Geospatial mapping of land use and land cover for watershed management in the Nidnegal micro watershed, Karnataka

Introduction

Watersheds are the fundamental hydrological units that sustain water resources, soil productivity, biodiversity, and agricultural livelihoods. However, rapid population growth, expanding agricultural activities, urbanization, and economic development have exerted increasing pressure on the limited land resources within watersheds, leading to significant changes in land use and land cover (LULC). These changes alter natural hydrological processes, accelerate soil erosion, reduce groundwater recharge, degrade water quality, and threaten the long-term sustainability of watershed ecosystems. A comprehensive understanding of land use and land cover dynamics is therefore essential for effective watershed planning and management. Land use reflects human activities carried out on the land, whereas land cover represents the physical and biological features present on the Earth's surface. The interaction between these components influences runoff generation, sediment transport, vegetation cover, and the overall ecological health of a watershed. Accurate and up-to-date LULC information provides a scientific basis for assessing watershed conditions, identifying environmentally sensitive areas, and planning appropriate soil and water conservation measures. Remote sensing technologies have a significant potential for mapping and extracting information about forest structure and biomass (Sinha et al., 2015) [6] . With advances in remote sensing and Geographic Information Systems (GIS), monitoring spatial and temporal changes in land use has become more efficient and reliable. The remote sensing data has many applications. One of the important applications is land use land cover classification (Koppad and Tikhile, 2013) [4] . These geospatial technologies facilitate the evaluation of land resource dynamics, support watershed prioritization, and aid in developing sustainable land management strategies. Integrating LULC assessment with watershed management enables informed decision-making for conserving natural resources, enhancing agricultural productivity, reducing land degradation, and ensuring the sustainable use of land and water resources for present and future generations. Considering the importance of accurate land use information for effective watershed planning and management, the present study was carried out to prepare a high-resolution land use and land cover map of the Nidnegal micro watershed using remote sensing and GIS techniques

Materials and Methodology Study area

The Nidnegal micro watershed is one of the seven micro watersheds delineated within the Hiremorab sub-watershed (4D4D2) of Hirekerur Taluk, Haveri district, Karnataka, India. It was identified as the highest-priority micro watershed based on morphometric analysis due to its greater susceptibility to soil erosion and land degradation. The watershed was selected for the present study to assess land use and land cover dynamics and to support sustainable watershed management through geospatial techniques.

Preparation of Land Use Land Cover (LULC) Map

The Land Use Land Cover (LULC) map of the Nidnegal micro watershed was prepared using the Google Earth Engine (GEE) cloud computing platform through a supervised classification approach employing the Random Forest (RF) algorithm. Highresolution QuickBird satellite imagery (0.61 m spatial resolution) was utilized for the preparation of the Land Use Land Cover (LULC) map of the selected micro watershed. Initially, cloud-free satellite imagery of the study area was collected and pre-processed in GEE to improve image quality and ensure accurate classification. Pre-processing included image filtering, mosaicking, clipping to the watershed boundary, and generation of spectral indices. Normalized Difference Vegetation Index (NDVI) and False Colour Composite (FCC) images were generated to facilitate the identification and visual interpretation of different land cover types. Training samples representing six major land use and land cover classes were selected using NDVI, FCC imagery, and field observations. Ground truth data were collected from representative locations across the micro watershed using GPS during field surveys. The combined use of field observations and very high-resolution imagery improved the accuracy and reliability of the training dataset. The training samples were converted into an ee.FeatureCollection() in GEE and used to train the Random Forest classifier. The classifier learned the spectral characteristics of each land cover class from the training dataset and subsequently classified all image pixels into their respective classes. The resulting classified image was exported as the final LULC map of the selected micro watershed.

Land Use Land Cover Classes

Based on spectral characteristics, field verification, and visual interpretation of QuickBird imagery, the study area was classified into six land use and land cover categories.

Accuracy Assessment

The accuracy of the classified LULC map was evaluated using an independent set of reference samples derived from field observations and high-resolution QuickBird satellite imagery (0.61 m spatial resolution). The reference samples were compared with the classified output to generate a confusion matrix in Google Earth Engine.

