Developing a New Planting and Harvesting Algorithm For SMAP Soil Moisture Retrievals

NASA’s Soil Moisture Active Passive (SMAP) satellite has proven to be a useful tool for measuring soil moisture in many parts of the world. However, certain limitations prevent SMAP from producing reliable soil moisture retrievals during some portions of the year in certain regions. In fact, dry biases, in which SMAP soil moisture retrievals are consistently drier than in-situ soil moisture measurements, have been seen throughout most of the year in the U.S. Corn Belt, particularly in the spring and fall seasons. We hypothesize this is due to the SMAP algorithm’s present inability to distinguish between dynamic soil surface roughness and developing vegetation throughout the year. The two main points of interest coincide with planting and harvest dates, which determine when vegetation is present or not present in the Corn Belt and when changes in soil roughness due to tillage and rainfall affect the remote sensing signal. When crops are present, SMAP should consider the effect of vegetation, namely the attenuation of microwave radiation emitted by the soil due to the water present in a crop canopy. However, when crops are not present, SMAP should instead account for changes in soil surface roughness and not vegetation. The end goal of this project is to enable SMAP to detect in real-time whether crops are present in a given area at any time of the year and include the effects of either water in vegetation or changes in soil surface roughness in its soil moisture retrieval algorithm. This should improve regional soil moisture estimates, particularly during the spring and fall seasons.

We are experimenting with a planting prediction algorithm using thresholds specified by the Agro-IBIS crop model, which determines an idealized crop planting date based on soil temperature and accumulated thermal time. When the average soil temperature is greater than or equal to 10°C for a 10-day period, and soil temperatures are ≥ 10°C for 10 consecutive days, planting can begin. We will test the Agro-IBIS temperature model for planting in all nine crop reporting districts in the Corn Belt state of Iowa by computing the average effective surface temperature of all cropland pixels within each district. SMAP does not provide soil temperature. However, it does output effective surface temperatures which have shown to approximate soil temperatures well. We will compare these effective surface temperatures and the output of the Agro-IBIS planting model with observed planting progress provided by USDA-NASS for each Iowa crop district to determine if SMAP effective surface temperatures can be used to predict planting dates in Iowa. Once the model has been implemented with effective surface temperatures, we will use SMAP retrieved soil moisture to determine the extent that excess soil moisture may be a further impediment to planting progress in Iowa. Finally, we will create harvest prediction algorithm using thermal time (growing degree days, GDDs). to predict when crops are mature and ready for harvest. Once a model is developed for Iowa using SMAP-retrieved surface temperature and later soil moisture, the model could be applied to the rest of the Corn Belt to correct SMAP soil moisture retrieval biases.

Implementing these conditions into the SMAP soil moisture retrieval algorithm will enable SMAP to use its dual-channel algorithm to retrieve both soil moisture and vegetation optical depth (VOD), the measure of vegetation attenuation, or soil surface roughness, which should result in improved soil moisture readings during the spring and fall, as well as new observations of soil surface roughness and potentially improved observations of VOD. It may also be important to consider that spatial and temporal variability in snow melting and other factors that influence human decisions relating to planting and harvest. These factors and their effect on SMAP soil moisture retrievals will be analyzed in future work.