In the Seeing Signal Through the Noise: Satellite Data Time Series for Agriculture webinar, we shared how multi-temporal data and statistical analysis is used to extract patterns and trends for monitoring crop growth and detect key events throughout the season. Watch the webinar on-demand now.
Below are answers to questions from the live event and resources to get you started.
Question: Can you please explain the relation of this and the hydraulics systems field?
Answer: In the context of hydraulics and water management, this data is often used to monitor irrigation efficiency and water distribution. By using Planetary Variables (like Soil Water Content) or vegetation indices, operators can identify leaks in hydraulic infrastructure (where vegetation is unexpectedly lush due to leaking water) or detect blockages where downstream crops are stressed. The daily cadence of PlanetScope is particularly useful here, as it allows for the detection of system failures in near real-time, rather than waiting for monthly composite images.
Question: Which is better when tracking loss or destruction of a Crop Biomass or NDVI?
Answer: It mainly depends on the issue causing the loss. With optical data, we are looking at reflectance; if the loss doesn’t immediately affect "greenness" (e.g., lodging, where the crop falls over, but stays green), NDVI might miss it. In this case, Crop Biomass (a Planetary Variable) is often a better option because it provides a more robust measure of the crop’s architecture and density, overcoming the saturation issues common with NDVI in dense canopies.
Question: Is satellite multispectral imagery subjected to reflectance calibration, so that pictures acquired one day can be compared with the rest of the time series? Kind of like drone MS acquisitions are calibrated using a reflectance panel before/after each flight.
Answer: Yes. Planet provides Surface Reflectance (SR) products which are atmospherically corrected. This process ensures that the values you see are consistent over time and comparable across different dates, similar to how you calibrate drone data with a panel. This rigorous calibration is what makes the data "analysis ready" for time-series analysis in Planet Insights Platform.
Question: We chose to reduce NIR by alpha instead of RED, is it because, crop reflects more NIR when at peak canopy stage?
Answer: Yes, exactly. Healthy vegetation reflects a significant amount of Near-Infrared (NIR) light. Adjusting the weighting of the NIR band allows for more sensitivity in high-biomass stages where traditional indices might otherwise saturate.
Question: Would Wide Dynamic Range Vegetation Index (WDRVI) be better when determining insurance claims and potential damage? For example, matching more of what's seen physically in a field when inspecting vs. the imagery and time series?
Answer: In general, for dense crops (i.e., with LAI values above 3-4), WDRVI is able to better capture subtle differences in crop biomass. For insurance claims, this is valuable because it helps distinguish between "high yield" and "very high yield" areas, or verify damage in late-stage crops that NDVI might show as a uniform block of green.
Question: Can you please explain more about Land Surface Temperature estimation from optical data?
Answer: The Land Surface Temperature (LST) Planetary Variable is a fusion product. It uses the high-resolution optical bands (like NIR/SWIR) to downscale coarser thermal data (typically from passive microwave sensors). Specifically, the 20 m LST product uses NDSWIR to spatially sharpen the 1 km product, giving you the best of both worlds: high revisit frequency and useful spatial resolution.
Question: You mentioned markers for certain events within Planet Insights Platform. How accurately can that predict events that might produce similar signatures? For example, mowing vs. harvest, or heavy stress vs. harvest. etc.
Answer: The accuracy comes from the temporal resolution. Because PlanetScope captures imagery near-daily, we can distinguish the "shape" of the event. A harvest is typically an instantaneous, uniform drop in vegetation signal across a defined field boundary. Biotic stress (like disease) usually appears as a slower, patchy decline. Mowing might show a sharp drop that partially recovers (in pasture), whereas harvest drops to bare soil. By using the Time Series capabilities in Planet Insights Platform, you can visualize these distinct signatures to categorize the event type.
Question: For the yield prediction, is it for wheat only? Or could it be applied for other crops?
Answer: While the specific case study may have featured wheat, the methodology is applicable to most broadacre crops (like corn, soy, barley, and canola). The underlying metrics—vegetation growth and biomass accumulation—are universal indicators of yield potential. However, the specific model connecting those metrics to tons/hectare must be calibrated for the specific crop type and region.
