Skip to content. | Skip to navigation

Personal tools
You are here: GES DISC Home News “Data Rods” from the NASA GES DISC

“Data Rods” from the NASA GES DISC

Transforming satellite observations of Earth into temporal data for specific locations

“Data Rods” from the NASA GES DISC

"Data Rods" provide a way for hydrologists and meteorologists to easily use data acquired by NASA satellites.

“Data Rods” from the NASA GES DISC


Imagine: You’re gazing down on the surface of the Earth from a very high place, such as a high-altitude aircraft or a satellite, and you notice a remarkable cloud pattern. What you see might look something like this (Fig. 1):


Figure 1. Moderate-Resolution Imaging Spectroradiometer (MODIS) image of a strong storm system over the U.S. midwestern and southern states on September 26, 2011. An associated cold front (indicated by the curving band of clouds southeast of the center of the storm’s circulation over southern Wisconsin) brought rain to Mississippi, Alabama, Tennessee, Kentucky, Indiana, Ohio, and Michigan.  (Click on the image to view a much larger version.)

Such a view is very useful, if you are interested in seeing a large area in its entirety at one specific time.  Other types of “view” are possible, especially with different kinds of instruments  that sense the Earth at specific wavelengths of radiation; these measurements can be transformed into data about the Earth, as shown in Figure 2. Again, though, this is an image of a specific data type over a large area at one specific time.

Figure 2. Precipitation data from the Tropical Rainfall Measuring Mission (TRMM), acquired on September 26, 2011, showing the rainfall associated with the storm system and cold front observed by MODIS in Figure 1.  (Click for full-size version.)


On the ground, meteorologists and hydrologists (scientists who study water) usually collect their data in a different way. Meteorologists measure temperature and collect rainfall at weather stations and create time-series of these location-specific measurements. Hydrologists monitor the heights of water in streams at regular intervals, using stream gages at individual monitoring stations, thus generating a time-series of the stream height. Or, hydrologists might measure the amount of moisture in the soil at a specific location. While this measurement might determine the soil moisture at several different depths below the surface, it is still what scientists call point data; that is, data measured at a specific geographical point on the Earth surface.

So, the problem is, how can the way that NASA satellite instruments collect data – i.e., observing and measuring a large area at one specific time – or the way related land surface models output data be linked to the way that hydrologists collect data, i.e., acquiring measurements at discrete points on the ground for many consecutive time intervals, which creates time-series of data for those specific locations?

One solution to this disparity between how NASA stores data (time-step arrays or maps) and how the hydrology and similar point-time-series-oriented communities prefer to access NASA data is the creation of data sets composed of data rods. Data rods are generated from satellite observational data files, by first “stacking” them to construct a sequential set of data files for the same area and then extracting the data for a specific location out of the files, resulting in a time-series of a specified data variable for that location. Data rods for many different locations can thus be created from the same stack of satellite observations. 

One way to visualize the process of generating data rods is to think of each satellite observational data file (or each model output data file) as a pancake. Stacking the pancakes on top of each other (a common way to serve them) creates the time-series of pancakes (oldest at the bottom). Now you can cut out a mini-stack of pancakes with a knife, leaving a hole behind (Fig. 3), resulting in a “pancake rod” of a specific location (~center of the pancake). (Incidentally, pouring syrup into the hole is a great way to eat pancakes with syrup.)

Figure 4 provides a schematic diagram of how to create “data rods” with actual satellite (or model) data!

How to make a "pancake rod"

Figure 3. How to make a "pancake rod."  (Top left) Stack of pancakes. (Top right) A "pancake rod" extracted from the stack, sampling the same location in each pancake.  Extracting the "pancake rod" leaves a hole for syrup.  (Bottom left) Closeup of the "pancake rod." (Bottom right) Stack of pancakes with syrup added in the middle and buttery spread.  Similar to the pancakes, satellite observations can be assembled to form a stack of data files ordered in time, from which “data rods” of specific locations can be extracted. 
Reorganizing Data for Use by Applications
Figure 4. Schematic diagram of how data rods are created from a set of stacked satellite observational data files.  (Click on the image to view a larger version.)

The NASA Goddard Earth Sciences Data and Information Services Center (GES DISC) has just released a Web page describing the data rods available in the GES DISC data archives. 


