229
In e na ional Jou nal o Ad ance and Applied Resea ch
www.ijaa .co.in
ISSN – 2347-7075
Impac Fac o – 8.141
Pee Re iewed
Bi-Mon hly
Vol. 6 No. 38
Sep embe - Oc obe - 2025
A i icial in elligence in clima e science and Long-Te m En i onmen al Heal h
and G een Fu u e -P edic ion Moni o ing and G een echnology
Sonali Pu usho am Kumawa
Assis an P o esso , Depa men o Compu e Science
D . D. Y. Pa il A s, Comme ce & Science College, Aku di, Pune
Co esponding Au ho – Sonali Pu usho am Kumawa
DOI - 10.5281/zenodo.17315552
Abs ac :
This pape discusses and examines ways o enhance en i onmen al sus ainabili y h ough he
use o A i icial In elligence (AI). The in eg a ion o A i icial In elligence con ibu es o en i onmen al
sus ainabili y by enabling he de ec ion and moni o ing o g een land, suppo ing e o s in clima e
s abilisa ion, sus ainable ag icul u e and ecosys em conse a ion. The esea ch ocuses on iden i ying
egions wi h exis ing ege a ion and de ec ing acan o unused land app op ia e o plan a ion
ac i i ies. I enables use s o easily iden i y egions ich in plan li e as well as hose lacking ege a ion,
aiding in s a egic en i onmen al planning. The app was de eloped o display he sa elli e images and
he co esponding ou pu .
By analysing sa elli e image y, he app p o ides a isual o e iew o plan dis ibu ion and
acan land. The p oposed applica ion uses AI and emo e sensing da a o p o ide accu a e ege a ion
mapping and iden i y emp y spaces.
By iden i ying bo h plan ed and ba en zones, he app con ibu es o be e land managemen
and ecosys em planning.
In oduc ion:
Wi h as u ban g ow h and clima e
change, making good land use plans is mo e
impo an now han e e . T adi ional ways o
checking land use o en depend on manual
su eys and old da a, which can be slow and
no e y accu a e.
T adi ional me hods o checking plan
co e o en ake a long ime because hey ely
on manual su eys.
AI is he main echnology ha allows
he app o spo and ell he di e ence be ween
a eas wi h plan s and hose ha a e emp y o
no used. By using AI, he app can quickly look
h ough big and complex da a se s, like sa elli e
images, d one pho os, o pho os aken om he
g ound, o gi e use ul in o ma ion o
managing how land is used. The abili y o ind
emp y spaces ha a e good o plan ing can
help wi h e o es a ion, making ci ies g eene ,
and a ming in a way ha is be e o he
en i onmen . This app uses sa elli e images
along wi h machine lea ning o show a clea
pic u e o land co e , helping o dis inguish
be ween a eas wi h plan s and a eas wi hou .
The app is use ul o di e en people, like
a me s ying o g ow mo e c ops, ci y
planne s looking o add mo e g een a eas, and
en i onmen alis s wo king o es o e habi a s.
This app helps by connec ing a ailable
da a wi h use ul ac ions, gi ing use s a ool o
de ec and wa ch o plan s in di e en kinds o
land.
I o e s a solu ion ha is scalable and
au oma ic, making i be e and as e .
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Sonali Pu usho am Kumawa
230
Resea ch looks a how AI can help
make en i onmen al sus ainabili y be e in
many a eas, such as p o ec ing biodi e si y,
main aining ecosys ems, e c.
A big pa o his is using AI o moni o
he en i onmen , which helps make be e
decisions and ake e ec i e ac ions. Because o
his, adding AI echnologies can g ea ly
imp o e how we wo k on en i onmen al
sus ainabili y.
As AI becomes mo e powe ul and
widely used, i s ole in helping indus ies
ocused on sus ainabili y is g owing.
