Galvez-Davison Index (GDI) as another tool to forecast convection in North America
June 13th, 2025 by José Manuel Gálvez
The Gálvez-Davison Index (GDI) (Gálvez and Davison, 2016) was developed by José Gálvez and Mike Davison at NOAA’s Weather Prediction Center in 2014, to improve the detection of environments favorable for tropical convection [Online access: https://www.wpc.ncep.noaa.gov/international/gdi/ ]. However, its skill does not only limit to tropical locations. It extends into mid-latitudes during the warm season. This means that it can be used in the United States during the summer, particularly in regions where warm and moist air masses develop. This short blog uses satellite imagery to show a simple example of (1) how the GDI has skill in parts the United States during the warm season and can be used as a forecasting tool and (2) the importance of incorporating an analysis of atmospheric dynamics to better assess where convection might form or continue.
Like most stability indices, the GDI is not a standalone index and should be analyzed in combination with atmospheric dynamics. The GDI only describes the static environment where convection might form, given that its calculation does not consider wind information. Consequently, it only provides information about whether or not atmosphere is capable of hosting convection and rainfall, but it does not really tell the forecaster whether there convection triggering mechanisms will be present. This is where an analysis of flow dynamics will enhance convection forecasts.
Let’s start with a quick description of the Galvez-Davison Index (Figure 1). The GDI is computed using temperatures and mixing ratios from 4 atmospheric levels: 950, 850, 700 and 500 hPa. It represents three processes that are summarized in three sub indices: (1) The CBI describes the availability of heat and moisture in the 950-500 hPa column, (2) the MWI incorporates the impacts of stabilization by warm air masses in the mid-troposphere and (3) the II represents the drying and stabilizing impacts of thermal inversions in the lower and low to mid troposphere.
How to interpret GDI values? A general interpretation diagram is available in Figure 2. The GDI is dimensionless. Low values relate to shallow convection and very little precipitation, while higher values suggest a higher potential for deep convection. In the color scale we often use, the chances for deep convection generally increase when greens and yellows appear (GDI=15 to 35), and once oranges and reds appear (GDI>35), the potential for deep convection increases. A potential for heavy rain producing convection appears as well when GDI approaches 35, and increases with higher values.
For a quick look into the skill of the GDI in North America, a random day when GDI was high in the United States and convection was present was chosen: 12 June 2025. Figure 3 summarizes the GDI field circa 18 UTC. This is close to midday or the early afternoon in the United States. We are using GDI calculated with GFS model data, 18 hour forecast, using the software Wingridds. According to the figure, deep convection is likely to develop in points A, B and D; and to a minor degree in point C. A higher chance for deep convection appears in region E, in south-central Mexico, where GDI values are high.
Now let’s look at where the convection was actually forming around 18 UTC using satellite products. We will use an animation of the 10.3 um channel or one of the longwave IR channels available in GOES-19 presented in Figure 4.
Figure 4. Animation showing the 10.3 um long wave IR band of GOES-19. Source: CIRA Slider.
Thanks to the satellite loop, we can verify that the GDI seems to be capturing the development of deep convection in points A and D. Using this band and the chosen color palette, we can generally interpret the presence of deep convection by oval-shaped areas colored green, yellow, orange and red that are generally growing. These colors represent cold cloud tops, which are present in deep convective cells. The strongest system appears to be located in southeast Texas in point A, while in point D, the satellite signature suggests diurnal convection already developing in Florida. But what is happening in the other points? And what about the deep convection forming in the Rocky mountains, west of point C, while nothing is occurring at point C itself where the GDI is higher? Let’s start answering this by a quick analysis of atmospheric dynamics.
Figure 5 is very similar to Figure 3, but contains additional model information to interpret atmospheric dynamics.
