Methods

This section describes in detail how do you perform the study.

It contains: your data source, your objects, method developed or used, experimental set up, the outcomes and statistical analysis used. The latter includes: data presentation, statistical test conducted and its significance, software used.

It is recommended that :

1. you follow strictly the presentation format shown in a journal that you are aiming at.

2. your procedure should be explained clearly for reproducibility.

Sources:

Chapter 9 – Writing for Impact: How to Prepare a Journal Article

Andrew M.Ibrahim, Justin B.Dimick

https://www.sciencedirect.com/science/article/pii/B9780128099698000097?via%3Dihub

The Road to Better Presentation of Data: The Do’s and Don’ts

Irem Bayindir-Buchhalterhttps://www.advancedsciencenews.com/road-better-data-presentation-dos-donts/

Example

Methods

Light data

Following Bennie et al. [3], we used nighttime stable lights annual composite images, created with data from the Defense Meteorological Satellite Program’s Operational Linescan System (DMSP/OLS), downloaded from the National Oceanic and Atmospheric Administration archives (1992–2012, n = 21). These images capture upwardly reflected and directed nighttime light. The images are nominally at 1 km resolution, but are re-sampled from data at an equal angle of approximately 2.7 km resolution at the equator. These images cover spectral responses from 440 to 940 nm with the highest sensitivity in the 500 to 650 nm region. The spectral range encompasses the primary emissions from the most widely used sources for external lighting in Brazil: low pressure sodium (589 nm), high pressure sodium (from 540 nm to 630 nm) and mercury vapour (545 and 575 nm) [1,28].

Each pixel is represented by a digital number (DN) of between 0 and 63. Zero represents no detectable upward radiance, while brightly lit areas saturate at values of 63. Images were inter-calibrated and drift-corrected following the method of Bennie et al. [3]. An average calibrated image for both the first (1992–1996) and the last (2008–2012) five years was created and the difference was calculated. To assess the changes over the full period time, we considered pixels increasing or decreasing by more than a threshold of 3 DN units of difference between the averages of the first and last years. It was previously observed that over 94% of observed increases in DN of more than 3 units and over 93% of observed decreases of the same magnitude were consistently related to the directions of changes on the ground (e.g., expansion or contraction of urban and industrial areas) [3]. Following Gaston et al. [29] and Duffy et al. [30], we considered pixels as exposed to artificial light when they had values higher than 5.5 DN units. By using a threshold effectively twice the detection limit for change, we defined a conservative estimate of lit area and limited the extent to which dark sites may be classified as lit due to noise in the data set or calibration errors [29,30].

Vegetation type data

We used the vegetation map produced by the Brazilian Institute for Geography and Statistics [31], which is recommended as a good basis to compare with data obtained from remote sensing images [32]. This map presents both original native vegetation and current vegetation and land cover. The former portrays the original vegetation classes in Brazil likely found at the time of Portuguese colonisation [31], and the latter describes the vegetation now present [31]. Original vegetation includes 24 wider classes while the current is more detailed, including 52 classes (Table 1). The shapefile was produced by IBGE—Brazilian Institute of Geography and Statistics and accessed through REDD-PAC website (http://www.redd-pac.org/new_page.php?contents=data.csv) in WFS (web feature service) format.

The IBGE map divides vegetation into two broad classes: forests and non-forests [33]. Forests are divided into Ombrophilous Forest and Seasonal Forest. The former is further divided into three physiognomies (Dense, Open and Mixed) and the last into two (Deciduous and Semi-deciduous). All of these can be classified by up to five formations: Alluvial, Lowland, Sub montane, Montane and High-montane (Table 1). Non-forests are divided into four formations: Campinarana, Savanna, Steppe-savanna, and Steppe, which in turn can be divided into up to four formations: Forest, Woody, Shrubland, and Grassland. The map also classifies pioneer formations—that encompass vegetation influenced by rivers (Alluvial Areas), by the sea (Restingas), and by both (Mangroves)—Ecotones, Relict Vegetation and Water. When considering the current vegetation, it also includes Agriculture and Secondary Vegetation classes (Table 1).

Processing

To define the proportional area of each vegetation type that has been exposed to artificial nighttime light, we overlaid both original and current vegetation shapefiles on the DMSP data for the most recent five years (2008–2012). We extracted both the number of lit pixels and the total number of pixels inside each vegetation type and divided the first by the second. To assess changes, we overlaid the two vegetation shapefiles on the difference between the first (1992–1996) and the last (2008–2012) five years of DMSP data. We extracted the number of increasing pixels, decreasing pixels and the total number of pixels inside each vegetation type. We divided the number of increasing and decreasing pixels by the total in each vegetation type, achieving the proportional area where artificial light has been increasing and decreasing respectively.

 

Exposure of tropical ecosystems to artificial light at night: Brazil as a case study

Juliana Ribeirão de Freitas, Jon Bennie, Waldir Mantovani , Kevin J. Gaston

https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0171655