Which crops to growafter each otherwhen water runs short
Show us a field in Germany. We replay 42 years of NASA weather on it and tell you which crop rotation keeps the harvest and the soil when the summer turns dry.
Three steps, about a minute
Show us where the field is and what grows there now. Everything else we look up ourselves.
Pick a field in Germany
Uelzen, Niedersachsen: 52.96°N, 10.55°E. Grey circles are regions that work offline; any other point in Germany is loaded live from NASA POWER.
Loading NASA POWER daily climate…
Your current rotation
Detect it from the Thünen crop maps below, or enter it yourself. Order does not matter; the rotation repeats.
Source: Thünen-Institut, Kulturartenkarte (CC BY 4.0), annual maps 2017–2021.
Pick at least the first crop to see how it scores.
Climate at this point
NASA POWER
Pick a point above; the first load takes a few seconds for a new location.
7-day forecast Open-Meteo
Loading forecast…
Tell us what matters to you. The ranking follows your priorities, not ours.
What matters most? pick up to two
Fine-tune the weights
Irrigation allowance YOUR INPUT
Only potato, sugar beet and maize are irrigated in the candidate rotations; cereals and rapeseed rely on rain. This is a scenario input, not a legal permit value.
Equipment and markets
Is there a local buyer for each crop?
Rotations that need missing equipment or a missing buyer are ranked last, not hidden.
Soil YOUR INPUT40% sand · 150 mm/m
SoilGrids has no value for this pixel, so defaults are used. Replace them with your soil test.
Economics (assumptions you can edit) YOUR INPUT
These are placeholders, not market data. Replace them with KTBL or your own farm figures; the income ranking depends strongly on them.
| Crop | Potential yield, t/ha | Price, €/t | Cost, €/ha |
|---|---|---|---|
| Winter wheat | 9 | ||
| Winter barley | 7.5 | ||
| Winter rapeseed | 4.2 | ||
| Grain maize | 11 | ||
| Sugar beet | 85 | ||
| Potato | 48 | ||
| Faba bean | 4.5 | ||
| Winter rye | 6 | ||
| Field pea | 4 | ||
| Lupin | 2.5 |
One button. We run every rotation through 42 years of real weather at your field and show the one that fits you best.
Uses your field's own location, the soil and slope at that spot, the crops that grew there in 2017–2021, your goals and your economics. Every farmer gets a result for their own field.
Choose your own field first: click it on the map, pick an example region, or draw its outline. Until then the map shows only a starting view, not your field. Go to the map
Press the button above to get a personal result with alternatives.
A scenario is not a forecast: it shows what the model gives if every past season were this much warmer and drier.
ASSUMPTION: the preset values are ours, not measured. Normal year: the weather record unchanged. Dry year: every past season with 20% less rain. Hot and dry year: 2 °C warmer and 20% less rain. The same shifts are used in the stress test; they are experiments, not forecasts.
Climate is still loading or failed for this point.
Scored on 0 past seasons of NASA POWER climate at this point.
Load a climate point first.
The model sees only seasons up to the cut-off year, ranks the rotations, and we then check that ranking against the seasons that really followed.
Move the cut-off so that at least 8 seasons are before it and 5 after.
Load a climate point first.
| Source | Used for | Resolution | Type | Status |
|---|---|---|---|---|
| NASA POWER daily climate (T2M_MAX, T2M_MIN, T2MDEW, WS2M, ALLSKY_SFC_SW_DWN, PRECTOTCORR, PS), 1984–2025 | Rain, temperature, humidity, wind, solar radiation, pressure | 0.5° (~50 km) | NASA DATA | Live; 8 regions bundled for offline use |
| NASADEM (SRTM-based) via Microsoft Planetary Computer | Slope for the erosion index | 30 m | NASA DATA | Live |
| AWS Terrain Tiles (terrarium) | Hillshade layer on the map | ~30 m | MODEL | Live |
| Thünen-Institut crop type maps 2017–2021 (Zenodo 10.5281/zenodo.10628809), read from Cloud-Optimized GeoTIFF | Crop layer, crop history of a field, crops nearby | 10 m (map overview 160 m) | MODEL | Live; CC BY 4.0 |
| BKG TopPlusOpen basemap and VG2500 borders | Basemap, Germany outline, federal states | 1:25 000 / 1:2.5 M | GROUND SURVEY | dl-de/by-2-0 |
| EOX Sentinel-2 cloudless 2016 (contains modified Copernicus Sentinel data) | Satellite basemap | 10 m | MODEL | Live; CC BY 4.0 |
| NASA GIBS: MODIS NDVI and land surface temperature, SMAP L4 soil moisture, GPM IMERG rain, GRACE groundwater | Optional map layers | 1 km to 300 km | NASA DATA | Live; NASA open data |
| BGR BÜK1000 and SQR1000 soil maps (WMS) | Soil map, yield potential, soil drought risk | 1:1 000 000 | GROUND SURVEY | Live; BGR terms |
| DWD RADOLAN rain radar (WMS) | Rain of the last hour | 1 km | GROUND SURVEY | Live; dwd.de/copyright |
| SoilGrids 2.0 (ISRIC) | Sand, clay, organic carbon → plant-available water | 250 m | MODEL | Live; often empty, then defaults |
| Open-Meteo (geocoding, forecast, ERA5 reanalysis) | Place search, 7-day alerts, ET₀ check | ~10 km | MODEL | Live |
| FAO-56 (Kc, stages) and FAO-33 (Ky) | Crop water use and yield response | - | MODEL | Built in; German dates are adapted |
| Prices, costs, yield potential, nitrogen demand | Income and nitrogen estimates | - | YOUR INPUT | Editable assumptions, not sourced market data |
Not used in the calculation: the NASA map layers are for context only. ECOSTRESS needs an Earthdata login, so it cannot run keyless in the browser.
