With all new technology, questions abound - so we've provided answers to the most common ones here.

What does GAIA stand for?

Geospatial and a little extra

The name was inspired by the Greek deity, Gaia. According to Wikipedia, she “is the personification of the Earth and one of the Greek primordial deities. Gaia is the ancestral mother of all life: the primal Mother Earth goddess.”

GAIA is also an acronym descriptive of the software: “Geospatial Artificial Intelligence for Agriculture”. The term ‘geospatial’ refers to geospatial analysis, which is an approach to applying statistical analysis and other analytical techniques to data which has a geographical or spatial aspect.

How Do I Interpret GAIA NDVI Imagery?

Dr Eriita Jones explains

The Normalized Difference Vegetation Index (or NDVI) is a function of the brightness of the surface at red and near-infrared wavelengths of light.

NDVI is sensitive to the condition and health of the vine and correlates strongly with many biophysical parameters such as canopy density, fruit yield, wine grade/quality, nutrient deficiency, and others.

High values of NDVI, shown here in blue colours, indicate healthier plant condition, greater vigor, and higher vegetation density (leaf and canopy area). Low values of NDVI, shown in reds and oranges, are indicative of plant stress (such as water stress), or lowered vegetation density. They may also highlight reduced vegetation health, due for example to the presence of disease or nutrient deficiency. Moderate values of NDVI are shown in greens and yellows.

GAIA web app screenshot
Screenshot taken from the GAIA app on 10 Nov 2018

Can GAIA monitor plant health?

Yes, specifically disease detection

Viral and fungal trunk diseases can and have been successfully detected using multispectral satellite imagery.

Additionally, detection has been achieved for a variety of diseases before symptoms are visible to the naked eye.

It has been demonstrated that vines infected with other viral and fungal diseases can be detected via multispectral remote sensing (e.g. grapevine leafroll disease GLD, grapevine leaf stripe disease GLSD, Flavescence dorée FD, and downy mildew Peronospora). Incidence of disease and stress results in spectral changes to the vegetation which can be detected at visual and near infrared wavelengths with a high degree of accuracy (e.g. one study reports >90% accuracy in remote sensing detection of GLD). These vegetation changes may be detected even before the foliar symptoms are observable. The vegetation index NDVI is sensitive to the presence of both current foliar symptoms (e.g. leaf stripes) and asymptomatic vegetation which showed symptoms in previous years. NDVI can also be used to map disease severity (ie. severity of foliar symptoms) and progression (demonstrated for GLSD). Other vegetation indices may provide enhanced sensitivity to the presence of grapevine infection compared to NDVI (demonstrated for FD). Therefore given similar studies, VNIR remote sensing can likely be utilised for detecting and mapping the incidence of Eutypa infection. However, research is needed to assess the full capabilities of remote sensing for Eutypa early detection and symptom tracking.

Can GAIA assess soil moisture?

Yes, in varying degrees

Mapped regions of relative high vs low moisture can be assessed without need for any field sampling at shallow root zone (±20cm depth).

When combined with modest field sampling, more area-specific soil metrics can be estimated.

The soil moisture available to the grapevine is a controlling factor in vine growth, root production, and vine productivity (fruit yield). Insufficient availability of soil water can cause water stress to the grapevine. Knowledge of the soil moisture informs water management practices (e.g. irrigation or drainage). Although soil moisture fluctuates with time, the patterns in soil moisture within a given vineyard block will be generally consistent and dependent on: the soil characteristics – such as soil type/ mineralogy (e.g. sandy, clay, silt) and soil texture; terrain attributes (e.g. slope and aspect); and climate. Soil moisture is typically measured through ground probes, which provide information only on the region in closest proximity to the probe. Airborne or ground-based ground penetrating radar (GPR) can provide both surface and deeper subsurface assessment of soil moisture, however the temporal frequency of such GPR measurements is limited, and the data processing required is non-trivial.

