Gaining a Clearer Picture of Phenology and Snow Depth Dynamics with NEON Phenocam Data

March 31, 2016

By Leslie Goldman

In a collaboration with the PhenoCam Network, NEON now streams phenocam image data to the Phenocam Gallery - including tower top images and ground level images.

PhenoCam is one of the largest cooperative, open access networks in the world of phenocam data. Phenocam both archives and distributes imagery and derived data products from digital cameras deployed at research sites across North America and around the world. Data are available via the Phenocam Gallery, a robust and open access database that provides image data of canopy phenology. You can also search and access the data via the NEON Data Portal. Phenocam image data from NEON field sites are hosted on the Phenocam Gallery in addition to similar datasets from other phenocam sites across the country allowing users to easily search and work with NEON’s data and other phenology image data.

Collecting NEON phenology images

NEON collects several types of phenology image data including the following data products:

Phenology images (NEON.DP1.00033): RGB and IR images of the plant canopy taken from an automated camera on the tower top of NEON flux towers. Towers rise above the plant canopy at each terrestrial field site to allow sensors to collect a full profile of atmospheric conditions. The height of the tower also provides a perfect view of the surrounding plant canopy.

Snow depth and understory phenology images (NEON.DP1.00042): Camera images of snow depth relative to mounted/calibrated depth stakes when snow is present; images used to track understory phenology when possible. These images are captured by a second camera that is mounted mid-way down the tower.

Land-water interface images (NEON.DP1.20002): RGB and IR images of the lake, river, or stream vegetation, stream surface, and stream gauge (where possible) taken from an automated camera. Images are collected every 15 minutes.

Images are collected every 15 minutes, throughout the day and night. In addition to the familiar RGB images, our cameras also collect IR imagery. NEON protocols closely follow protocols of the Phenocam Network.

Gaining a clearer picture of phenology and snow depth dynamics

NEON phenocams will collect multi-year time series image data from each NEON field site creating a wealth of data from many different ecosystems. Over time, these data will reveal trends in the greening of trees, herbs and shrubs at NEON sites. Additional data collected near the phenocams such as atmospheric data, organismal data, soil data and remote sensing data from the NEON AOP will provide a wealth of data for ecologists to better understand ecological change over time.

NEON scientists also place several snow gauges, appearing as yard sticks sticking out of the ground, within the field of view (FOV) of the ground-based cameras at terrestrial sites. These snow gauges are placed in the FOV so that scientists can estimate snow depth over the non-growing season. Understanding the dynamics of snow cover are important because snow cover and snow melt have an important influence on things like soil temperature, gas exchange between the soil and the atmosphere and nutrient cycling.

Innovation in image processing to automate snow depth estimation

Snow depth measurements can be gleaned from the ground-based phenocam images manually by opening each image and looking at the lengths of the snow gauges not covered by snow.  But doing this manually for a place where winter lasts five months would entail a lot of manual labor, as five months generates roughly 14 thousand images for a single site! As an alternative, NEON staff scientists have been experimenting with image processing techniques that automate snow depth estimation. The procedures are still preliminary, but prototype code has performed fairly well. 

At the SERC site, for example, the code has been used to produce time series showing snow cover and snow depth before, during and after snow events (Figure below).

The automated estimates look pretty accurate when compared to those generated manually. And it only takes about a minute to process roughly 1000 images.

Snow depth python script

This experimental code is written in the Python programming language and needs a fair amount of fine tuning. It's a prototype script that can be used with NEON phenocam data to automate snow depth estimation. The procedures are still preliminary, but the code has performed fairly well. Feel free to download it, play with it, and, of course, improve it!


What types of scientists might be interested in these data?

According to Andrew Richardson, principal investigator of the Phenocam Network, there are a lot of potential applications for phenocam data. “Phenology data derived from camera imagery tell us a lot about the seasonality of ecosystem-atmosphere CO2 fluxes and ecosystem productivity. The data can be used to test and improve models of vegetation phenology. Ecologists can quantify the differences in phenology among competing species. The remote sensing community can use the data to validate satellite data products. Earth system modelers can use the data to benchmark land surface models.”

Accessing NEON phenocam data

To explore NEON phenocam data and other phenocam data images, go to the PhenoCam Gallery, and scroll through the core sites for sites beginning with NEON. To explore other NEON phenology data products, visit the NEON Data Portal.

Dialog content.