Selecting control sites for post-fire ecological studies using biological criteria and MODIS time series data

LANDI M.A., DI BELLA C., OJEDA S., SALVATIERRA P., ARGAÑARAZ, Y BELLIS L.M.

Fire Ecology

The Association for Fire Ecology

Año: 2017

1933-9747

Wildlandfires play a key role in the functioning and structure of vegetation. Theavailability of sensors aboard satellites, such as Moderate Resolution Imaging Spectroradiometer (MODIS), makespossible the construction of a time series of vegetation indices (VI) and themonitoring of post-fire vegetation recovery. One of the used techniques tomonitor post-fire vegetation is the comparison of a burned site with anadjacent unburned control site. However, to date, there is no objective methodavailable for selecting these unburned control sites. Here, we propose threebiological criteria that the unburned sites must meet to be considered controlsites, as well as statistical methods based on the analysis of the propertiesof the Quotient Vegetation Indices time series (QVI), to detect unburned sitesthat meet the proposed criteria. We also test the performance of the proposedmethod by checking the pre-firedifference between burned and unburned sites, assuming that thehigher the number of met criteria, the greater the similarity. Therefore, we compare the differences between VItime series of burned sites and VI time series of unburned sites with the same vegetationcover, that met three, two, one and none of the proposed criteria. In addition,we compare the quality of QVI time series that met three, two, one and none of the proposed criteria. Ourresults show that for NormalizedDifference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) data, the difference betweenthe time series of burned and unburned sites gradually decrease with the increase of met criteria. A gradual increase is alsoobserved in the quality of the QVI time series with the increase of met criteria. Despite thelimitations present in the proposed method, our model represents an advancefrom the conceptual and methodological standpoints, since this is the firstproposal of a statistical method for selecting unburned control sites based onbiological criteria.