View source for SME FAQ

Frequently Asked Questions for the Soil Moisture Estimator
== I am a county extension agent. How can I use the SME to help the farmers in my county? ==

Ideally, an extension office would have a catalogue of working crop coefficients and soil attributes for local farms. 
The farmer would use the SME with these input files, along with irrigation events, to monitor a crop's water usage throughout the growing season.

Presently, we are in the initial stages of exercising the SME in the field and so we ask extension agents to:

- get familiar with the SME tool. Learn how to construct input files and how to use the output files (eg. import into spreadsheets).

- get local rain data and compare with the WRF model rain data. We are interested in compiling wide-scale data for analysis. Sharing comparison data with us would help in model adjustments. 

- compile a set of input files of local crop cultivars and associated coefficients.  See this link.(not yet active)

- show farmers how to derive soil attributes used in the soil model. See this link.(not yet active)

- ask farmers with soil moisture sensors to use the SME and share their experience. We actively seek the participation of farmer-scientists for feedback.

Assuming a set of working input files, a useful exercise for farmers is to review the past growing season using different irrigation schedules to see how the soil responds. Crops can also be changed to see what their water usage would have been. See the question below.

== I am an irrigation specialist at a large farm. How can I use the SME ? ==

The purpose of the SME is to provide a working estimate of water usage for different crops and soils. This allows for general monitoring of soil moisture and to apply irrigations as needed to avoid plant stress. 

Because the SME can be run historically, it is possible for the irrigation specialist to run "what if" scenarios. 
For example,

- What if the crop was different, how much water would that crop have used in our soil? 

- What if I irrigated small amounts more often or large amounts less often? Is there a sweet spot between the two?

- What is the optimal irrigation amount to give available water to roots and not drain excessively beyond the root zone? 

The SME, with working parameters, can be used to address and give insight to the above questions.

== I am an agricultural professor. How can I use the SME? ==
Here are some ways to use the SME: <br><br>
- show your students the website tools and indicate that source code is available for review and improvements.

- incorporate the tools into classroom lectures/exercises/labs for critique. We would appreciate feedback on usage.

- suggest projects/studies to students regarding comparison with data (soil moisture, ETo, ETc, precipitation, etc.) from research stations currently covered geographically. 

- generate WRF data for areas of interest (sectors). See below on generating data.

- consider implementing other backend soil moisture models for research and comparison purposes.

== I am an agricultural student with programming skills. How can I use the SME? ==
Here are some ways to use the SME: <br><br>

- download and install the code from our [https://gitlab.com.agrineer repository] and exercise the command line versions of GDC and SME.  Learn the processing stream on a local platform to get familiar with the model. You do not need to generate data as it can be downloaded. Linux navigation skills are required. 

- inspect, in detail, parts of the model code that are of interest to you and look for improvements. For example, look at the [https://gitlab.com/agrineer/eto ETo] code and improve on relative humidity calculations, or you could implement crop coefficients using smooth curves, instead of discrete steps. There are many aspects of the implementation stream that are interesting and improvable. This type of exploration gives a student an in depth understanding of modelling. Python/NumPy programming skills are required.

- compile and compare aspects of predicted and sensed data for local areas, eg. How close is historical WRF modelled rain to measured rain in your pixel (area)? Many components can be compared that would be available with regular weather sensors, such as temperatures, relative humidity and wind speed.

- graduate endeavours could include long term studies to compare sensed soil moisture to different SME outputs with various parameters, ie.crop coefficients, soil descriptions, etc.

- consider implementing other backend soil moisture models for research and comparison purposes.

- become the administrator for data contribution for a sector. See below.

== I do not see a sector defined that I am interested in. Can I generate data for that sector? ==

Yes.

We currently cover only half of the western US and invite data contributions from third parties. We have released a Docker file package for container implementation of the data generation, where necessary code is automatically downloaded and bundled for use. All software is free. This requires a Docker friendly platform, with multi-core CPU, and of course somebody to maintain it (eg. check that data gets delivered, etc.). We have tested the container on a Linux platform, but not yet on Macs or Windows. The computer does not need to be dedicated to generate the data, as only ~4hrs is used per sector per day. Docker claims it should work on Mac and Windows.

== Can I download the input and output files from an SME run? ==
Yes. Click on the "Download" checkbox next to the "Apply" button. You will be asked where to download the zipped file containing all input and output files.

== Your model is historically based using WRF in hindcast mode. Can the SME run in forecast mode? ==
Yes, and it would be useful in deciding near future irrigation scheduling for harvest, market, and efficiency purposes.

It would only require a small modification to the GFS ingestion. All other aspects would remain the same. The forecast implementation requires WRF data generation into the future for some interested amount of time. So, instead of using reanalyzed GFS files the model would use predicted forecast GFS files. 

It would also be possible to merge both hindcast and forecast data for an SME run. So, data contributions could be based on either hindcast (yesterday and before) and/or some number of days into future, including today. Daily forecast files would incrementally be replaced with the hindcast, reanalyzed, version. However, the addition of future days would dramatically increase the need for data contributions. For example, if we wanted to predict seven days into the future it would take 8 data contributions per sector per day: 1 hindcast + 7 future days, as opposed to just 1 hindcast contributor per sector per day. Implementing this "merged" version is possible with enough data contributors.

<!-- 
== How do I setup the input files for the SME tool? == 
== Where can I find crop coefficients? == 
-->