Especially in a eas wi h a big impac on
he en i onmen , AI can b ing big posi i e
changes. Righ now, we a e a a key poin in
ime: “ad ances in big da a”, “compu e
ha dwa e”, and “AI algo i hms” ha e made i
possible o sol e en i onmen al p oblems ha
we e once hough impossible. Wi h AI, aking
ca e o ou plane – including he oceans –
“ eels mo e achie able han e e .” In his s udy,
we look a i e di e en ways AI can help wi h
en i onmen al sus ainabili y, gi ing a ull
o e iew o i s po en ial e ec in majo a eas.
Remo e Sensing Founda ions o Vege a ion
Mapping:
Remo e sensing echnologies, including
sa elli e images and d one da a, a e cen al o
mode n land assessmen .
Op ical emo e sensing uses isible and
nea -in a ed ligh o cap u e in o ma ion ha
helps measu e ege a ion heal h and s uc u e
h ough ools like NDVI and EVI. Rada -based
me hods, such as Syn he ic Ape u e Rada
(SAR), add o his by o e ing images in all
wea he condi ions and de ailed insigh s in o
o es s uc u es.
T adi ional Classi ica ion Me hods:
Ea lie , mapping e o s used ule-
based o h eshold-based me hods, like using
decision ees based on NDVI o objec heigh .
These we e as and easy o unde s and
o he b oad ca ego isa ion o ege a ion,
hough hey we e no e y accu a e. Be e
me hods, like supe ised machine lea ning
classi ie s such as Suppo Vec o Machines
(SVM), Random Fo es (RF), decision ees,
and k-Nea es Neighbou s, ha e imp o ed
accu acy and adap abili y, especially when
he e is a lo o da a. Techniques like ensemble
lea ning, especially Random Fo es , ha e
inc eased pe o mance and eliabili y ac oss
di e en en i onmen s.
Deep Lea ning and Ad anced AI
Techniques:
The use o deep lea ning has g ea ly
imp o ed emo e sensing analysis.
Models like Con olu ional Neu al
Ne wo ks (CNNs), 1D/2D/3D a ian s,
au oencode s, and GANs a e now common in
p ocessing hype spec al images and classi ying
land co e . Models such as DeepLabV3+ use
ad anced con olu ion echniques in an
encode -decode se up o p ecisely iden i y
di e en ege a ion a eas.
Me a-Analy ical Re iews and
Me hodological Insigh s:
S udies ha e e iewed he la es
de elopmen s in emo e sensing and AI
de ec ion. These highligh key challenges, such
as de ec ing objec s a di e en scales, in
o a ed posi ions, o ha a e small and ain .
They also sugges ways o imp o e de ec ion
accu acy and model eliabili y in he u u e.
Ano he e iew ocuses on using deep lea ning
o change de ec ion, classi ying me hods in o
supe ised, unsupe ised, o ans e lea ning
app oaches.
Senso Fusion and Explainable AI (XAI) in
Mapping:
Using bo h op ical and ada da a
imp o es classi ica ion and allows o
con inuous moni o ing, e en when he e a e
clouds.
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Sonali Pu usho am Kumawa
231
Explainable AI echniques, like SHAP
alues, a e being used mo e o make AI
decisions easie o unde s and – his is
especially impo an o impo an asks like
iden i ying di e en ee species o acking
ege a ion.
Real-Wo ld Applica ions and New Tools:
AI has eal-wo ld uses in many a eas:
Ag icul u al mapping: AI helps in
p ecision a ming by showing land
co e , which helps assess c op heal h
and manage na u al esou ces be e .
Soil and wa e managemen : Machine
lea ning models like Random Fo es ,
SVM, and CNNs a e used o
moni o ing la ge a eas o soil salini y
and wa e esou ces wi h geospa ial
da a. a ccjou nals.com.
Clima e- isk and inancial
modelling: AI-based emo e sensing
links en i onmen al changes, such as
pa e ns o u ban hea o a eas p one o
looding, wi h social and economic
isks in housing and ci y planning.
