New Mexico State University makes weather data publicly available through the ZiaMet website. Numerically predicted values of ETo from the SME platform were compared to weather-based ETo calculations (Hargreaves and Samani) over nine ZiaMet weather stations throughout New Mexico with dates ranging from 2017-05-25 to 2018-07-31. Not all stations spanned the given range and one station was not used at all due to insufficient data. Results are given in the table below:
|Site||Regression||R^2||Samples||SME total ETo (mm)||H&S total ETo (mm)||% Error|
|Artesia ASC||0.837x + 0.315||0.90||274||1305||1181||10.5|
|Clovis ASB||0.824x + 0.449||0.86||432||2095||1921||9.0|
|Fabian Garcia SC*||0.706x + 0.978||0.88||432||2435||2143||13.6|
|Farmington ASC||0.861x + 0.483||0.94||432||1887||1834||2.9|
|JTH Forestry RC||0.880x + 0.407||0.89||285||947||950||0.3|
|Leyendecker II PSRC||0.824x + 0.695||0.86||432||2251||2156||4.4|
|Los Lunas ASC||0.933x + 0.465||0.88||307||1190||1254||5.1|
|NMSU Main*||0.717x + 1.024||0.88||432||2435||2190||11.1|
|Tucumcari ASC||0.818x + 0.468||0.83||432||2151||1961||9.7|
You can download the daily data CSV files here.
1) This study brought out the fact that previous SME calculations over-estimated the ETo values due to an error in the net radiation calculations. This has been corrected and the results given above are from the new ETo calculations from 2017-05-25 to 2018-07-31.
Data for the currently available 10 sectors will be updated using the corrected ETo in two phases. The first phase will be for dates 2017-01-01 through 2018-07-31. The second phase will be for dates 2018-08-01 through 2019-05-31 and will include new physics options (see below). As of 2019-06-01 current data server uploads will use the corrected ETo and new physics options.
2) There still appears to be a small bias toward over-estimating ETo, when compared to the Hargreaves and Samani calculation. This is probably due to WRF implementation. We will be adjusting WRF physics model input parameters to determine appropriate values. For example, WRF is known to over-estimate night time wind speeds and so options will be investigated to address this. Another example is to bring in more cloud factoring which will diminish radiation values on cloudy days.
3) This is a preliminary study. An augmented study for ZiaMet will include data from 2017-01-01 to 2017-05-24 and will be available as soon as the data are ready.