Landsat images of Kozhikode City and its suburbs for the year 1993, 2003, 2013 and 2023 were downloaded from USGS to analyze land use land cover change. A False Color Composite (FCC) image was created by stacking multiple bands of satellite images into a single image using the ERDAS software's layer stack tool (Figure 3). These images have been undergone atmospheric, radiometric and geometric correction. Supervised classification with the Maximum Likelihood Classifier was employed. NRSC Level I classification was adopted for the study.

For understanding the changes in LULC pattern, the study area was categorized into five LULC classes- built up, agriculture land, vegetation, barren land and water bodies. LULC maps prepared for the year of 1993 and 2003 is shown in Figure 4.

Figure 3

Accuracy assessment of the classified image was done. The ground truth was verified using Google Earth, SOI Toposheet and field survey. Satellite images of 1993, 2003 and 2013 were mostly verified using Google Earth and SOI Toposheet. Classified image for the year 2023 is validated with field verification. Sample points of latitude and longitude transferred to handheld GPS for field verification. The accuracy assessment was carried out by taking 220 sample points randomly. 52 samples for built up, 43 samples for agriculture, 44 samples for vegetation, 41 samples for barren land and 30 for water bodies and were taken. The researcher has visited all the selected sample area for ground truthing. Then, the Error Matrix and Kappa Co-efficient were generated for assessing the accuracy of classified images with ground verification. The error matrix or confusion matrix signifies the classified and actual result in table format. Classified LU/LC is placed in rows and ground verified samples are placed in columns. Cohen’s Kappa Coefficient is also widely accepted method for the classified LU/LC accuracy assessment by many researchers (Sanjoy Roy et al., 2015 and Arumugam et al, 2021). Kappa Co-efficient computes degree of agreement between two variables that measure the same land use. The result is less than or equal to one with the values near to one represents the perfect agreement and near to zero represents the worst agreement. Error matrix of LULC accuracy assessment is given in Table 1.

Figure 4 Table 1

	Built up	Agriculture	Vegetation	Barren land	Water bodies	Total (User)	User accuracy
Built up	50	1	1	0	0	52	96.15
Agriculture	1	42	0	0	0	43	97.67
Vegetation	2	0	42	0	0	44	95.45
Barren land	1	0	0	40	0	41	97.56
Water bodies	0	0	0	1	29	30	96.67
Total (Producer)	54	43	43	41	29	210
Producer Accuracy	92.59	97.67	97.67	97.56	100.00
Overall Accuracy		= 96.67 = 0.94
Kappa Co-efficient

In the year 1993, built up land covered an area 1371.12 hectares and shared only 3.59 % of study area. Area under agriculture was 7765.73 ha (20.36%), vegetation was 26536.73 ha (69.56%), barren land was 345.83 ha (0.91%) and water bodies were 2129.24 ha (5.58%). In the year 2003, area under built up, agriculture, vegetation, barren land and water bodies was 3621 ha (9.49 %), 6952.90 ha (18.23%), 25206.27 ha (66.07%), 269.87 ha (0.71%) and 2098.13 (5.50%) respectively. In the year 2013, area under built up was 5006.96 ha (13.12%) and agriculture was 6644.77 ha (17.42%). Vegetation covered 63.24% of study area with an area of 24126.92 hectares, whereas barren land and water bodies covered an area of 269.41 ha (0.71%) and 2100.60 ha (5.51%) respectively. In the year 2023 built up area encompassed 34.63 % of study area with an area of 13211.38 ha while agriculture was 6032.06 ha (15.81 %). Area under vegetation was 16790.58 ha (44.01). Barren land occupied only 98.19 ha (0.26%) and water bodies covered 2016.44 ha (5.29%). Table 2 and Figure 5 shows land use land cover of Kozhikode City and its suburbs.

The result of spatio temporal changes of LULC classification has shown in Table 3 and Figure 6 reveals that Kozhikode city and suburbs witnessed drastic changes. During the period 1993-2003, there was rise in built up land, whereas agriculture, vegetation, barren land and water bodies declined. Built up area shows positive growth and increased by 2250 ha (164.13%). Agriculture, vegetation, barren land and water bodies decreased by 812.83 ha (10.47%), 1330.46 ha (5.01%), 75.96 ha (21.96%) and water bodies 31.11 ha (1.46 %) respectively. During the period 2003 and 2013, built up area expanded by 1385.47 ha (38.26 %) and water bodies shows slight increase of 2.47 ha (0.12%). Area under agriculture, vegetation and barren land declined by 1079.34 ha (4.28 %), 308.13 ha (4.43%), and 0.47 ha (0.17 %) respectively. during 2013 -2023, built up area drastically expanded by 8204.42 ha (163.86%) whereas vegetation, agriculture, barren land and water bodies declined by 7336.34 ha (30.14 %), 612.71 ha (9.22%), 171.21 ha (63.55%) and 84.16 ha (4.01%) respectively.

Table 3

LULC Class	1993-2003		2003-2013		2013-2023
	Changes in Area (Ha)	Changes in Area (%)	Changes in Area (Ha)	Changes in Area (%)	Changes in Area (Ha)	Changes in Area (%)
Agriculture land	-812.83	-10.47	-308.13	-4.43	-612.71	-9.22
Built-up land	+2250.37	+164.13	+1385.47	+38.26	+8204.42	+163.86
Vegetation	-1330.46	-5.01	-1079.34	-4.28	-7336.34	-30.41
Barren land	-75.96	-21.96	-0.47	-0.17	-171.21	-63.55
Water bodies	-31.11	-1.46	+2.47	+0.12	-84.16	-4.01

Figure 6 Figure 7

Land Use Land Cover of Kozhikode city and suburbs have been analyzed for the years 1993, 2003, 2013 and 2023 using Landsat images. The findings of the study revealed that built up area increased due to rapid urbanization and huge amount of vegetation and agricultural land were converted to built up land. Real estate pressure leading to the encroachment of agricultural land is a significant problem in Kozhikode city, as it is in many urban areas. As cities expand, the demand for land rises, agricultural land and vegetative area become prime target for real estate and illegal construction. Thus, the findings of this study are significant in implementing sustainable planning interventions in Kozhikode City and suburbs.