Air temperature estimation techniques in Minas Gerais state, Brazil, Cwa and Cwb climate regions according to the Köppen-Geiger climate classification system

Santos, Pietros André Balbino dos; Monti, Cassio Augusto Ussi; Carvalho, Luiz Gonsaga de; Lacerda, Wilian Soares; Schwerz, Felipe

doi:10.1590/1413-7054202145023920

Acessibilidade / Reportar erro

Brasil

Español English

sumário « anterior atual seguinte »

Sumário

Agricultural Sciences • Ciênc. agrotec. 45 • 2021 • https://doi.org/10.1590/1413-7054202145023920 copy

Air temperature estimation techniques in Minas Gerais state, Brazil, Cwa and Cwb climate regions according to the Köppen-Geiger climate classification system

Técnicas de estimativa da temperatura do ar no estado de Minas Gerais, Brasil, em regiões de clima Cwa E Cwb segundo sistema de classificação climática de Köppen-Geiger

Authorship SCIMAGO INSTITUTIONS RANKINGS

ABSTRACT

Air temperature significantly affects the processes involving agricultural and human activities. The knowledge of the temperature of a given location is essential for agricultural planning. It also helps to make decisions regarding human activities. However, it is not always possible to determine this variable. It is necessary to make a precise estimate, using methods that are capable of detecting the existing variations. The aim of this study was to develop models of multiple linear regression (MLR), artificial neural network (ANN), and random forest (RF) to estimate the mean (Tmean), maximum (Tmax), and minimum (Tmin) monthly air temperatures as a function of geographic coordinates and altitude for different localities in Minas Gerais state, Brazil, with climatic classification Cwa or Cwb. The average monthly data (Tmean, Tmax, and Tmin), over a period of 30 years, were collected from 20 climatological stations. The MLR was able to estimate the Tmax with accuracy. However, the predictive capacity of estimating Tmean and Tmin was low. The algorithms RF and ANN were used to estimate Tmean, Tmax, and Tmin with high accuracy. The best results were obtained using the RF model.

Index terms:
Artificial neural network; random forest; multiple linear regression; geographic coordinates

Editora da Universidade Federal de Lavras Editora da UFLA, Caixa Postal 3037 - 37200-900 - Lavras - MG - Brasil, Telefone: 35 3829-1115 - Lavras - MG - Brazil
E-mail: revista.ca.editora@ufla.br

Acompanhe os números deste periódico no seu leitor de RSS

[1] ^*Corresponding author: pietros.balbino@gmail.com

ID	Climatological station location	Latitude (S°)	Longitude (W°)	Altitude (m)	Climatic classes	Period (years)
1	Araxá	19.56	46.93	1004	Cwa	1887 - 2017
2	Bambuí	20.00	45.98	661	Cwa	1887 - 2017
3	Barbacena	21.25	43.76	1126	Cwb	1890 - 2017
4	Bom Despacho	19.68	45.36	695	Cwa	1887 - 2017
5	Caparaó	20.51	41.86	843	Cwb	1890 - 2017
6	Caratinga	19.80	42.15	609	Cwa	1887 - 2017
7	Coronel Pacheco	21.58	43.25	453	Cwa	1887 - 2009
8	Conceição do Mato Dentro	19.03	43.43	652	Cwa	1887 - 2017
9	Diamantina	18.25	43.60	1296	Cwb	1887 - 2017
10	Florestal	19.88	44.41	760	Cwa	1887 - 2017
11	Ibirité	20.01	44.05	815	Cwa	1887 - 2015
12	Itamarandiba	17.85	42.85	1097	Cwb	1887 - 2017
13	Juiz de Fora	21.76	43.35	940	Cwa	1887 - 2017
14	Lavras	21.23	45.00	919	Cwa	1888 - 2017
15	Machado	21.66	45.91	874	Cwa	1891 - 2017
16	Poços de Caldas	21.91	46.38	1150	Cwb	1892 - 2015
17	São Lourenço	22.10	45.01	900	Cwa	1887 - 2017
18	São Sebastião do Paraíso	20.91	47.11	820	Cwb	1887 - 2013
19	Sete Lagoas	19.46	44.25	732	Cwa	1892 - 2015
20	Viçosa	20.75	42.85	690	Cwa	1890 - 2017