ET₀ check against ERA5
Satellite or ground measurement?
- Satellites: A satellite sees a whole region again and again, but it measures light, heat or microwaves from space, not your soil. The picture is blurred over hundreds of metres to kilometres and clouds can hide it. Here: MODIS greenness (NDVI, 250 m), GPM rain (about 10 km), NASA POWER weather (about 50 km).
- Resolution rule: NDVI 250 m: measures the field; GPM 10 km: context only; SMAP 9 km: context only; GRACE 25 km: context only. A source whose cell is larger than the field describes the surroundings, not the field itself.
- Ground measurements: A soil survey, a rain gauge or a farmer's declaration is exact at one place but says little about the field next door. Here: the BGR soil survey, rain radar adjusted with gauges, and the Thünen crop maps built from satellite images and farm declarations.
- Models: A model joins the two: it turns weather and soil into a water balance and yield. A model can be wrong even when its inputs are right, so we check it against independent satellite data and label every number as NASA DATA, GROUND SURVEY, MODEL or YOUR INPUT.
- How to use them: Use satellites for the big picture and for trends, ground information for your own field, and local advice for the final decision.
Physics, statistics and learning: what each does here
| Approach | Used for | How well it works |
|---|---|---|
| Physical model (FAO-56 water balance) | Water use, yield response and scenarios for every rotation. | Evaporation matches the ERA5 reanalysis (r 0.989 to 0.996); modelled dry years match satellite greenness and NASA soil wetness. |
| Statistics | Back-tests, rank correlations and checks against independent satellite data. | Rotation ranking stays stable over time (Spearman 0.93 in the Hannover back-test); satellite rain agrees with model rain (0.93 to 0.96). |
| Learning (AI) | Only one thing: what farms in each federal state plant after each crop, learned from 565,718 map pixels. | It names the next crop first in 40% of cases and within the top three in 69%, against 20 to 23% for simple guessing. It shows practice, not what is best. |
The numbers in your result come from the physical model and your inputs, so each can be traced; learning is used only to show regional practice.
Glossary
- NDVI
- Greenness index from satellite images, 0 to 1: higher means denser green plants. A year with a lower value than usual is a negative anomaly. It also depends on which crop grew.
- Soil wetness (MERRA-2)
- NASA POWER root-zone soil wetness: 0 is dry, 1 is saturated. A year below its usual level is a moisture anomaly.
- Water balance
- A water account for the soil: rain and irrigation come in; evaporation, plant use, runoff and drainage go out; what stays is available to the plants.
- Frost
- Air temperature below 0 °C. A late spring frost can damage winter crops that have restarted growth; an early autumn frost ends the season for crops such as maize.
- Anomaly
- The difference from the long-term average, for example a year drier or greener than normal.
- Rank correlation (ρ)
- How well two series move together: +1 the same order, −1 the opposite order, 0 unrelated.
- ET₀
- Reference evapotranspiration: how much water a well-watered grass surface would use. Computed with FAO-56 Penman-Monteith.
- Kc / ETc
- Crop coefficient and crop water use (ETc = Kc × ET₀).
- Ky
- Yield response factor (FAO-33): how strongly yield drops when water use falls short.
- P10
- The value that 10% of seasons fall below; a bad-year measure.
- Match
- Weighted average of the four scores using your weights.
Every score is relative to the rotations compared here, not an absolute grade.