Catchment land cover is one of the important factors that influence the conversion of rainfall into runoff. It provides information on the infiltrability of the soil and the ability of the catchment to store and release rainwater. In the Philippines, particularly in Cebu Island, flooding incidences have been experienced at present for rainfall events of magnitude that happened in the past but did not create flooding then. This phenomenon has a connection to changes in land cover especially from vegetation to pavement. This research utilizes satellite images from Landsat-5, Landsat-7, and Landsat-8 in generating historical land cover maps from 1993-2019. The study area was the Butuanon river catchment located in Central Cebu, which has been significantly altered by urbanization in the past two decades. Mosaiced images were made from the said satellite imageries to produce a clear image of the catchment. To achieve this, pansharpening and atmospheric correction were applied to improve image quality, while clouds which came with the images were removed with the QGIS Cloud Masking Plugin. The resulting voids from the masking process were filled in with the temporally closest masked images. Land cover classification was performed with eCognition Developer 9.0, with the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Built-up Index (NDBI) as the primary parameters used. Maps were then processed and created using ArcMap 10.3.1, and the percentages for bare soil, vegetated, and developed areas were determined. Five complete images were formed, composing of data from 1993-1995, 2001-2002, 2007-2011, 2014-2016, and 2019, respectively. It was found that for the duration specified, the percent land developed rose over time, with the biggest increase (8.062%) happening earlier this decade. On the other hand, the vegetated areas had a generally decreasing trend, with the exception of the earliest years, where the opposite occurred. Lastly, bare soil areas had no specific pattern to its changes. Manual digitization was then performed on a 1-square km sample area to validate the results of the classification from eCognition. It is recommended that the land classification procedures used be improved to include more land cover types. This study did not deal with tree cover, but doing so would make the results much more accurate.

1.
Barsi
,
J.A.
, “
The Spectral Response of the Landsat-8 Operational Land Imager
,”
Remote Sens.
6
,
10232
10251
(
2014
).
2.
X.C.
Llano
,
QGIS Plugin: CloudMasking (Version 18.11.15.Q3), SMByC-IDEAM and FAO, 2018, see
https://smbyc.bitbucket.io/qgisplugins/cloudmasking
3.
Phiri
,
D.
, “
Effects of pre-processing method on Landsat OLI-8 land cover classification using OBIA and random forecast classifier
,”
Int J Appl earth Obs Geoinformation
73
,
170
178
(
2018
).
4.
Trimble
Geospatial
,
eCognition
9
(
2019
), available at http://www.ecognition.com/suite/new/whats-new-version-9
5.
USGS
,
Landsat: Landsat Collection 1 Level 1 - Landsat 8 OLI/TIRS, Landsat 7 ETM+ & Landsat 4-5 TM (1993-2016)
, available at https://earthexplorer.usgs.gov
6.
USGS
,
What are the best Landsat spectral bands for use in my research?
(
2019
), available at https://www.usgs.gov/faqs/what-are-best-landsat-spectral-bands-use-my-research?qt-news_science_products=0#qt-news_science_products
7.
R.E.
Crippen
, “
Calculating the vegetation index faster
,”
Remote Sensing of Enviromnment
34
,
71
73
(
1990
)
This content is only available via PDF.