Multispectral Aerial Imaging for Agriculture: Fundamentals, Challenges and Solutions

A new type of aer­i­al pho­tog­ra­phy that helps to mon­i­tor the con­di­tion of plants, pre­vent prob­lems in the ear­ly stages and, as a result, increase yields, has been heard by many work­ers and lead­ers of the agri­cul­tur­al sec­tor who fol­low the devel­op­ment of mod­ern tech­nolo­gies. In this arti­cle, we will talk about the phys­i­cal mean­ing, appli­ca­tion and fea­tures of mul­ti­spec­tral shoot­ing using the pro­fes­sion­al drone DJI P4 Mul­ti­spec­tral as an exam­ple and touch on points that have not been suf­fi­cient­ly cov­ered in arti­cles in Russ­ian before.

Why are there 6 cameras on one gimbal?

Instead of the usu­al small cam­era of the Phan­tom 4 drone, we see a mas­sive device with six lens­es on the drone’s three-axis gim­bal. Each of the cam­eras, unit­ed by a com­mon shut­ter, has a matrix that is sen­si­tive in a cer­tain region of the elec­tro­mag­net­ic spec­trum, also called a chan­nel.

The P4 mul­ti­spec­tral cam­era takes pic­tures in the fol­low­ing chan­nels (the wave­length in nanome­ters is indi­cat­ed in paren­the­ses):

• red edge (RE, 730 nm ± 16 nm): in this chan­nel, locat­ed at the bor­der of infrared radi­a­tion, the pig­ment chloro­phyll is most notice­able, with the par­tic­i­pa­tion of which pho­to­syn­the­sis occurs;
• near infrared (NIR, 840 nm ± 26 nm); the chan­nel is used for sim­i­lar pur­pos­es and is locat­ed on the bor­der of vis­i­ble light and mid-infrared radi­a­tion;
• Green (G 560 nm ± 16 nm): used to visu­al­ize the veg­e­ta­tive activ­i­ty of plants and their aging process, impor­tant in mat­u­ra­tion and prepa­ra­tion for har­vest­ing crops;
• Red (R 650 nm ± 16 nm): in the agri­cul­tur­al sec­tor it is of para­mount impor­tance for the analy­sis of the qual­i­ty and con­di­tion of the soil, it also allows visu­al­iz­ing man-made objects on a mul­ti­spec­tral image;
• Blue (B 450 nm ± 16 nm): used to dis­play water on images, as well as to study the bot­tom of reser­voirs;
• Vis­i­ble radi­a­tion: Take stan­dard RGB pho­tos.

Normalized Relative Vegetation Index (NDVI) and other indices

The nor­mal­ized rel­a­tive veg­e­ta­tion index (also in Russ­ian it can be called the nor­mal­ized rel­a­tive veg­e­ta­tion index) is cal­cu­lat­ed from the results of shoot­ing in the red and near infrared chan­nels, it ranges from 0 (no veg­e­ta­tion) to 1 (max­i­mum veg­e­ta­tion) and indi­cates the rel­a­tive vol­ume of green plant bio­mass, in real con­di­tions can­not exceed 0.95. The val­ue of the index indi­cates the state of health of plants and their devel­op­ment. The soft­ware used for the P4 Mul­ti­spec­tral drone allows you to visu­al­ize NDVI on the map in var­i­ous col­ors, includ­ing in real time. The spe­cial­ist can eas­i­ly find the prob­lem areas of the field on the orthopho­to map and pro­vide infor­ma­tion to his col­leagues.

The DJI Ter­ra soft­ware (1‑year basic license includ­ed with the pur­chase of the drone) that works with the P4 Mul­ti­spec­tral also cal­cu­lates and visu­al­izes the fol­low­ing index­es:

• GNDVI (green nor­mal­ized rel­a­tive veg­e­ta­tion index) — sim­i­lar to NDVI, but using a green chan­nel instead of a red one. It char­ac­ter­izes the pho­to­syn­thet­ic activ­i­ty of plants and is used to search for plan­ta­tions expe­ri­enc­ing a defi­cien­cy or excess of mois­ture; plants of oth­er species (weeds) locat­ed on the ter­ri­to­ry of agri­cul­tur­al land are also deter­mined;
• NDRE (Nor­mal­ized Red Edge Dif­fer­ence Index) — serves to deter­mine the nitro­gen con­cen­tra­tion in the leaves by assess­ing the activ­i­ty of pho­to­syn­the­sis. It is used main­ly in the study of old and unsat­is­fac­to­ry crops;
• LCI (soil col­or index) — designed to assess soil fer­til­i­ty;
• OSAVI (veg­e­ta­tion index with opti­mized soil) — used to study the con­di­tion of sparse young green veg­e­ta­tion, tak­ing into account the influ­ence of the soil.

