The Mongolian Data Cube; First Test Products Produced

The Mongolian Data Cube will form a central component of the SIBELIUs infrastructure which will allow time series of satellite data and derived products, for example relating to pasture and snow, to be queried by staff at NAMEM and by other project stakeholders.

The implementation of the Mongolian Data Cube within an initial test environment is almost complete, with most of the scripts and database structures now in place. The Open Data Cube architecture, pioneered by Geoscience Australia and now its own development entity, has the stated aim of opening up access to analysis ready data (ARD) to a wider range of users. Once an Open Data Cube is deployed covering a given geographical region, it allows for the rapid expansion in the use of satellite data in that area, often including unforeseen applications that are stimulated once different user groups see what data and resultant information is available.

This increase in capability is due to the structured deployment of satellite data within the Open Data Cube’s database which allows for the easy viewing of data both spatially and temporally. For example, allowing users to view data either over large areas of the Earth or stretching back into the past to gain an insight into how systems have changed over time. Crucially, it allows a single system to be deployed that allows users to rapidly interrogate data across the different dimensions of the Cube, spatially and or temporally, with minimal training required.

The Mongolian Data Cube (MDC) is based on the Open Data Cube concept, as pioneered by the Australian Geoscience Data Cube, and will supply products at a higher spatial and temporal resolution than was previously possible to the SIBELIUs partners both in Mongolia and the UK. The SIBELIUs project will include two Data Cubes: a “high” resolution Data Cube and a “medium” resolution Data Cube. The high-resolution MDC draws upon data acquired by the Landsat8 and Sentinel-2 satellites, while the medium-resolution MDC incorporates data from the satellites carrying the MODIS and VIIRS sensors. The products from these systems will then be served out to stakeholders via web applications for further analysis and to support their operations. The initial SIBELIUs MDC products focus on measuring two environmental conditions: pasture and snow.

The processing chain that transforms the initially downloaded satellite data into useful products has several stages. The first stage is to generate the analysis ready data, which includes removing or the masking of unwanted atmospheric effects. The next stage is to generate several intermediate indices. The pasture products are mainly derived from an index known as the normalized difference vegetation index (NDVI), which uses two satellite bands in the visible red and near infrared wavelengths. The normalized difference between these bands is correlated with the vegetation present on the surface of the Earth, which in Mongolia is mainly the pasture covering the country’s extensive rangelands. The snow products likewise use the normalized difference snow index (NDSI), which works in a similar way to the NDVI but instead compares data between the near infrared and short-wave infrared bands.

A range of pasture products will be produced by the MDC on a regular 10-day cycle between April and November. The main products are the absolute pasture biomass, the pasture anomaly and the pasture trend. Pasture anomaly compares the current state of grazing pasture to the long-term average for that time of the year, allowing herders and decision makers the opportunity to judge whether the pasture conditions are currently better or worse than they usually are. The pasture trend compares the current pasture conditions to those from 10-days ago. This can help give an indication of whether pasture conditions are improving or worsening and can allow for herders to move their herds to regions where the pasture is improving, to give their animals a chance to put on weight and gain strength prior to the harsh Mongolian winters.

The pasture products will provide valuable information during the summer, however the MDC will also provide help for herders and decision makers in winter during which time snow is a critical issue, its depth and penetrability determining whether livestock is able to reach the pasture beneath it. The initial snow products will illustrate where snow is currently present, but also to help herders to see the trends developing in snowfall over the course of a winter. The basic snow product, derived straight from the NDSI will show where snow is currently present. Several more advanced products are also being developed, which will illustrate the length of time for which different regions have been covered by snow. This snow persistence product, which updates throughout the winter season will indicate the where the effects of the snow are being felt most severely.

One of the next key phases in the development of the pasture and snow products is to ensure they are successfully validated against ground truth data collected on the ground. The SIBELIUs project’s partners within Mongolia will be crucial to achieving this aim. The pasture and snow products will form key building blocks, from which several other more advanced products will be derived, including dzud risk maps, which will aim to give advanced warning of very severe environmental conditions posing a risk to livestock.

A pasture anomaly image from the Delgerkhaan province in the Khentii region, comparing the mean pasture value in the time period 9th – 19th August 2018 to the long-term pasture averages for that same time period. Green areas represent higher than average pasture, while red highlights where there is less pasture than typical.
A pasture trend image in the Delgerkhaan province of the Khentii region, comparing the mean pasture in the time period 9th – 19th August 2018 to the preceding time period 30th July – 9th August. Green areas indicate where the pasture is improving and red areas where the pasture is degrading compared to the previous time period. The white areas indicate that no meaningful data was available during this period, mainly as a consequence of cloud cover. In the future, the use of more satellite sensors will mean that cloud cover becomes less of a problem.
%d bloggers like this:
search previous next tag category expand menu location phone mail time cart zoom edit close