Skip to main content

Interpreting Env-DATA results and FAQ

Here's where your expertise comes in. The purpose of the Env-DATA System is to save you many time-consuming steps in linking tracking data to environmental conditions. We hope this makes it easier for you to test hypotheses and leaves you with more time and money for designing novel studies and collecting high-quality data. However, we're certainly not the experts for every species, study site, and environmental dataset, and so we leave users with the responsibility to put results in an interesting and realistic biological context. To get you started, we provide a few tips below, and leave the fun part to you!
 
When interpreting your results,
  • Remember that the annotated data are estimates that reflect the resolution and accuracy of the original datasets. Be sure to check the resolution of the original datasets and carefully consider what variability and trends are biologically meaningful. For example, weather conditions from global atmospheric models do not describe local conditions and so are only an approximation of what animals actually experienced. Also consider the accuracy of your tracking data—if the location estimates for your animals have an error of 1 km, annotating 30-m elevation estimates or categorical land use estimates in heterogeneous environments might not be useful.
  • Note that accuracy will also depend on the variability in your original datasets. For example, elevation estimates will typically be better over a flat plain than in a mountain range.
  • Check the documentation in the readme file to make sure you know the correct units for each attribute you have annotated.
  • If you plan to present or publish analysis done with your annotated data, be sure to check the use conditions for the datasets used (also available in the readme file) and give the data providers proper acknowledgement.
  • Note about Excel: If you use Excel to view your results, be aware that it can corrupt timestamps and truncate values in the annotation results. See here for instructions about how to avoid this problem if you do not have an alternative software option.

For more details about how the Env-DATA Track Annotation Service works, see Dodge et al. 2013.

The following are commonly asked questions about Env-DATA results.

Where can I find the units, citation, and other information about my results?

See the readme file that was included in the zip file with your results. Information about attributes, like units, is included under "File attributes", and information about products, like the resolution of the original dataset and terms of use, are provided under "Environmental data services".

Why are my results a file with a header but no data?

The most common reason to get results with just a header row is that the user selects to annotate data for animals but there are no data associated with those animals. The solution: Upload data, if you haven't already, and make sure that they are correctly linked to animals using deployments (read more about deployments here). You can verify what data are deployed in the study by selecting Download > Download Data and downloading the data without clicking on "include undeployed locations" and can compare deployments and available data in the Deployment Manager.

Why do I have lots of "NaN" values in my results?

Nodata values ("NaN") are due to missing information in the original datasets requested, for example due to cloud cover when a satellite was overhead. Reasons for many or all values being NaN include

  • The source file has periods or regions with nodata values, for example, if you request a vegetation index for a period when the animal was flying over the ocean, or sea surface temperature for a period when the animal was on land. For locations near the poles, remember that there are times of year when it will be nearly always dark and so there will be little or no satellite imagery.
  • The requested records fall outside the temporal or spatial range of the dataset: See our data products summary for the area and time covered by each product. For products that are available to the "present", there is a lag between today and when today's measurements are available as processed data files, so if you have live data feeds, the most recent few weeks will typically return NaNs for these products.

Solutions: See what obvious explanations there might be for your results as described above. You can also try requesting using a different interpolation method—as described in Env-DATA Interpolation Methods (look for "missing data"), nearest neighbour or inverse distance weighted interpolation will give a valid result even if some of the neighboring data values are missing, while bilinear interpolation will not. Also consider whether there are related variables you could request to compare or get more information, for example quality control variables for the product that you could request to learn more about specific data quality and reasons for missing data.

How do I calculate wind speed and direction from the u and v wind?

A good resource with instructions that is recommended by NARR's support team is here. See the section on "geographic wind direction" and if you are using a spreadsheet program like Excel, see their warning under "two argument arctangent function" to make sure the equations work correctly.

How can I get more help?

If you receive an email saying or request has failed, have not gotten results within a few days, or have questions about the results of your request not addressed above, please send the "access key" for the request to support@movebank.org so we can take a closer look. This is in the third paragraph of the readme file, which can also be found by going to Env-DATA > Show My Requests and clicking on Details for the request, as shown here:

Movebank study name: Golden Eagles in Alaska
Annotated Animal IDs: 12345
Requested on Tue Oct 27 20:00:33 CET 2015
Access key: 5485945721593819135
Requested by: user name