Our goal is to provide a generic measure of plant growth which allows comparison between farms growing different species of crops, in different places around Australia, over many years. With such a measure, we can enable quantitative assessment and benchmarking of the productive value of agricultural land in different regions, and of the way land has improved or degraded over time in its productive capacity.
As a first step to doing this, in collaboration with CSIRO, we've developed a standardised measure of plant productivity based on NPP (net primary productivity), which uses NDVI (normalised difference vegetation index) and other measurements to produce a more accurate picture of land potential. In this article I’ll cover what this measure is, how it’s different to NDVI, the ways that this measure can currently be used, and our future plan for this approach.
What is NPP?
Plants absorb carbon through photosynthesis - the rate at which they do this is called gross primary productivity (GPP). The plant then burns some of this carbon off as CO2, through the process of respiration. The net remaining rate of carbon accumulation is called net primary productivity (NPP). This is the amount of carbon available for building new plant mass, and constitutes approximately half of plant biomass growth, with the other half being all the other elements in plants, like nitrogen, phosphorous and potassium.
Because NPP describes the rate of plant growth, it is also strongly correlated with both crop yield and pasture production (though we don’t currently recommend using NPP directly to predict crop yield).
Our approach
To measure NPP, we start with remotely sensed data about a particular location, e.g. greenness (NDVI), air temperature and elevation. We then pass this data into a biophysical model which estimates carbon turn-over and accumulation based on the physiology of grasses - a category which includes, among others, all cereal crops and most common pastures. The model then generates estimates of the NPP, or the rate of growth, of every satellite observed area, measured in grams of carbon per square metre per day.
This model provides output every 16 days at a 250 metre resolution, across all of Australia, from 2001 to 2017. Validation of this model has shown (despite the difficulty of measuring NPP directly) that it is highly accurate for a tropical pasture in NT, a savannah grassy woodland in NT, a tussock grassland in QLD and a modified temperate grassland in VIC.
Why NPP instead of NDVI?
When estimating productivity, many organisations use NDVI (normalised difference vegetation index) by itself. While NDVI is a decent measure of plant health, it does not give the full picture of what's going on on a farm.
NDVI isn’t directly measuring growth. NDVI is most often used as a measure of vegetation health. However, by itself, it is not always a good indicator of how much plants are actually growing, as it does not take into account plant life-cycle and biology. For example, NDVI values in mature forests will be high, as there is a lot of green leaf cover, despite the fact that these plants are not growing fast. At the other end of the spectrum, NDVI values for a newly planted crop will be low, as there is little leaf cover, despite the fact that these plants are growing rapidly.
NPP estimates use NDVI. NDVI is a primary measure used in the estimation of NPP, which is a more reliable measure for plant growth, taking into account plant physiology and other environmental conditions.
How to use this data
Note: NPP can be negative. Most locations will experience negative NPP values at some time in the year, when respiration rates are higher than photosynthetic rates, such as occurs over winter in the temperate zone and in the dry season in the tropics.
To access productivity data for a given farm in the DAS platform, click that farm's outline on the map, then click the Production tab. Note that all measurements are in units of kg biomass per hectare per day.
Long term average productivity. This is the mean productivity over the timeframe selected (see Timeframe below).
Area used for agriculture. The area on this farm which is used for agriculture, according to our land use data.
Timeframe. Use this to select the timeframe over which you'd like to view the data. All values will automatically update when a new timeframe is selected.
Chart. The chart shows the productivity of this farm over the selected time frame. You can click or hover over the line to view the productivity values at different points in time.
Rather than provide a point-in-time measure of plant health, our measure of plant productivity is aimed at assessing and comparing overall property productivity. Given the parameters of the model (16 day time step, 250 meter resolution, continental coverage), we recommend using this data for the following things:
Assessing the long term trend in productivity increase / decline on a farm.
Comparing productivity between farms growing similar crop compositions (see “What’s next” section below).
We don’t recommend using this data for:
Assessing productivity on farms smaller than 10 hectares.
Trying to directly estimate crop yield.
Comparing productivity between farms growing dissimilar crop compositions (see “What’s next” section below).
What’s next?
Our next step in making this data more useful will be introducing productivity benchmarking - taking sets of similar farms in terms of soil, rainfall, crop compositions and elevation, and ranking them against each other in order to provide objective scores for agricultural land, both in terms of current productivity and long term sustainability.