Estimated Shallot Yield Area Using the Rapid Classification of Croplands Method

Abstract

Shallots are one of the horticultural commodities that have fluctuating prices. Market integration occurs horizontally but not vertically due to poor information systems at the producer and consumer levels. This study aimed to estimate the area of shallot land quickly using the rapid classification of croplands method. The research was conducted in Merek District, Karo Regency, North Sumatra. Primary data obtained from survey activities were processed using the Google Earth Engine platform. Classification and regression trees (CART) and random forest (RF) algorithms were used to classify land cover as onion and non-onion classes. The shallot land area based on this method was 74.4 hectares, with an area accuracy of 95% (RF) and 24% (CART) and a location accuracy of 92% (CART and RF). The rapid classification of croplands method can estimate land area quickly. It helps stakeholders who need information on shallot production projections and can be developed to improve the vertical market integration information system (market integration between producers and consumers). Some areas for improvement of this method are limited access and resolution, inability to describe up to the level of garden bunds, and the condition of the area covered by clouds, which will affect the accuracy of the results. Keywords: shallots, production estimation, google earth engine, remote sensing