Question: How can you get reflectance values which are ranging from 0 to 1 instead of a digital number in the range around 5000-10000?
Answer: The raw data (Digital Numbers or DNs) are scaled to optimize storage. To get the 0-1 reflectance values, you simply apply the radiometric scaling coefficients provided in the image metadata (XML/JSON files). If you are using Planet Insights Platform (formerly Sentinel Hub) or our APIs, you can often request the data to be returned directly in float32 format (0-1) or use a simple Evalscript to divide the DN by the scaling factor (typically 10,000) automatically.
Question: For people whose organizations do not have access to the analysis-ready data product, what do you recommend as basic steps to go from PlanetScope daily scenes to a basic time series corrected somewhat for atmospheric effects and jitter effects?
Answer: If you are starting with basic Ortho Scenes, you will need to perform your own atmospheric correction (converting Top of Atmosphere radiance to Surface Reflectance), though this can be complex. To handle spatial "jitter" or misalignment, you should ensure you are using the most recent co-registration tools available in your GIS software. However, we highly recommend upgrading to the Surface Reflectance asset class if possible, as Planet has already solved these complex physics and co-registration challenges for you, saving significant data preparation time.
Question: How much historical information do you have? Does it vary by location?
Answer: Planet has one of the deepest high-resolution archives in the world. Our PlanetScope archive generally provides daily, global coverage starting around 2016. We also possess the RapidEye archive, which provides 5-meter data going back to 2009. Because PlanetScope is an "always-on" mission, this archive is consistent globally and does not have the gaps common in tasking-based satellite constellations.
Question: Does this cover any area in the world, like South Egypt?
Answer: PlanetScope monitors the entire landmass of the Earth almost every day. We have extensive coverage of South Egypt, including agricultural reclamation projects, reservoirs, and desert regions. You can view existing coverage anytime via Planet Explorer.
Question: I'm from New Zealand. I want to be able to remotely assess Pasture Cover on livestock grazing farms. Do you have a product that will convert NDVI to pasture cover assessed as kgDryMatter/ha? Or is calibration at paddock scale required?
Answer: Planet provides the high-frequency NDVI data which correlates strongly with pasture cover. However, converting that optical signal into a specific weight (kgDM/ha) usually requires a local calibration (a regression model) because the relationship varies by pasture species and season. Many of our partners in the ag-tech space in New Zealand have already built these specific "NDVI-to-Dry Matter" models on top of Planet data to provide that exact metric.
Question: I'm looking at the ipynb for agricultural time series data. The primary metrics--NDVI, MSAVI, NDRE-- are all extremely highly correlated, so are they really useful as separate metrics?
Answer: While they are correlated, they are sensitive to different stages of growth. MSAVI is designed to minimize soil background noise, making it more useful during early emergence. NDVI is great for mid-season but saturates at high canopy density. NDRE (using the Red Edge band available on PlanetScope) is more sensitive to chlorophyll content and nitrogen status in late-stage, dense crops where NDVI flatlines. Using the right index for the right growth stage is key to precision.
Question: We have been using drones and multispectral images and have created an index to determine N, P and K in ag crops. We would like to get access to certain light bands from satellite data. How can we do this and what resolution is available?
Answer: You can access these specific spectral bands via the PlanetScope 8-Band product. This includes the Coastal Blue, Green II, Yellow, and Red Edge bands, in addition to standard RGB and NIR. The Red Edge band is particularly critical for Nitrogen (N) modeling. The resolution for these SuperDove satellites is roughly 3 meters. This allows you to scale your drone-based logic to a regional level without flying every field.
Question: Do you think the accuracy of satellite data is better at measuring yield reductions due to drought and flooding, for example, and is not as accurate at measuring yield improvements that are largely driven by increases in seeds per plant?
Answer: Generally, yes. Satellites essentially measure "biomass" and "photosynthetic activity." Drought and flooding cause massive, visible reductions in biomass, which satellites detect with very high accuracy. Conversely, yield improvements driven purely by genetics (e.g., higher harvest Index/more seeds per pod) without a corresponding increase in leaf area can be harder to detect optically. However, high-frequency data can sometimes detect these improvements by measuring the duration of the grain-filling period (stay-green), which is often linked to higher yields.
Resources
