Workflow

Data preperation

• It is recommended to use Microsoft's Open R disriubtion
• Download and install R version 3.5.1 if not already present on system
• If you want to validate the model against remote sensed surface tempeatures, this approach does so using ASTER On Demand Surface Kinetic Temperature
• You could also potentially use Landsat 7/8 but would need to updates scripts accordingly
• If validating, custom locations for validation should be chosen such that the match the formant of "data/validation_sites.csv"
• ASTER LST data info: https://lpdaac.usgs.gov/dataset_discovery/aster/aster_products_table/ast_08
• If you want to replicate using MesoWest API to reterive historical weather data for model forcing, you will need an API key from Synoptic Labs
• MesoWest/ Synoptic info: https://developers.synopticdata.com/mesonet/
• Store raw ASTER data from desired periods after unzipping in "data/aster/raw"
• Store API token in file named local-token.txt in main directory (it is git ignored by default)

R scripts

1a-sat-data-preprocess.R

• Should only need to be run once
• This script simply imports all ASTER .tif data, converts all values to to deg C
• Then batch dumps in a single folder "data/aster/all"

1b-satellite-LST-extraction.R

• This script reterives metadata for all ASTER scenes and cleans it
• Then according to user inputs, selects a small subset of desired scenes for use in validation
• Default is no cloud cover, then THREE custom scenes are selected that have good coverage of sites
• Because of weird projection issues of ASTER that R libraries have trouble interpreting,
• A python script through QGIS is used to fix the projection issue within this script
• This must be done manually before the second half of script currently **
• Will also plot selected scenes using a predefined shapefile of the metro region (Maricopa County UZA in our case)

2-weather-data-retrieval.R

• This script requires the API key from Synoptic (free, need an account)
• This script will pull all weather data from all stations in a predefined area and all corresponding weather data
• Default is to reterive all station data availabe in Phoenix metro for 2000 to 2018

3-heat-transfer-model.R

• This is the heat transfer model
• Primary weather inputs: temperature, solar rad, humidity/dew, wind
• If using MesoWest data from 2__.R, it should be properly cleaned already
• Inputs of desired parameters of materials (such as asphalt concrete pavement specifications)
• Inputs of model run parameters (iterations, day simulation lenght, etc)

4-model-post-process

• computes summary statistics
• plots simulated pavement temperature profiles
• compares remotely sensed validation LST data to simulated data