MDPI.
Regene a i e ag icul u e and soil
ca bon mapping: O ganisa ions a e
using ML o c ea e de ailed maps o
soil ca bon and educe he need o
adi ional sampling, making hese
p ocesses mo e scalable and accu a e,
Reu e s.
Me hodology:
Machine Lea ning: Sa elli e image y
1) Image inpu aken om a sa elli e.
2) The applica ion de eloped uses
echnology.
A) Backend: Py hon, Django Res
F amewo k, NumPy, Pandas, OpenCV,
SciKi Lea n, DeepLabV3+ wi h
ResNe -101 Backbone.
B) F on end: Reac JS, Ma e ial UI,
Ja aSc ip , Axios ( API Calls )
3) Inpu is applied o an AI applica ion.
4) Implemen ing he CNN, which uses he
DeepLabV3+ model.
5) Implemen ing he LULC me hod o
ca ego ise and map di e en ypes o land
su ace ea u es.
6) Display he ou pu om he p oposed
implemen a ion.
S udy A ea:
Loca ion: Flame Uni e si y A ea.
Model: DeepLabV3+ wi h ResNe -101
Backbone
Lib a y:
segmen a ion_models_py o ch(SMP)
Da ase : ADE20K ( The MIT Scene Pa sing
Benchma k)
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Sonali Pu usho am Kumawa
232
Resul s:
Resul Analysis:
Inpu Image
Ou pu Image
Pa ame e
Value
Value
Wid h x
Heigh
1349 x 708
1349 x 708
To al Pixels
9,55,092
9,55,092
Fo ma
JPEG / JPG
JPEG / JPG
Pixel Densi y
Va iable
Va iable
Pale e
Na u al Colou
Spec um
ADE20K
Colou Scheme
PSNR
SVM
P edic ion label
SVM (Class) P obabili y
SVM
Decision Value
11.315 dB
1
(0.44 0.64)
0.99999968
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Sonali Pu usho am Kumawa
233
Conclusion:
This wo k p esen s an AI-based
app oach o iden i ying and moni o ing
plan ing and non-plan ing zones by u ilising
spa ial da a ex ac ed h ough dimension
p edic ion echniques om Google Maps.
The me hodology combines sa elli e
image y analysis wi h machine lea ning o
au oma ically de ec land usabili y based on
su ace pa e ns, ege a ion co e age, and
spa ial a ibu es.
By applying AI models o
geospa ial inpu s, he sys em e icien ly
classi ies la ge a eas wi hou elying on
manual ieldwo k. The in eg a ion o
Google Maps enhances he amewo k by
p o iding consis en and widely accessible
mapping da a, making he app oach scalable
and adap able ac oss di e en egions.
The p oposed sys em demons a es
s ong po en ial o applica ions in p ecision
ag icul u e, land esou ce planning, and
en i onmen al moni o ing. Fu u e
enhancemen s may in ol e in eg a ing deep
lea ning models, inco po a ing ime-se ies
da a,
o
e ining
he
sys em
o
egion-speci ic ag icul u al p ac ices.
Re e ences:
1.
Manish Yada 1, Gu jee Singh2: 1S uden
o MCA 3 d Sem., Lo ds Uni e si y,
Chikani Alwa , Rajas han-301028 2Dean,
Lo ds School o Compu e Applica ions
& IT, Lo ds Uni e si y, Chikani, Alwa ,
Rajas han-301028
2.
EPRS |Eu opean Pa liamen a y Resea ch
Se ice Au ho philip bouche Scien i ic
o esigh uni (STOA)PE 641547 June
2020
3.
Sa ke , I. H. (2021). Machine lea ning:
Algo i hms, eal-wo ld applica ions and
esea ch di ec ions. SN compu e
science, 2(3), 160
4.
Abdullah, A. Y. M., Mas u , A., Adnan,
M. S. G., Baky, M., Al, A., Hassan, Q. K.,
e al. (2019). Spa io- empo al pa e ns o
land use/land co e change in he
he e ogeneous coas al egion o
Bangladesh be ween 1990 and 2017.
