Hardware Requirements

Note: the below text applies only for those who want to build models and predict for themselves. The hardware requirements for the open access method is negligible and will work on any laptop or phone. For the open source part of Forest Foresight you will need to process the data yourself. This requires a little bit of hardware but is for most areas doable on a normal computer.

Storage

At time of writing the data of Forest Foresight comprise of approximately 1.3GB per tile. Have a look here to see how many tiles cover the area you are interested in. This is the approximate size of the data if you download the data for the whole time series, meaning everything from 1-1-2021 till to date. If you want only the most recent data it will be less, about 400MB if you just download the latest year.

Memory

When you start processing you will load training data into your computer. Assume approximately 1.2GB per tile per year of training data with a data sample size of 0.3, which is the standard for ff_run, if you will use the area of the entire tile and all features. You can also set the auto_sample in ff_run to TRUE to automatically determine the sample size that your computer can handle. Be aware that this might then be very small if you want to process on a large area for a long time series with a small RAM.

CPU

The CPU is not really important for Forest Foresight, most of the processing time comes from reading and writing from your disk. So almost all CPU’s will do.