Aqualinc delivered a comprehensive update to the National Irrigated Land Spatial Dataset (originally created by Aqualinc in 2017), providing critical information for effective water resource management and environmental reporting in New Zealand.

Knowing where and how land is irrigated is a key input for many projects, including catchment-scale modelling, water allocation and economic studies.

Aqualinc was commissioned by the Ministry for the Environment to update the National Irrigated Land Spatial Dataset for the 2019-2020 irrigation season. This project involved integrating multiple spatial data sources and applying expert judgement to map areas equipped for irrigation, categorised by system type, across mainland New Zealand. The updated dataset provides a crucial national overview of irrigated land, building upon the initial dataset created by Aqualinc in 2017.

The primary objective of this project was to create a comprehensive and up-to-date spatial dataset of irrigated land across all mainland regions of New Zealand. To achieve this, Aqualinc employed a multi-faceted methodology, integrating several key spatial data sources. High-resolution aerial photographs and satellite imagery were utilized to visually identify irrigation systems such as centre pivots, lateral move irrigators, and other visible equipment. The interpretation of these images allowed for the determination of irrigation system types based on visual cues like wheel tracks and irrigation patterns.

In addition to aerial imagery, multi-spectral satellite imagery from Sentinel-2 was a crucial input. This imagery was used to calculate the Normalised Difference Vegetation Index (NDVI), which highlights the contrast between actively growing vegetation and drier areas, indicating potential irrigation. Due to drier than average conditions in early 2020 across much of New Zealand, this period offered an opportune time for NDVI analysis to identify irrigated land. For Southland, where early 2020 was unusually wet, 2018 data was used instead. The spatial resolution of the Sentinel-2 imagery (10 m) represented an improvement over the Landsat data used in the 2017 irrigated land dataset.

Resource consents data helped identify farms and areas with legal authorization to take water for irrigation, particularly when combined with land slope data to identify potentially irrigable areas. Land parcel boundaries were used to define farm extents, considering ownership information to account for irrigation across multiple parcels under the same ownership.

The update process built upon the 2017 irrigated land dataset. Areas were prioritized for review based on several factors, including identifying potential new irrigation outside the 2017 boundaries using NDVI and imagery, identifying irrigation associated with water take consents granted more recently, and noting any changes in irrigation system types (such as conversion from border-dyke to centre-pivot). Region-specific datasets, such as a vineyards layer from Marlborough District Council and a field-verified irrigated area layer for the Manuherikia catchment in Otago, were also incorporated.

Qualitative confidence levels (high, medium, low) were assigned to each newly mapped area, and quantitative uncertainty estimates were applied based on the irrigation system type and the regional NDVI contrast (high, moderate, low). Regions with high NDVI contrast (Canterbury, Marlborough, Otago) generally allowed for more accurate mapping.

The 2020 update to the National Irrigated Land Spatial Dataset reveals a total mapped irrigated area of 903,465 ha, representing a 14% increase from the 2017 dataset. While some of this growth is due to new irrigation developments, it is also attributed to improved availability and resolution of aerial and satellite imagery and the dry conditions in early 2020 which facilitated the identification of irrigated land.

Canterbury, Marlborough, and Otago collectively account for 77% of the total mapped irrigated area, with Canterbury alone comprising 61%. The most prevalent irrigation system type is centre-pivot, making up 39% of the total irrigated area.

The total mapped area in 2020 is 21% greater than the 2017 Agricultural Production Statistics (APS) data, though direct comparison is limited by differing methodologies and time lags. Several regions, including Hawkes Bay, Gisborne, Northland, and Tasman, showed large percentage increases from the 2017 mapping, but their totals did not closely align with the APS data, suggesting a combination of actual increases, under-reporting in 2017, and identification uncertainties.

The spatial dataset and technical report are available from MfE’s data service.