Remo e Sensing 11, 790. doi:
10.3390/ s11070790
5.
Da id Popp, The Role o G een
Technology T ans e in Clima e Policy,
RESOURCESMAG.ORG (2010),
h ps://www. esou cesmag.o g/common-
eso u ces/ he- ole-o -g een- echnology-
ans e -i n-clima e-policy/ (las isi ed
Jun 28, 2010).
6.
Robe Richa dson, DEPLETING
EARTH‟S RESOURCES,
MSUTODAY.MSU.EDU (2018),
7.
h ps://msu oday.msu.edu/news/2018/depl
e i ng-ea hs- esou ces/ (las isi ed Sep
1, 2018).
8.
Robe Richa dson, Yes, humans a e
deple ing Ea h‟s esou ces, bu
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Sonali Pu usho am Kumawa
234
„ oo p in ‟ es ima es don‟ ell he ull
s o y, THECONVERSATION.COM
(2018), h ps:// hecon e sa ion.com/yes-
humans-a e-deple ing-ea hs- esou ces-
bu - oo p in -es ima es-don - ell- he- ull-
s o y-100705 (las isi ed Jul 31, 2018).
9.
A Adnan A. Y. Mus a a, Linda G. Shapi o
and Ma k A. Gan e ,"3D Objec
Recogni ion om Colo In ensi y
Images", 13 h In . Con . on Pa e n
Recogni ion, Vienna, Aus ia, pp. 25-30,
/Augus , 1996.
10.
Wang Xiang-yang, Sun Wei-wei a, Wu
Zhi- ang, Yang Hong-ying, Wang Qin-
yan “Colo image segmen a ion using
PDTDFB domain hidden Ma ko ee
model” Applied So Compu ing 29
(2015) 138– 152.
11.
Yang Haibo, Wang Zongmin, Zhao
Hongling, Guo Yu “Wa e body
Ex ac ion Me hods S udy Based on RS
and GIS ” 2011 3 d In e na ional
Con e ence on En i onmen al Science
and In o ma ion Applica ion Technology
(ESIAT 2011).
12.
Yuqiang Wang, Renzong Ruan, Yuanjian
SHE, Meichun YAN “Ex ac ion o
Wa e In o ma ion based on Rada sa Sa
and Landsa ETM+ ” 2011 3 d
In e na ional Con e ence on
En i onmen al Science and In o ma ion
Applica ion Technology (ESIAT 2011).
13.
U. S. Geological Su ey (USGS)”
Landsa O ho ec i ied ETM+ Pan
Sha pened” Sioux Falls, SD USA,
USGS and Science Cen e (EROS),
h ps://l a.c .usgs.go /T i_Dec_GLOO.
14.
Xiaoxiao Lia, B., Soe W. Myin b, Yujia
Zhangb, Ch is ophe Galle ib, Xiaoxiang
Zhangc, Billie L. Tu ne II
“Objec -based land-co e classi ica ion
o me opoli an Phoenix, A izona,
using ae ial pho og aphy” In e na ional
Jou nal o Applied Ea h Obse a ion
and Geoin o ma ion 33 (2014) 321–330.
15.
Usama Fayyad, G ego y Pia e sky-
Shapi o, and Padh aic Smy h “F om Da a
Mining o Knowledge Disco e y in
Da abases “Ame ican Associa ion o
A i icial In elligence. All igh s
ese ed. 0738-4602- 1996.
16.
I. Wi en, E. F ank “Da a Mining:
p ac ical Machine Lea ning Tools and
Techniques”.
17.
J. Zahang, W. Hsu and M. L. Lee. “An
in o ma ion-D i en F amewo k o
image Mining”, in p oceedings o 12 h
In e na ional Con e ence on Da abase and
Expe Sys ems Applica ions.
