Recommending Environmental Big Data Using Semantically Guided Machine Learning - Large Scale and Big Data: Processing and Management - page 482

freely available in the NASA website for different products. The instrument type
on-board is a medium-resolution, multispectral, cross-track scanning radiometer
with daylight reflection, and day/night emission spectral imaging. The MODIS
instruments acquire data in three native spatial resolutions: Bands 12 in 250 m, Bands
37 in 500 m and Bands 836 in 1000 m. The images are freely available in NASA
website for different products. For this study, we have downloaded the MODIS/
Terra Land Surface Temperature and Emissivity (LST/E) 8-day L3 Global 0.05Deg
CMG product (MOD11C2) version 5 for Tullochgorum, Tasmania during the period
December 2010 to June 2012. This product provides per-pixel temperature and emis-
sivity values in a sequence of swath-based global products. MOD11C2 product com-
prises the Science Data Set (SDS) layers for day time and night time observations:
LSTs, quality control assessments, observation times, view zenith angles, clear sky
coverage, and emissivity for bands, that is, 20, 22, 23, 29, 31, and 32 (Figure 15.10).
This was explicitly mentioned in the Land Processes Distributed Active Archive
Center (LP DAAC) portal of NASA that MOD11C2 products are ready for use in sci-
ence applications. Time series were created from these downloaded image data files
based on image processing techniques (Figure 15.11) [14,69-71]. Similarly, Tropical
Rainfall Measuring Mission (TRMM) 3B42 satellite-based precipitation prod-
ucts were constructed from the post real-time TRMM Multi-Satellite Precipitation
Analysis (TMPA) product, 3B42. The purpose of the 3B42 algorithm is to produce
TRMM-adjusted merged-infrared (IR) precipitation and root-mean-square (RMS)
precipitation-error estimates. The algorithm consists of two separate steps. The first
step uses the TRMM VIRS and TMI orbit data (TRMM products 1B01 and 2A12)
and the monthly TMI/TRMM Combined Instrument (TCI) calibration parameters
(from TRMM product 3B31) to produce monthly IR calibration parameters. The sec-
ond step uses these derived monthly IR calibration parameters to adjust the merged-
IR precipitation data. These gridded estimates are on a 3-hour temporal resolution
and a 0.25° × 0.25° spatial resolution, which provided the adjusted merged-IR pre-
cipitation (mm/hr) and RMS precipitation error [14,39,69-71]. Figure 15.12 shows an
example of MODIS production on a global scale.
15.4.5 u nsuPerviseD k nowleDge r eCommenDation
15.4.5.1 Time Series Integration
As described in the previous section, all time series data were integrated together to
form a complete list of environmental data set on a dynamic time scale. Extracted
and integrated time series data were processed through the harmonization layer,
semantic cross validation layer, and feature representation layer to form a representa-
tive semantic feature based. Refreshed, complemented, cross-validated, and normal-
ized integrated time series matrix was formed, which had 40 different environmental
variables from five different big environmental data sources. Semantic attribute
matrix had 40 columns and n rows, where n was representing dynamic total number
of days during which the integration was performed. Uniqueness of this processing
was that depending on the total time of integration the whole matrix could dynami-
cally adapt to represent most recent and most relevant data matrix (Figure 15.13).