National Mapping of New Zealand Pasture Productivity Using Temporal Sentinel-2 Data

Alexander C Amies,Stella E Belliss,John R Dymond,Coby Hoogendoorn,James D Shepherd,David Pairman,Marmar Sabetizade

doi:10.3390/rs13081481

Alexander C Amies, Stella E Belliss + Show 5 more

Open Access

https://doi.org/10.3390/rs13081481

Copy DOI

Abstract

A national map of pasture productivity, in terms of mass of dry matter yield per unit area and time, enables evaluation of regional and local land-use suitability. Difficulty in measuring this quantity at scale directed this research, which utilises four years of Sentinel-2 satellite imagery and collected pasture yield measurements to develop a model of pasture productivity. The model uses a Normalised Difference Vegetation Index (NDVI), with spatio-temporal segmentation and averaging, to estimate mean annual pasture productivity across all of New Zealand’s grasslands with a standard error of prediction of 2.2 t/ha/y. Regional aggregates of pasture yield demonstrate expected spatial variations. The pasture productivity map may be used to classify grasslands objectively into stratified levels of production on a national scale. Due to its ability to highlight areas of land use intensification suitability, the national map of pasture productivity is of value to landowners, land users, and environmental scientists.

Highlights

Van HoolstNew Zealand agricultural land, which historically has primarily been used for sheep grazing, has seen increased utilisation for dairy in recent decades
To generate a map of pasture productivity from this model, a national map of median generate a map of pasture productivity from Zealand this model, a national map ofmedian median
Median Normalised Difference Vegetation Index (NDVI) in the corresponding range of the input data could be used in the model to estimate pasture yield

Summary

Introduction

New Zealand agricultural land, which historically has primarily been used for sheep grazing, has seen increased utilisation for dairy in recent decades. The land area of dairy farming increased by 33% between 1999 and 2019 [1,2]. Land use intensification in dairy farming areas has occurred, with an increase from 1.9 to 2.8 cows per hectare between 1991 and 2019 [2,3]. There have been regional patterns to New Zealand dairy conversions, typically based on the suitability of land for cattle grazing. One metric of interest in the prediction of land suitability for high-intensity farming is pasture productivity, or pasture yield, expressed in units of tonnes per hectare per year (t/ha/y).

Methods

Results

Discussion

Conclusion