Table 1.
Predictor and outcome variables used in BRT models.
| Variable description | Source information | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Type | Variable | Descriptions | Units | Time-dependent variable | Data collection | Source | Spatial resolution | Temporal resolution | Units |
|
Environment predictor variable |
FC | Forest cover in a 2-km side square grid around village | % | yes (year) | UMD/hansen/global_forest_change_2015 | Univ. Maryland | 30 m | year | % per output grid cell |
| FL | Annual forest loss in a 2-km side square grid around village | % | yes (year) | UMD/hansen/global_forest_change_2015 | Univ. Maryland | 30 m | year | % per output grid cell | |
| CAR | Cumulative annual rainfall. (average in a 2-km side square grid around village) | mm | yes (year) | TRMM/3B42 | TRMM | ~27 km | 3hrs | mm/3hrs x pp | |
| LST | Land surface annual mean temperature | °C | yes (year) | MODIS/006/MOD11A1/LST_Day_1 km | NASA, MODIS, LST | 1km | 1 day | °C | |
| NDVI | Normalized difference vegetation index. (average in a 2-km side square grid around village) | index | yes (year) | MODIS/006/MOD13Q1 | NASA, MODIS, Vegetation Index | 250 m | 8 days | index | |
| NDWI | Normalized difference water index. (average in a 2-km side square gride around village) | index | yes (year) | LANDSAT/LC05C01/T1_T1 LANDSAT/LC08/C01/T1_T1 LANDSAT/LE07/C01/T1_T1 | NASA, LANDSAT | 30 m | 16 days | index | |
| SDR | Euclidean shortest distance to rivers | kilometers | no | JRC/GSW1_0/GlobalSurfaceWater (occurrence) | JRC/GSW Historical data | 30 m | once | kilometers | |
|
Social predictor variable |
TPV | Travel time to major populated villages/towns | minutes | no | Oxford/MAP/accessibility_to_cities_2015_v1_0 | Oxford (MAP), Google, (JRC) & Univ. Twente | 1Km | once | minutes |
| POPD | Population in a 5-km side square grid around village) | log (number people) | no | WorldPop/POP | WorldPop 2015 | ~100 m | once | people in ~100 × 100 m grid cell | |
| Outcome variable | Malaria high-risk | Village with API** > 10 cases/1000 people | binary (yes/no) | yes (year) | — | Surveillance system of Peruvian ministry of health | — | week | reported cases |
| Malaria very-high-risk | Village with API > 50 cases/1000 people | binary (yes/no) | |||||||
**API = Confirmed cases in a year *1000 / total population.
*PP: Per-pixel.