Abstract:
Floods pose a global threat and are responsible for major risks inflicted on communities
worldwide. Timely and accurate mapping of floods plays a crucial role in effective
hazard management, assessing risks, and developing plans for hazard mitigation. This
can be realized through the right application of integrated research that could fill
key research gaps, such as the limitation of present techniques, lacking capacity, and
mapping challenges of flooding in vast semiarid areas coupled with poor validation of
Synthetic Aperture Radar (SAR)-based flood mapping methods. In response to these
challenges, this research focused on the development of advanced flood mapping solutions
that leverage the capabilities of remote sensing technologies, specifically multitemporal
SAR and multi-spectral imagery, to improve the timeliness and accuracy of
flood mapping. The analysis conducted between September 10, 2022, and September
30, 2022, using Sentinel-2 MS imagery with NDWI and Sentinel-1 SAR imagery revealed
significant flood events in various areas of Pakistan. The flood mapping results
from both datasets were compared, showing relatively consistent estimates of the flood
extent and the number of exposed people across different provinces. However, there
were noticeable differences in the estimated affected cropland, highlighting the variability
in flood mapping results obtained from different satellite imagery sources. In the
SAR imagery, an area of about 1,258,544 hectares and, according to MS-based NDWI
analysis, 1,245,226 hectares, was detected to be flooded all over Pakistan. Much of the
impact was on agricultural land, affecting about 303,306 hectares and 515,455 hectares
of cropland for SAR and MS-based NDWI results, respectively. As the final steps,
this study compared the kappa coefficient and overall accuracy statistics of the flood
mapping results from the different thresholds of SAR and NDWI. The best accuracy
was noted when the threshold value in SAR was 1.19, and for an NDWI threshold of
0.10, overall accuracy equalled 87.12% with a kappa-coefficient equaling 0.7424. The
accuracy assessments are very important to ascertain the efficacy of the suggested flood
mapping methodologies. These findings are very relevant to disaster response and mitigation
strategies. The purpose is to reduce the vulnerability of various communities
to future flood events in Pakistan.