Experimental Results

The Land Use Land Cover (LULC) map of the Nidnegal micro watershed (MWS-5) was prepared using high-resolution satellite imagery (QuickBird -0.61 m resolution) through visual image interpretation and digitization in a Geographic Information System (GIS) environment. Based on the image interpretation, the study area was classified into different land use land cover categories and the resulting LULC map is presented in Fig.

Discussion

The land use and land cover (LULC) analysis of the Nidnegal micro watershed identified six major land cover classes, with agricultural land occupying the largest area (488.75 ha; 52.95%) followed by scrub forest (372.35 ha; 40.34%) while plantation (2.65%), built-up area (2.23%), water bodies (1.73%) and fallow land (0.11%) occupied comparatively smaller proportions of the watershed. The dominance of agricultural land indicates that the watershed is primarily agrarian whereas scrub forests are largely confined to hillock regions with relatively steeper slopes and shallow soils. Similar land use patterns were reported by Ravikumar, (2020) [1] in the Koranahalli sub-watershed, where cultivated land (60.15%) and scrubland (36.52%) were the predominant land cover classes. The LULC classification achieved an overall accuracy of 91 per cent with a Kappa coefficient of 0.89, indicating a high level of agreement between the classified map and reference data. Comparable classification accuracies have been reported by Rawat and Kumar (2015) [2] (90.29-92.13% overall accuracy; Kappa 0.82-0.91), Das and Angadi (2020) (92.16-95.16 per cent overall accuracy; Kappa 0.90-0.94), and Koppad et al. (2020) [3] , who achieved an overall classification accuracy of 97.09 per cent using IRS LISS-IV imagery. These findings confirm that the generated LULC map is sufficiently accurate and reliable for subsequent analyses such as carbon stock assessment, soil erosion modelling, land suitability evaluation and watershed management planning.

Conclusion

The LULC analysis indicated that the Nidnegal micro watershed is predominantly characterized by agricultural land and scrub forest, which together occupy more than 93 per cent of the total geographical area. The dominance of cropland reflects the agricultural importance of the watershed, while the extensive scrub forest on hillocks plays a significant role in protecting the landscape from soil erosion and maintaining ecological stability. The comparatively smaller extent of plantations, built-up areas, water bodies and fallow land suggests limited urban expansion and localized land-use diversification. Furthermore, the high overall classification accuracy (91.00%) and Kappa coefficient (0.89) confirm the reliability of the generated LULC map, demonstrating its suitability as a baseline dataset for soil erosion assessment, land suitability evaluation for agroforestry interventions and watershed management planning.

Acknowledgement

AcknowledgmentThe study was part of REWARD project, funded by the World Bank and Government of Karnataka and Centre of Excellence on Watershed Management, GKVK, Bangalore. The authors duly acknowledged the technical and financial support.

References

1. Ravikumar D. Micro level land and water resource development plan for Koranahalli subwatershed in Tarikere Taluk, Chikmagalur District using remote sensing and GIS techniques. Karnataka: Kuvempu University; 2020.
2. Rawat JS, Kumar M. Monitoring land use/cover change using remote sensing and GIS techniques: A case study of Hawalbagh Block, District Almora, Uttarakhand, India. Egyptian Journal of Remote Sensing and Space Science. 2015;18(1):77-84.
3. Koppad AG, Banavasi P, Sarfin S. Assessment of land use land cover and carbon sequestration in forests of Joida Taluk of Uttar Kannada district using remote sensing technique. Journal of Applied and Natural Science. 2020;12(3):344-348.
4. Koppad AG, Tikhile P. Influence of land use land cover classes on carbon sequestration in soils of Sirsi and Siddapur Taluka of Uttara Kannada District, India. International Journal of Current Research. 2013;5(4):10121015.
5. Kiran VSS, Srivastava YK, Jagannadha Rao M. Utilization of resources at LISS IV data for infrastructure updation and land use/land cover mapping: A case study from Simlipal Block, Bankura District, West Bengal. International Journal of Advance Remote Sensing and GIS. 2014;3(1):592-597.
6. Sinha S, Jeganathan C, Sharma LK, Nathawat MS. A review of radar remote sensing for biomass estimation. International Journal of Environmental Science and Technology. 2015;15(12):1779-1792.