITING

Presentation Script:

“Based on the results, Google Earth Engine effectively filtered Sentinel-2 images within the defined area
of interest and removed cloud contamination.

The application of the median composite technique helped normalize radiometric differences across
multiple scenes.

Visual inspection confirmed that major clouds and cloud shadows present in the original images were
successfully removed.
As a result, the final output is a high-quality composite that is suitable for subsequent spatial analysis
and modeling tasks.”

SLIDE: RESULTS AND DISCUSSION – NDVI INTERPRETATION

Presentation Script:

“The NDVI results further support the reliability of the composite image.

Areas with high NDVI values correspond to dense vegetation, while low NDVI values represent built-up
areas, exposed surfaces, or bare soil.

This behavior aligns with the expected characteristics of NDVI, validating that the spectral information in
the composite remains accurate and consistent.”

SLIDE: RESULTS AND DISCUSSION – NDWI INTERPRETATION

Presentation Script:

“The NDWI results were used to assess the presence and spatial distribution of water bodies.

Water features were clearly highlighted within the area of interest, confirming that the composite
imagery preserved natural water reflectance properties.

This further demonstrates that the cloud-free processing did not distort important spectral signals.”

SLIDE: RESULTS AND DISCUSSION – APPLICATIONS & ADVANTAGES

Presentation Script:

“The cloud-free composite provides continuous and gap-free spatial coverage across the entire area of
interest.

This directly addresses cloud-related challenges that are common in tropical regions.

The output is useful for various operations such as land-cover interpretation, image segmentation, and
feature extraction.

High-quality input data is essential for deep learning architectures, and cloud-free composites
significantly enhance model reliability by reducing noise.”
SLIDE: RESULTS AND DISCUSSION – LIMITATIONS

Presentation Script:

“However, there are limitations to the median composite approach.

While it smooths temporal information across multiple images, it is not suitable for applications that
require temporal monitoring, seasonal analysis, phenological studies, or change detection.

This is because temporal dynamics are lost due to the averaging process.”

SLIDE: CONCLUSION

Presentation Script:

“In conclusion, the cloud-free Sentinel-2 composite generated using Google Earth Engine produced a
reliable and high-quality dataset suitable for spatial analysis, land-cover mapping, and crop classification.

The visualization results, along with NDVI and NDWI analysis in QGIS, confirmed that vegetation, water,
and other land-cover features were accurately represented, indicating that the imagery retained its
natural spectral characteristics.

Although the composite is not suitable for studies requiring seasonal or time-based analysis, it provides
a practical and effective solution for overcoming cloud interference in optical satellite imagery.”

Output checking

The quality of the generated cloud-free composite will be evaluated through spectral index analysis to
ensure that the imagery retains accurate spectral and physical characteristics after processing.
Vegetation and water indices will be computed using QGIS to validate whether the cloud masking and
compositing procedures preserve the expected reflectance behavior of major land cover types within the
area of interest.

NDVI
Vegetation conditions will be assessed using the Normalized Difference Vegetation Index (NDVI), which
will be derived from the red and near-infrared bands of the composite image. Higher NDVI values are
expected to correspond to areas with dense and healthy vegetation, while lower or negative values will
indicate non-vegetated surfaces such as built-up areas or bare soil. The resulting NDVI distribution will
be analyzed to verify the spectral consistency and reliability of the composite imagery.

NDWI

Water features will be identified using the Normalized Difference Water Index (NDWI), which will be
calculated using the green and near-infrared bands. Positive NDWI values will be used to delineate water
bodies, whereas negative values will represent vegetation and dry land surfaces. The spatial pattern of
NDWI values will be examined to confirm that water-related spectral properties are accurately retained
in the cloud-free composite.

Overall, the results of the NDVI and NDWI analyses will serve as validation measures to ensure that the
processed imagery is suitable for subsequent spatial analysis, land cover interpretation, and agricultural
applications.

R&D 1

The application of Google Earth Engine enabled efficient filtering of satellite scenes and effective
removal of cloud and shadow contamination within the defined area. By integrating automated masking
algorithms with a median compositing approach, radiometric inconsistencies among multiple scenes
were minimized. Visual inspection confirmed that cloud artifacts present in individual images were
largely eliminated, resulting in a seamless and analysis-ready composite.

The generated indices further supported the reliability of the composite imagery. NDVI outputs reflected
realistic vegetation distribution patterns, while NDWI results accurately represented surface water
features. These outcomes demonstrate that the cloud-free composite retained natural spectral
characteristics, making it suitable for various spatial analysis tasks.

The continuous and gap-free coverage produced by the composite is particularly valuable in tropical
regions where persistent cloud cover limits the usability of optical satellite data. Such high-quality inputs
are essential for applications including land cover interpretation, image segmentation, and feature
extraction. Moreover, cloud-free imagery enhances the robustness of machine learning and deep
learning models by reducing noise and uncertainty in training data.

However, while the median compositing method improves spatial completeness, it also introduces
limitations. Temporal information is averaged across multiple acquisition dates, making the output
unsuitable for applications that require time-series analysis, seasonal monitoring, phenological
assessment, or change detection. As a result, the composite is best suited for static mapping and
baseline spatial analysis rather than dynamic temporal studies.