Show simple item record

dc.contributor.advisorFaqih, Akhmad
dc.contributor.authorParwoko, Iqbal Dony
dc.date.accessioned2024-08-13T06:29:18Z
dc.date.available2024-08-13T06:29:18Z
dc.date.issued2024
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/157267
dc.description.abstractPertanian di Pulau Jawa sangat dipengaruhi oleh variabilitas iklim musiman. Prediksi panjang musim hujan sangat penting untuk perencanaan pertanian dan pengelolaan sumber daya air. Studi ini bertujuan mengembangkan model prediksi panjang musim hujan di Pulau Jawa menggunakan Convolutional Neural Network dan mengevaluasi model tersebut. Prediktor yang digunakan adalah empat data curah hujan luaran North American Multi-Model Ensemble dengan issued time Bulan Juni, Juli, Agustus, dan September serta lead time Bulan September hingga Mei. Penelitian ini berhasil mengembangkan prediksi panjang musim hujan di Pulau Jawa menggunakan CNN. Korelasi Pearson bernilai 0,38 hingga 0,51 dengan nilai korelasi tinggi terjadi pada issued time Bulan Juni di Jawa Timur. RMSE bernilai 3,29 hingga 3,80 dengan nilai berada di Jawa Barat bagian selatan. Standar deviasi ternormalisasi bernilai 0,68 hingga 0,94 dengan nilai terbaik berada di Jawa Barat dan Jawa Tengah. Model memiliki kemampuan moderat dalam menangkap hubungan linear antara data aktual dan prediksi tetapi kurang dalam menangkap nilai ekstrem dari data aktual. Hasil skill score cenderung merata di seluruh pulau dengan model CanSIPS-IC3 memiliki skill yang terbaik. Perbandingan panjang musim hujan menunjukkan bahwa prediksi model cenderung mengikuti pola aktualnya tetapi terdapat beberapa perbedaan nilai prediksi yang overestimate ataupun underestimate daripada nilai aktualnya.
dc.description.abstractAgriculture on Java Island is highly influenced by seasonal climate variability. Predicting the length of the rainy season is crucial for agricultural planning and water resource management. This study aims to develop a rainy season length prediction model in Java Island using a convolutional neural network and evaluate the model. The predictors used are four rainfall data outputs from the North American Multi-Model Ensemble with issued times in June, July, August, and September and lead times from September to May. This research successfully developed a rainy season length prediction model in Java Island using a CNN. Pearson correlation values ranged from 0.38 to 0.51, with high correlation values occurring at the issued time of June in East Java. RMSE values ranged from 3.29 to 3.80, with the highest values located in the southern part of West Java. Normalized standard deviation values ranged from 0.68 to 0.94, with the best values found in West Java and Central Java. The model has a moderate ability to capture the linear relationship between actual and predicted data but is less effective in capturing extreme values from the actual data. The skill score results tend to be evenly distributed across the island, with the CanSIPS-IC3 model having the best skill. A comparison of the rainy season length shows that the model's predictions tend to follow the actual pattern, but there are some differences in predicted values that are overestimated or underestimated compared to the actual values.
dc.description.sponsorship
dc.language.isoid
dc.publisherIPB Universityid
dc.titlePengembangan Model Prediksi Panjang Musim Hujan di Pulau Jawa Menggunakan Convolutional Neural Networkid
dc.title.alternativeDevelopment of a Rainy Season Length Prediction Model in Java Island Using Convolutional Neural Network
dc.typeSkripsi
dc.subject.keyworddeep learningid
dc.subject.keywordevaluasi performaid
dc.subject.keywordliebmannid
dc.subject.keywordmodel output statisticid
dc.subject.keywordnorth american multi-model ensembleid


Files in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record