The classic way of predicting fores

The classic way of predicting forest fire behavior relies on the evolution results provided by a certain forest fire
spread simulator. Typically, the input parameters needed by the underlying fire simulator such as the initial state
of the fire front (RF = real fire), terrain characteristics, vegetation types, meteorological information and so on are
obtained/estimated at a certain time ti, fed into the simulator in order to provide a fire front evolution at a later time
ti+1. Comparing the simulation result (Simulated Fire=SF) from time ti+1 with the advanced real fire (RF) at the same
instant, the forecasted fire front tends to differ to a greater or lesser extent from the real fire line. One reason for this
mismatch is that the classic prediction of the fire is based on a single set of constant and uniform input parameters. To
overcome this drawback, a simulator independent data-driven prediction scheme was proposed to optimize dynamic
model input parameters [6]. Introducing a previous calibration step as shown in Figure 1, the set of input parameters
is calibrated before every prediction step. The solution proposed come from reversing the problem: how to find
a parameter configuration such that, given this configuration as input, the fire simulator would produce predictions
that match the actual fire behavior. This process is defined as a parameter calibration process. Once, the input
parameter set that best describes the current behavior of the fire has been determined, this set of parameters could
also be used to describe best the immediate future, assuming that meteorological conditions remain constant during

Results (Indonesian) 1:

The classic way of predicting forest fire behavior relies on the evolution results provided by a certain forest firespread simulator. Typically, the input parameters needed by the underlying fire simulator such as the initial stateof the fire front (RF = real fire), terrain characteristics, vegetation types, meteorological information and so on areobtained/estimated at a certain time ti, fed into the simulator in order to provide a fire front evolution at a later timeti+1. Comparing the simulation result (Simulated Fire=SF) from time ti+1 with the advanced real fire (RF) at the sameinstant, the forecasted fire front tends to differ to a greater or lesser extent from the real fire line. One reason for thismismatch is that the classic prediction of the fire is based on a single set of constant and uniform input parameters. Toovercome this drawback, a simulator independent data-driven prediction scheme was proposed to optimize dynamicmodel input parameters [6]. Introducing a previous calibration step as shown in Figure 1, the set of input parametersis calibrated before every prediction step. The solution proposed come from reversing the problem: how to finda parameter configuration such that, given this configuration as input, the fire simulator would produce predictionsthat match the actual fire behavior. This process is defined as a parameter calibration process. Once, the inputset parameter yang paling tepat menggambarkan perilaku api saat ini telah ditetapkan, set parameter bisajuga dapat digunakan untuk menggambarkan terbaik masa depan, dengan asumsi bahwa kondisi meteorologis tetap konstan selama