Colloff et al. Evaporation from irrigation dams and channels in the northern Murray-Darling Basin

We mapped and quantified 2,786 storages and 10,173 km of irrigation channels and estimated the amount of water lost each year to evaporation.

Methods: we mapped all on-farm storages in cotton-growing catchments in the northern Murray-Darling Basin using Goofle Earth Pro, calculated areas and volumes (based on Lidar data) and copiled a database, including year of contrruction so we could estimate cumulative growth in storage numbers and capacity over time. To estimate rates of evaporation from storages and channels, we downloaded the time series of the percentage of surface area observed as wet (i.e. holding water) from the Digital Earth Australia website, converted it to an Excel spreadsheet, coded dates of observation by month, recorded number of days between observations and converted the percentage area wet into hectares. We deleted dates with no data due to cloud cover. For each observation we recorded daily evaporation, expressed as the monthly mean, from the nearest weather station where pan evaporation was recorded. We applied a correction factor of x0.75 to account for higher evaporation from a Class A Pan caused by solar heating of its metal surfaces. We multiplied the area of each storage observed as wet by the number of days this area was recorded and multiplied the product by the mean monthly rate of evaporation to obtain the number of hectares to a depth of 1 mm. We converted this area to volume of evaporation (in megalitres) by multiplying by 0.01 (where 1 ha of 1 mm depth = 10,000 L or 0.01 ML). We summed the volume of evaporation for each water year (July-June) and estimated the total annual volume of evaporation and the mean and total volume over the lifetime of the storage. To estimate mean annual volume of evaporation for each catchment or sub-catchment, we calculated annual volume of evaporation from a representative sample of storages (n = 414; 14.9% of 2,786 storages), constructed a calibration curve of storage area v. evaporation and used the regression equation for the line of best fit to calculate evaporation for the other storages.To estimate total evaporation for each sub-catchment and catchment over time, we used data on annual evaporation for each storage in the representative sample and divided it by storage area to obtain the annual rate of evaporation per hectare of storage. We then calculated the mean for all storages for each year and multiplied this by total area of storages in the catchment for that year. We estimated annual volume of evaporation from major irrigation channels (i.e. that held water most of the time, linked to the main channel, other storages or extended from storages onto paddocks). We excluded minor channels that contained water only during the growing season. Channels were traced onto topographic maps and their length measured. We used Google Earth Pro to check they were major channels containing water and determine mean width (n = 10 per sub-catchment or catchmen). Total channel area was calculated and evaporation estimated based on mean annual rate for each sub-catchment and catchment (Table S1). Channel capacity was calculated from the area and an estimate of mean depth of 1.5 m.

Results: annual mean total evaporation was 1,203 GL (969 GL from storages, 234 GL from irrigation channels), representing 38% of the mean annual surface water take from the cotton-producing catchments. On-farm storages covered 94,632 hectares, with a capacity of 3,296 GL; an exponential increase since 1995 when a cap on irrigation diversions was introduced. Storages contained water for 89% of the time and were full or near-full 31% of the time. If evaporation from storages and channels is taken into account, it takes an average of 11.8 ML to grow a hectare of cotton: almost twice that claimed by the cottton industry.