These GES DISC data rods were created in a project supported by the NASA ACCESS Program. This data rods project had two primary goals: (1) to reorganize two large hydrological data sets as time-series for more efficient access by the Consortium of Universities for the Advancement of Hydrologic Science, Inc. (CUAHSI) and the larger hydrology community and (2) to integrate these time-series data, in the form of  data rods, into hydrology community tools, such as the CUAHSI Hydrologic Information System (CUAHSI HIS) and the EPA BASINS (Better Assessment Science Integrating point & Non-point Sources). These data rods contain data for various water cycle variables obtained from the North American Land Data Assimilation System (NLDAS) and the Global Land Data Assimilation System (GLDAS). 

Figure 5 shows a plot of an example data rod for surface runoff, for the location of Goddard Space Flight Center in Greenbelt, Maryland. (Surface runoff is the amount of water that runs off the surface of the Earth into streams and rivers following a rain shower or storm.) Data rods for NLDAS and GLDAS data from the GES DISC respectively cover the time periods of 1979-present and 2000-present. Note that, by specifying the start date and the end date (startDate and endDate), a data rod for a specific time period can be extracted. Figure 6 shows the same data as those for Figure 5 but only for the period 1996-1999, when surface runoff was elevated in the region.

Time series of surface runoff for GSFC lat/lon coordinates

Figure 5. Time-series for NLDAS surface runoff, 1979-present, for the geographical location of Goddard Space Flight Center in Greenbelt, Maryland. During the mid-1990s, surface runoff was elevated by several events.  (Click on the plot to see the original larger version of this time-series plot.)
Time series of surface runoff for GSFC lat/lon coordinates, 1996-1999
Figure 6. Time-series for NLDAS surface runoff, 1996-1999, for the geographical location of Goddard Space Flight Center in Greenbelt, Maryland. Two of the largest events generating surface runoff occurred in early winter (December 1996) and late autumn (November 1997). In November 1997, about three times more rain than average for the month fell in the Washington D.C. region, caused by a slow moving low-pressure system off the southeast U.S. coast. (Click on the plot to see the original larger version of this time-series plot.)
The tables on the Data Rods (Time Series Data) page provide information on the elements that are used to construct a URL request for the Data Rods Web services.
How the URL is constructed for the plot in Figure 6 (hourly time-series data of NLDAS surface runoff) is described below
Here is the complete URL:, 39.0)&startDate=1996-01-01T00&endDate=1999-12-31T23&type=plot
The first section of the URL is where the data are stored at the GES DISC:
timeseries.cgi” invokes the time-series (data rods) retrieval function.
The next section specifies the data set and the surface runoff variable.
The location can be specified with either NLDAS or GLDAS grid IDs or with latitude and longitude, as shown below. Note that the longitude value comes before the latitude value.  (Negative longitudes indicate the Western Hemisphere, and negative latitudes indicate the Southern Hemisphere.)
&location=GEOM:POINT(-76.84, 39.0)
The next section specifies the starting date and ending date. Because the data are hourly, the first hour and the last hour are specified as well, by the two digits following the “T” (00 to 23).
The last section of the URL requests a plot of the data.
The only change needed to request ASCII data instead of a plot is as follows:
Rui, H., R. Strub, W.L. Teng, B. Vollmer, D.M. Mocko, D.R. Maidment, and T.L. Whiteaker, 2013. Enhancing access to and use of NASA earth sciences data via CUAHSI-HIS (Hydrologic Information System) and other hydrologic community tools, 2013 AGU Fall Meeting, San Francisco. (PDF)
Rui, H., B. Teng, R. Strub, and B. Vollmer, 2012. Data reorganization for optimal time series data access, analysis, and visualization, 2012 AGU Fall Meeting, San Francisco. (PDF)
Teng, W., H. Rui, R. Strub, and B. Vollmer, 2015. Optimal reorganization of NASA earth science data for enhanced accessibility and usability for the hydrology community, Journal of American Water Resources Association, in review.

We would like to thank the other members of the “data rods” project team for contributing their parts of the project, to which the data rods are connected: David Maidment, Tim Whiteaker, David Arctur, and Gonzalo Espinoza Davalos of the University of Texas at Austin; and Christa Peters-Lidard, David Mocko, and Dalia Kirschbaum of the NASA GSFC Hydrological Sciences Laboratory.


Questions or comments? Email the NASA GES DISC Help




Document Actions
NASA Logo -
NASA Privacy Policy and Important Notices
Last updated: Aug 11, 2015 11:33 AM ET