Banana is one type of fruits that is very potential to grow to support food security because it contains source of vitamins, minerals, and carbohydrates. Jenawi District has a dry land area that is potential for the development of banana plants. The purpose of this research was to plan the development of banana plants in accordance with the land conservation aspect against the threat of the erosion. The research was conducted by making land unit map, field survey, analysis of soil sample in the laboratory, and data analysis of erosion hazard level. Land unit mapping was made by overlay method so there were 12 units of land with the same land conditions. Field survey and sample analysis at the observation point was conducted to obtain the parameters: slope gradient, depth of solum, surface unity, slope length, bulk density, soil texture, soil C-organic, and soil permeability. The analysis of land conservation aspects for banana plants was determined from the level of erosion hazard with the USLE (Universal Soil Loss Equation) method. The results showed that the threat of moderate to severe erosion hazard for banana crops was overcome by the efforts of land conservation techniques using a good construction bench terrace. The threat of erosion hazard after land conservation directives is very low to low. The amount of the erosion prediction for banana plant planning in garden is 35.80 t/ha/yr (unit B unit), land use of dry fields is 29.82 t/ha/yr (land unit I), 31.54 t/ha/yr (land unit K), and 13.72 t/ha/yr (unit land L).

The existence of vegetation in the watershed area as a function of maintaining the continuity of the process of soil infiltration, erosion resistance, and landslide resistance is essential. Studies that measure the existence of vegetation through vegetation index parameters by utilizing optical-based satellite imagery maps, as well as radiometric waves or Radio Detection and Ranging (RADAR), have been widely carried out. However, attempts to test it to formulate it into vegetation resilience indices on the scale of functional areas such as watersheds are still a challenge in itself. In this study, the vegetation resilience index was measured by looking for spatial change patterns from Normalized Difference Vegetation Index (NDVI) and the Soil-Adjusted Vegetation Index (SAVI) in 2015-2021. This spatial pattern change is associated with anthropogenic factors represented by the dynamics of changes in vegetation to housing and population distribution. The use of residential land is obtained from the interpretation of optical-based satellite imagery maps from Google Earth sources mitigated by the administrative boundaries of villages. The population change was obtained from statistical data on the village’s population in the same year. The change in vegetation index, residential, and inhabitants were examined by the Spatial Autocorrelation (SA) using the Global Moran’s Index (GMI). Finally, the changes in the spatial patterns in vegetation indices, residentials index, and inhabitant index were conducted with correlation analysis to determine the relationship between these parameters. The results of GMI indicate a tendency towards dispersion in NDVI and SAVI, which showed a linear index decline in the dry season but weak linear indices incline in the wet season. The trend of dispersion patterns also occurs in residential and inhabitants, showing a linear index decline with a high confidence level (R2=0.9171 in residentials and R2=0.5872 in inhabitants). The relationship between pattern changes in NDVI and SAVI and Residential and Inhabitant found by the following formula; NDVI = 0.214173 -3.12952 * Residential + 6.457788 * Inhabitant; SAVI = 0.101544 + 0.120294 * Residential + 1.963223 * Inhabitant.