| dc.description.abstract | Basalt Stem Rot (BSR) caused by the pathogen Ganoderma boninense is a
threat to the sustainability of oil palm agribusiness, especially in Indonesia and
Malaysia. This disease results in damage to plants, decreased production, and
death of plants, causing economic losses. Until now there has been no effective
control measures. Research is needed based on its characteristic which is
asymptomatic; symptoms are difficult to recognize early, so that fast, accurate and
wide-scale detection and classification is important. BSR G. boninense is also
known as soil borne disease; the occurrence of infection is influenced by the
properties of the soil.
The research was carried out in 2 stages 1) Detection and mapping using
remote sensing techniques and 2) Characterization of physical, chemical and
biological properties of the soil. The aims of this study were 1) To identify and
map the incidence of disease using remote sensing techniques using UAV, to
analyze vegetation indices (NDVI, GNDVI, SAVI, SR, CIgreen), application of
machine learning algorithms (CART, RF, SVM) 2) To analyze several properties
physical (texture, BD, porosity), chemical properties (pH, N, P, K, Ca, Mg, Cu,
Zn), biological properties (microbial population and Trichoderma) and 3)
Analyzing soil properties that correlate with disease incidence BPB G. boninense.
The research was conducted in Pabatu Serdang Bedagai Estate (Typic
distrudepts) 247ha and Tinjowan Estate (Typic Paleudults) 138ha; a total of
385ha, in oil palm plantation areas year 2004 and 2005. Image recording in June
2021, soil sampling in April 2022 at points of healthy (H) and infected (I) plants.
Analysis of physical and biological properties in the USU laboratory and chemical
properties in the PPKS laboratory.
The results of disease incidence research in Pabatu 24.1% and in
Tinjowan 27.2%. The lowest reflectant value is in R band, the highest is NIR;
reflectant in infected plants was lower in comparison in Pabatu for R (56.7%), G
(66.4%) and NIR (44.6%); in Tinjowan R (96.6%), G (97.2%) and NIR (91.9%).
The best vegetation index in Pabatu is SR, in Tinjowan GNDVI. The best
algorithm is SVM with 93.5% accuracy in Pabatu and 88.3% in Tinjowan. With
the SVM algorithm, it is possible to predict disease incidence in other estate
blocks whose images have been recorded with UAV, resulting in a distribution
map.
The dominant soil texture in Pabatu is sandy clay loam, in Tinjowan
sandy, BD is in the medium category and has good porosity. The pH value is in
the slightly acidic category, the parameters classified as low and very low are N,
Ca, CEC, Cu; high and very high categories are K, Mg and Zn. The difference in
P2O5 levels in Tinjowan is more than 75.4%. The population of microbes and
Trichoderma in Tinjowan was more than in Pabatu with a difference of +203.67%
and +63.06%. The correlation pattern of soil properties parameters that have an
effect on increasing the incidence of G. boninense BPB disease is soil porosity, CEC and those that have an effect on decreasing are BD, Cu and total
Trichoderma spp. The disease incidence distribution map is useful for controlling
and calculating plant productivity. It is expected that the results of this study will
contribute to the control of BSR G. boninense disease. | en_US |
