OBJECT AND PIXEL BASED LANDUSE LANDCOVER MAPPING AND COMPARISON, BASED ON MEDIUM AND HIGH RESOLUTION SATELLITE IMAGERY A CASE STUDY OF DEH AKRO-II AND CHOTIARI WETLAND COMPLEX, SINDH, PAKISTAN.

Climate change and anthropogenic activities badly affects the wetlands worldwide. Mapping of such changes set base of the recovery planning. This study aims to make a comparison between object based and pixel based techniques applied on medium, high and very high resolution images of Landsat 5 TM, Sentinel 2A and Planet Scope, respectively for developing Land Use and Land Cover (LULC) as they prove to be really effective in mapping and monitoring of wetland. LULC of two most badly effected wetlands complexes of Sindh, Pakistan, namely Deh Akro-II wetland complex and Chotiari wetland complex, have been developed and analyzed for changes over 30 years. For object based classification decision rule was applied on Landsat- 5 image of year 1990 and Sentinel 2A image for year 2020. However, supervised and unsupervised classification, Maximum likelihood and ISO cluster techniques have been used on very high spectral resolution imagery of Planet Scope. After classifications, accuracy assessment was performed by using random sampling technique. High accuracy of overall and kappa coefficient values shows that LULC classes were successfully developed by both pixels based and object based classification. Supreme level of kappa accuracy with 92.34% has been achieved when object based classification was performed on sentinel 2 image of Chotiari wetland complex, with overall accuracy of 94.22%. Whereas, maximum likelihood classification also resulted with very accurate results with over all accuracy of 96.40% and its Kappa coefficient is 93.49% for Deh Akro-I Wetland complex. Results revealed that major changes have been observed in water bodies and agriculture classes of both areas as water level of Deh Akro-II wetlands has been severely declined by 9.12 km2, while agriculture has been surprisingly increased up to 191.93 km2 in 30 years’ time span. Contrary, water level of Chotiari wetland complex has been shockingly increased up to 82.69 km2 after the construction of reservoir in 2003. Results also illustrated that rest of the land covers of Chotiari wetland complex has been greatly overwhelmed by upsurge of water level. This study revealed that remote sensing techniques are very effective for improved monitoring and mapping of wetlands and for betterment of wetlands health and flora and wildlife depends upon them.