How can you tell the relationship between satellite images and sensors for intelligent irrigation?
This is a critical area or research at the moment. There are several institutions currently with projects trying to relate the two. Currently, I am involved in a project where we are using satellite data like Normalized Difference Water Index and ECOSTRESS to correlate with individual soil moisture sites in the field. We will use the trending information from the field data with the infrequent snapshots of the satellites to combine for a complete picture (we hope). Since they are massively different scales, this effort will be challenging.
Where could I find the protocols and good experimental design to publish scientific research articles for watering requirements and watering optimization?
I would focus on a literature review that takes in papers on water requirements and optimization and carefully study their protocols and match your efforts with their designs, with improvements. Generally speaking, good sensor-to-soil contact and carefully derived models for water uptake along with weather data for water use using crop coefficients and ET are things to consider.
Are there any special considerations for very rocky soils or areas where the soil water content is typically very low as in the Mojave desert?
Low water content is not a problem and can be measured accurately by the dielectric sensors. Rocky soils are difficult for all soil sensors though. Best-practice installation techniques of inserting sensors into undisturbed soils may not be possible in rocky soils. You may have to remove some rocks and install the sensor into re-packed soil with no rocks. This will affect accuracy some, but precision should still be good.