date
October 2026
title
Tarlac Province Historical Flood Hotspot Analysis
version
2.2 (GAUL boundary fix and 60 m Z-score analysis)
file name
Tarlac_Province_Flood_Hotspots_60m_2015_2026
rationale
The analysis uses Sentinel-1 imagery across defined flood-event windows. It counts how often flooding is detected, then applies a Getis-Ord Z-score to identify statistically significant flood hotspots and separate them from places with only occasional or random inundation. The raster reports flood counts, observation counts, detection rates, and hotspot intensity for disaster risk reduction and land-use planning.
resolution
60 m
description
Historical flood hotspot analysis for Tarlac Province, Philippines, using Sentinel-1 SAR. It maps areas that flood repeatedly and groups them by hotspot intensity.
inputs used
Copernicus Sentinel-1 SAR, Ground Range Detected (COPERNICUS/S1_GRD)SRTM digital elevation model, used for slope (USGS/SRTMGL1_003)JRC Global Surface Water, optional permanent-water mask (JRC/GSW1_4/GlobalSurfaceWater)Rice paddy map, optional (projects/ee-brybaltazar/assets/SEA_Phil_Rice)FAO GAUL 2015 level 2 administrative boundaries for Tarlac Province (FAO/GAUL/2015/level2)
methodology
Data source. Copernicus Sentinel-1 Ground Range Detected imagery, VH polarization, Interferometric Wide (IW) mode, with an optional orbit-pass filter.Preprocessing. Each image is smoothed with a focal median filter to reduce speckle, then compared with a median pre-event image from the lookback period.Flood detection. A pixel is marked flooded when backscatter drops sharply from pre-event conditions and falls below the water threshold during the event window. Isolated patches below the minimum connected-pixel count are removed.Masking. Steep slopes are excluded. When enabled, permanent water is excluded where JRC Global Surface Water seasonality is 10 months or more. When enabled, areas outside mapped rice paddies are excluded.Event aggregation. Each pixel stores the number of event windows with detected flood (flood window count) and the number of windows with a valid Sentinel-1 observation (observed window count).Hotspot analysis. A Getis-Ord Z-score is calculated on the flood window count to find clusters of unusually high or low flood frequency. Classes are Low, Moderate, High, and Very High, based on standard deviations from the mean.Priority hotspots. Areas with a Z-score above 1.65 (High and Very High) are kept when they also meet the minimum valid-observation count and connected-pixel size.
raster bands
flood_window_count: detected flood events at the pixel, from 0 to 11.observed_window_count: valid Sentinel-1 observations at the pixel across all event windows.detected_pct: percentage of observed windows in which flood was detected.z_score: Getis-Ord Z-score. Positive values mark flood hotspots. Negative values mark coldspots.z_hotspot_class: 1 = Low (Z below -1.65), 2 = Moderate (Z from -1.65 to 1.65), 3 = High (Z above 1.65 up to 2.58), 4 = Very High (Z above 2.58).E2015_LANDO: Lando (Koppu). 0 = no flood, 1 = flood, 255 = no observation.E2016_KAREN_LAWIN: Karen / Lawin. 0 = no flood, 1 = flood, 255 = no observation.E2018_HABAGAT: Habagat / Henry / Inday / Josie. 0 = no flood, 1 = flood, 255 = no observation.E2018_OMPONG: Ompong (Mangkhut). 0 = no flood, 1 = flood, 255 = no observation.E2020_TY: Quinta / Rolly / Ulysses. 0 = no flood, 1 = flood, 255 = no observation.E2023_EGAY: Egay / Falcon / Habagat. 0 = no flood, 1 = flood, 255 = no observation.E2024_CARINA: Carina / Habagat. 0 = no flood, 1 = flood, 255 = no observation.E2024_KRISTINE: Kristine / Leon. 0 = no flood, 1 = flood, 255 = no observation.E2025_HABAGAT: Crising / Dante / Emong. 0 = no flood, 1 = flood, 255 = no observation.E2025_UWAN: Uwan (Fung-wong). 0 = no flood, 1 = flood, 255 = no observation.E2026_HABAGAT: August–September Habagat. 0 = no flood, 1 = flood, 255 = no observation.
no data value
-9999
coordinate system
EPSG:4326
events considered
Lando (Koppu), 16–27 October 2015 (E2015_LANDO)Karen / Lawin, 12–28 October 2016 (E2016_KAREN_LAWIN)Habagat / Henry / Inday / Josie, 16 July–14 August 2018 (E2018_HABAGAT)Ompong (Mangkhut), 13 September–1 October 2018 (E2018_OMPONG)Quinta / Rolly / Ulysses, 24 October–20 November 2020 (E2020_TY)Egay / Falcon / Habagat, 24 July–8 August 2023 (E2023_EGAY)Carina / Habagat, 20 July–3 August 2024 (E2024_CARINA)Kristine / Leon, 21 October–3 November 2024 (E2024_KRISTINE)Crising / Dante / Emong, 17 July–2 August 2025 (E2025_HABAGAT)Uwan (Fung-wong), 7–20 November 2025 (E2025_UWAN)August–September Habagat, 16 August–7 September 2026 (E2026_HABAGAT)