Incorporating an analysis of the structure of the upper flow (yellow streamlines and wind barbs) as well as the low-level flow (black streamlines and barbs) provides an important amount of additional information. It shows that the convection in point A is also being stimulated by a robust upper trough extending from eastern Nebraska across southwest Oklahoma into south-central Texas. Embedded in this large upper trough, a negatively tilted short wave trough structure (light blue dashed line) is propagating northeastwards across east Texas. The term “negatively tilted” trough is generally used to describe a trough that has its warm tier ahead or downstream of its cold tier. This configuration tends to associate with enhanced upper divergence (white contours) and enhances ascent in the column. In the low-levels, the upper trough has induced a surface trough and it associates with enhanced moisture convergence (red contour). Thus according to the model, there is important dynamical forcing in point A. Satellite imagery in Figure 4 supports this, although precise location of the strongest convection does not match what the model suggests with high precision. This partly relates to complex mesoscale structures associated with the mesoscale convective system in southeast Texas. But what is happening in the other points? Let’s dive into the Day Cloud Phase Distinction RGB (Figure 6) to explore convective initiation.
Figure 6. Animation showing the Day Cloud Phase Distinction RGB (JMA color table) available from GOES-19 data. It also contains CIRA’s GLM Group Energy Density product overlaid, to detect thunderstorms. Source: CIRA Slider.
The Day Cloud Phase Distinction RGB is a great tool to detect convective initiation, or the beginning stages of clouds growing vertically into mid and upper portions of the troposphere. Clouds that contain water droplets look cyan in this product. But when convection initiates and clouds grow, they reach higher altitudes where temperatures are colder. As a result, they develop ice. In this RGB, thick ice clouds appear yellow. The animation shows that in points B, E and F, convection is only starting to form. This suggests that the diurnal cycle in these locations could matter. In point C, however, atmospheric dynamics seem to be tampering the impacts of high GDI values. Looking back into the flow analysis presented in Figure 5, region C (eastern Colorado and west Kansas) lies under an upper ridge and in a region where low-level moisture convergence is not enhanced. Low-level divergence could be even present, but it is not plotted. To the west of point C, however, deep convection is rapidly forming along the Rocky Mountains. Figure 5 suggests that this could be responding to enhanced low-level convergence (red contours) and an approaching short wave upper trough marked with a yellow dashed line.
Since we would like to evaluate the role of the diurnal cycle, lets look at the evolution of convection later in the day using a similar product, but covering the 19-23 UTC period (Figure 7).
Figure 7. Similar to Figure 6, but covering the 19-23 UTC period. Source: CIRA Slider.
Once we look at the evolution past 19 UTC, we verify that – indeed – the diurnal cycle of solar heating seems to play a role as a convection trigger. Rising thermals enhanced by solar heating past 19 UTC have the ability to break the cap or convective inhibition region in the lower troposphere, reaching altitudes where they acquire the ability to use high GDI to produce convection. This is especially true in mountainous regions such as Mexico and the Rocky mountains (points G and H), given the enhanced role of diurnal heating in elevated terrain and continental regions. The western location of these mountain ranges also plays a role in the later occurrence of convective initiation. But evaluating the dynamics also matters. In points G and H, added only in Figure 7, enhanced low-level moisture convergence associated with moist diurnal upslope breezes seems to be important. In point B, convection developed in a scattered to isolated manner. Looking at the GDI alone could had been a little misleading. But when considering atmospheric dynamics, the basic analysis that Figure 5 allows suggests that there might be a general lack of significant dynamical forcing. Yet, the confluence of low-level flow in eastern Tennessee apparently resulted in a region of enhanced low-level moisture convergence producing the cluster of scattered thunderstorms in the region.
Note on the Day Cloud Phase Distinction (JMA) Product: A final aspect of the Day Cloud Phase Distinction Product, not related to the GDI, is the change of coloration as the sun sets. The reason is that this product uses reflective bands that depend on solar radiation. However, areas with cold cloud tops are visible throughout the night, but acquire a red coloration. This is caused by the 10.3 um band being included in the red component of the RGB. During nighttime, the green and blue components of the RGB disappear due to the lack of solar radiation. Only cold cloud tops are visible at night, and they look red.
Figure 8. Recipe to code the Day Cloud Phase Distinction RGB. Source: CIRA.
Thank you for reading.