Georeferencing and Positioning Accuracy

Since the main goal of mul­ti­spec­tral imag­ing is to detect a prob­lem at an ear­ly stage and pre­vent its devel­op­ment, it is crit­i­cal to know the exact loca­tion of the area in the field that requires increased atten­tion in order to decide on fur­ther actions. For this, the Real Time Kine­mat­ic (RTK) on-board mod­ule and the D‑RTK 2 ground sta­tion are used, using the sig­nals of all com­mon satel­lite sys­tems and stor­ing data to work in con­di­tions of inter­fer­ence and low sig­nal. The Time­Sync sys­tem syn­chro­nizes the work of posi­tion­ing sys­tems and cam­eras, as a result, pro­vid­ing an accu­ra­cy of snap­ping images on the ground up to 1 cm. In prac­tice, this means that in the process of post-pro­cess­ing you will find not only prob­lem areas, but also indi­vid­ual plants.

Research at every stage of crop growth

1. Prepa­ra­tion for the sow­ing cam­paign: at this stage, maps of the land plot are com­piled, soil fer­til­i­ty is assessed using the LCI index, and a deci­sion is made to apply fer­til­iz­ers to cer­tain areas;
2. Ear­ly and mid­dle stages of growth: using the NDVI, GNDVI and OSAVI indices, the devel­op­ment of crops is ana­lyzed, the qual­i­ty of irri­ga­tion is assessed, plants affect­ed by dis­eases and par­a­sites are detect­ed at ear­ly stages for their fur­ther removal, or spray­ing pes­ti­cides and liq­uid fer­til­iz­ers, water­ing of select­ed soil areas is added or reduced;
3. Har­vest prepa­ra­tion: a com­pe­tent analy­sis of the NDVI and NDRE indices allows you to start har­vest­ing in a time­ly man­ner, pre­vent­ing over­ripeness and pre­ma­ture aging of crops.

Working in tandem with DJI Agras MG-1P agricultural drone

The DJI Agras MG-1P octo­copter has a 10-liter tank and adjustable noz­zles for spray­ing pes­ti­cides, her­bi­cides and liq­uid fer­til­iz­ers with a capac­i­ty of 2.8–4 ha/h. The effi­cien­cy, accu­ra­cy and speed of aer­i­al chem­i­cal work can be increased if they are car­ried out only where it is need­ed. The help, which can hard­ly be over­es­ti­mat­ed, is pro­vid­ed in this mat­ter by the analy­sis of mul­ti­spec­tral images and veg­e­ta­tive indices. In order to coor­di­nate, plan and mon­i­tor the col­lab­o­ra­tion of two or more DJI drones, as well as eval­u­ate and archive its results, the use of ded­i­cat­ed DJI Flighthub soft­ware is rec­om­mend­ed.

Increased yield and profit

The intro­duc­tion of unmanned tech­nolo­gies is a big step towards the emer­gence of agri­cul­tur­al enter­pris­es and farms, where auto­mat­ed robots will take over most of the func­tion­al­i­ty of peo­ple. Despite the rel­a­tive­ly high cost of the DJI P4 Mul­ti­spec­tral drone with the D‑RTK 2 ground sta­tion, the invest­ment will pay off in the first year of use due to the increase in quan­ti­ta­tive and qual­i­ta­tive crop indi­ca­tors.

The device was cre­at­ed on the basis of the pop­u­lar shoot­ing quad­copter, which is also intend­ed for ama­teurs. This means that P4 Mul­ti­spec­tral, thanks to its ease of oper­a­tion and user-friend­ly inter­face, is avail­able for mas­ter­ing even by a user who does not have expe­ri­ence in oper­at­ing mul­ti­copters.

You might be inter­est­ed:

Dig­i­tal World with DJI Ter­ra — New Soft­ware
Shoot­ing with a ther­mal cam­era
Col­lect­ing water sam­ples with drones

__