Multispectral remote sensing at VNIR and thermal wavelengths can provide spatially extensive mappings of soil moisture and its variability across vineyard blocks, with high temporal frequency. A number of spectral indices exist for accurate measurement of soil moisture content of both bare soil, and sparsely to partially vegetated surfaces (e.g. interrow with grass or covercrop), from optical and thermal satellite data. These include: soil wetness index (SWI; uses VNIR and thermal wavelengths), temperature vegetation dryness index (TVDI; VNIR and thermal), tasseled cap transformation wetness (TCTW; VNIR wavelengths), normalized difference water/wetness index (NDWI; VNIR wavelengths), and normalized multi-band drought index (NMDI; utilises SWIR wavelengths). The reliability of soil moisture retrieval from these indices is likely limited to the top ~20cm of the subsurface. Limited ground-based measurements of soil moisture would be required to establish the numerical relationship between spectral indices and volumetric soil moisture. Although these indices have been shown to provide accurate soil moisture estimates in a number of settings, their specific applicability to measuring vineyard interrow soil moisture is yet to be investigated.

Nighttime thermal inertia, derived from thermal infrared measurements, has been shown to have a moderate correlation with inter-vinerow soil moisture, even in the presence of grass cover. Field measurements can be employed to derive the quantitative relationship between remotely sensed thermal inertia and volumetric soil moisture, or the thermal data can be used qualitatively to delineate zones of similar soil characteristics.

Can GAIA gauge fruit quality?

Yes, to a degree

GAIA can distinguish areas of good quality from areas of poor quality, based on satellite imagery alone. This can be used to inform split-picking schedules, and other critical decisions.

Over time, as GAIA continually learns and becomes smarter, we expect her to offer more predictive insight into the grade of fruit for specific areas.

Grape – and subsequently wine – quality is a complex function of a number of grape attributes, including sugar content, acidity, anthocyanin concentration and water content. These biochemical factors can be quantified through handheld spectrophotometers, however quality is frequently assessed through subjective expert tastings. Fruit quality directly impacts the value of the wine. The accumulation of sugars and flavour compounds is predominately light limited and so is directly related to the amount of shading experienced by the fruit, and hence correlates with measures of vine canopy coverage and density. Grape quality is also impacted by management practices (inputs such as fertilizers and other chemicals, vegetation cropping and canopy management, etc.), weather patterns and microclimates, soil characteristics, terrain, and any factors which alter the physiological characteristics of the vine. A single vineyard block can show significant variation in grape quality, with intra-block variation of more than 10% in quality parameters observed (e.g. Baumé and phenolics) and fruit spanning several quality grades. “Uplift” is the process of increasing the quality (grade) of grapes to increase the supply of higher retail price wines (e.g. increasing from C & D grade fruit to A & B grade).

The character and quality assessment of grapes is correlated with the Normalized Difference Vegetation Index (NDVI) and Ratio Vegetation Index (RVI or plant cell density PCD). Higher quality fruit originates from areas of lower values of the vegetation indices (VIs). The qualitative relationship between VIs and quality can be used to delineate relative quality boundaries (ie. high, moderate, low) for each vineyard block. These boundaries would be expected to be relatively temporally stable and can be utilised for targeted harvesting “split picking” across seasons.

The relationship between grape composition and vine physiology is complex. Canopy coverage and density can be remotely sensed, however manipulation of the vine canopy (e.g. vine balance) has no consistent significant impact on fruit quality parameters (anthocyanin, phenolic or tannin concentration) and is not a reliable mechanism for “uplift”. Therefore there is unlikely to be a consistent quantitative relationship between singular measures of vine vigour (such as VIs) and grape composition and wine quality. Other attributes of the vineyard site (terroir) are more significant in determining fruit quality, and these may be captured through multispectral remote sensing. A more complex multivariate measure of: the vegetation canopy, combining measures of vigor and biomass (e.g. leaf area); the leaf biochemical composition (e.g. water, chlorophyll, protein and lignin content); and the underlying soil or covercrop background can be derived from multispectral VNIR and SWIR satellite imagery. Such a measure has the capability of being able to quantitatively discriminate crops of different grape quality by providing sensitivity to both the supply of energy and organic carbon to the vine, and the various other aspects of terroir that influence grape quality. Once established, a quantitative relationship could be used to systematically inform “uplift” practices.

Want to Know More?

Get in touch to find out how GAIA can improve the decisions of your agricultural organisation.