Penilaian Tingkat Keparahan Area Pascakarhutla berbasis Poligon pada Aplikasi Mobile SIPAKARHUTLA

Abstract

Kebakaran hutan dan lahan (karhutla) merupakan permasalahan yang terus berulang di Indonesia dan menimbulkan dampak luas terhadap lingkungan maupun sosial ekonomi. Penilaian tingkat keparahan area pascakarhutla menjadi dasar penting dalam upaya rehabilitasi. Aplikasi mobile SIPAKARHUTLA telah dikembangkan untuk mendukung penilaian tersebut, tetapi masih memiliki keterbatasan pada pengisian lokasi saat offline, penyimpanan data, representasi lahan, serta penambahan plot pada laporan yang telah dikirim. Penelitian ini bertujuan menyempurnakan fitur aplikasi mobile SIPAKARHUTLA dan mengintegrasikan model Convolutional Neural Network (CNN) untuk klasifikasi tingkat keparahan karhutla menggunakan metode prototyping. Penelitian ini menghasilkan aplikasi mobile SIPAKARHUTLA versi terbaru dengan fitur meliputi pengisian lokasi offline berbasis koordinat, penyimpanan data otomatis, representasi lahan berbentuk poligon dengan validasi spasial, penambahan plot pada fitur edit laporan, peta laporan tingkat keparahan, dan klasifikasi tingkat keparahan melalui foto berbasis CNN. Pengujian black box testing terhadap 120 skenario menunjukkan 119 skenario berhasil dijalankan, dengan tingkat keberhasilan sebesar 99,17%, sedangkan pengujian usability dengan Post-Study System Usability Questionnaire (PSSUQ) memperoleh skor overall 1,51 pada skala 1 sampai 7. Hasil tersebut menunjukkan aplikasi berfungsi dengan baik dan mudah digunakan untuk mendukung penilaian pascakarhutla di lapangan.

Forest and land fires (karhutla) are a recurring problem in Indonesia with wide-ranging environmental and socio-economic impacts. Severity assessment of post-fire areas is an important basis for rehabilitation efforts. The SIPAKARHUTLA mobile application has been developed to support this assessment, but it still has limitations in offline location input, data storage, land representation, and the addition of plots to submitted reports. This study aims to enhance the features of the SIPAKARHUTLA mobile application and integrate a Convolutional Neural Network (CNN) model for karhutla severity classification using the prototyping method. This study produced an updated version of the SIPAKARHUTLA mobile application with features including coordinate-based offline location input, automatic data saving, polygon-based land representation with spatial validation, plot addition in the report editing feature, a severity report map, and CNN-based severity classification through photos. Black box testing on 120 test scenarios showed that 119 scenarios were successfully executed, achieving a success rate of 99.17%, while usability testing using the Post-Study System Usability Questionnaire (PSSUQ) obtained an overall score of 1.51 on a scale of 1 to 7. These results indicate that the application functions well and is easy to use in supporting post-fire assessment in the field.