Evapotranspiration Network Design Study: Ground Sensor Network Design for Air Temperature and Relative Humidity

Accurate measurement of air temperature and relative humidity via the optimal number and location of ground sensors is important to accurately compute evapotranspiration over a large area. The goal of this project was to develop an optimal ground sensor network for the climate variables (relative humidity and air temperature) for the South Florida Water Management District (SFWMD) using the available ground sensor and NARR (North American Regional Reanalysis) datasets for a period of 10 years (i.e., from 1996 to 2005). Based on trend analyses of the historical datasets, wet season data were chosen for the network design. The complete datasets covering the entire SFWMD area included 29 ground sensors and 55 NARR pixels of 32 km x 32 km each for relative humidity and air temperature. The NARR pixels were divided into 4 km x 4 km grids, rearranged to 48 analysis blocks, and then used for the network design. The analysis for the network design included geo-statistical analysis for spatial variation and a variable density analysis block approach to determine the optimal number of ground sensors for each block. The annual mean range parameter of the semivariogram models varied from 23 to 28 km for the temperature and relative humidity within the analysis blocks. The corresponding standard errors ranged from 0.9°C to 2.0°C for temperature and 2.8% to 5.0% for relative humidity. The optimum number of required ground sensors was calculated for a range of standard errors for both variables. The recommended standard error for the optimal network design was 1.6°C for air temperature and 4.2% for relative humidity. At the recommended standard errors, the optimum number of ground sensors is 48 for temperature and 65 for relative humidity. This would require installation of 19 additional sensors for air temperature and 41 additional sensors for relative humidity within the SFWMD.