In swinersha/DAFmap: Direct forest patrols from threats

Introduction

DAFmap is a package that was written specifically to generate maps of forest threats to optimise patrol effort allocations and track patrol performance. Maps are generated for patrol sectors which are devised to describe the nuances of the area being protected (e.g. forest reserve, national park) in terms of the distibution of patrol teams, the location of settlements and access routes.

Threats can be measured for a period of intest and are assessed using Global Land Analysis and Discovery (GLAD) change alerts in combination with threat observations from patrols.

Threat assessment

Threats are standardised as % forest loss and as observations per 100 hours of patrolling. Observations should be grouped according to a generic threat type; at the moment these are limited to forest clearance (from GLAD), encroachment, illegal logging, and hunting, in order to aid interpretation. At present patrol data can be uploaded as .csv files describing the number of observations per observation type and the patrol effort in hours within each patrol sector. These can be extracted directly from SMART or be created by the user.

Standard threat metrics are then classified into four threat levels (low, medium, high and extreme) using user presecribed cut points. The treat levels can then be plotted onto the patrol sectors to give a map of threat intensity. Of course this can be done to some in SMART but the advantage of DAFmap is that remote sensing data are directly integrated. The total threat level for each patrol sector is calculated according to the maximum threat level observed across all observation types.

Directing patrols

Once the treat level has been classified for patrol sectors it is then possible to develop priorities for patrolling according to the human resources available. At the moment we use field offices as administrative units, which are responsible for several patrol sectors and each have a single patrol team.

By assigning an importance weighting to each treat level from low to exteme we can then calculate the total threat level for all the sectors allocated to each field office. From this the number of patrols that should be allocated to each sector in order to properly distribute according to threat level is calculated. This is particularly useful because it create a key performance indicator type framework for directing patrol effort and assessing the performance of patrol teams.

Report generation

Template scripts are present which generate an automated report based upon the input data. The report shows the threat level both graphically and numerically, along with the metrics that go into the threat level calculations and the derived key performance indicators.

Aspirations

At the moment, these scripts are very specific to our site, Hutan Harapan, in Sumatra and the reports generated are in Bahasa Indonesia, but I hope that conceptually they are useful to others who face similar problems. Illegal forest clearance is a massive problem and it can be extremely difficult to motivate and manage forest patrol teams to directly address this issue, particularly when perpetrators and well organised and violent. Any feedback and suggestions would be extremely welcome.

I am sorry that no example data are available but my data are sensitive and I don't have time to generate some hypothetical data at present. If you need any help getting the code to run please don't hesitate in getting in touch.

swinersha/DAFmap documentation built on May 6, 2019, 7:33 a.m.