	Tmean		Tmax		Tmin
	ANN1	ANN2	ANN1	ANN2	ANN1	ANN2
Learning rate	0.3	0.3	0.3	0.3	0.3	0.3
Momentum	0.2	0.2	0.2	0.2	0.2	0.2
Number of training epochs	500	1000	500	500	500	1000
Number of hidden layers	1	1	1	1	1	1
Number of neurons into the hidden layer	5	6	6	5	6	6

	Tmean		Tmax		Tmin
	RF1	RF2	RF1	RF2	RF1	RF2
Break ties randomly when several attributes look equally good		x		x	x	x
Size of each bag, as percentage of the training set size	100	100	100	100	100	100
Number of iterations	500	500	500	500	100	500
Minimum number of instances per leaf	1	1	1	1	1	1
Maximum depth of the tree	unl.	unl.	unl.	unl.	unl.	unl.

Month	MLR method coefficients				MAE	RMSE	R²	r	d	c
Month	(β₀)	Alt (β₁)	Lat (β₂)	Lon (β₃)	(°C)	(°C)
Tmean
Jan	26.18	0.2980*	-0.144*	0.004*	0.25	0.33					0.88	0.94	0.97	0.90
Feb	26.41	-0.0030*	-0.346*	0.143^ns	0.33	0.41	0.75	0.86	0.92	0.80
Mar	27.05	-0.0040*	-0.311*	0.114^ns	0.28	0.34	0.85	0.92	0.96	0.89
Apr	22.87	-0.0046*	-0.363*	0.214*	0.38	0.45	0.82	0.90	0.95	0.86
May	28.84	-0.0026*	-0.760*	0.162^ns	0.47	0.56	0.75	0.87	0.92	0.80
Jun	33.19	-0.0011^ns	-0.917*	0.080^ns	0.62	0.72	0.65	0.80	0.88	0.71
Jul	25.62	-0.0011^ns	-0.718*	0.156^ns	0.86	0.99	0.38	0.62	0.73	0.45
Aug	21.73	-0.0014^ns	-0.779*	0.307^ns	0.75	0.87	0.51	0.71	0.81	0.58
Sep	17.94	-0.0028*	-0.828*	0.491*	0.63	0.72	0.71	0.84	0.91	0.77
Oct	18.72	-0.0046*	-0.613*	0.442*	0.44	0.49	0.85	0.92	0.96	0.88
Nov	18.29	-0.0061*	-0.409*	0.387*	0.27	0.32	0.93	0.97	0.98	0.95
Dec	21.63	-0.0053*	-0.173*	0.205*	0.23	0.30	0.91	0.95	0.98	0.93
Tmax
Jan	35.07	-0.0051*	-0.468*	0.179^ns	0.52	0.61	0.76	0.87	0.93	0.81
Feb	36.87	-0.0037*	-0.483*	0.141^ns	0.60	0.70	0.63	0.80	0.88	0.70
Mar	31.98	-0.0058*	-0.332*	0.199^ns	0.46	0.51	0.83	0.91	0.95	0.87
Apr	25.54	-0.0061*	-0.511*	0.407*	0.50	0.58	0.83	0.91	0.95	0.87
May	25.88	-0.0055*	-0.885*	0.509*	0.49	0.58	0.86	0.93	0.96	0.89
Jun	23.46	-0.0068*	-0.753*	0.514*	0.62	0.69	0.83	0.91	0.95	0.87
Jul	20.14	-0.0059*	-0.747*	0.577*	0.55	0.66	0.82	0.91	0.95	0.86
Aug	13.30	-0.0067*	-0.704*	0.762*	0.53	0.64	0.86	0.93	0.96	0.89
Sep	15.71	-0.0054*	-0.992*	0.855*	0.64	0.74	0.84	0.91	0.95	0.87
Oct	15.63	-0.0069*	-0.690*	0.753*	0.61	0.69	0.84	0.92	0.95	0.88
Nov	17.02	-0.0071*	-0.480*	0.609*	0.52	0.60	0.85	0.92	0.96	0.88
Dec	22.13	-0.0071*	-0.101*	0.336*	0.45	0.51	0.84	0.92	0.96	0.88
Tmin
Jan	19.08	-0.0038*	-0.071^ns	0.090^ns	0.35	0.42	0.71	0.84	0.91	0.77
Feb	22.79	-0.0025*	-0.188^ns	0.031^ns	0.49	0.56	0.47	0.69	0.80	0.55
Mar	25.67	-0.0023 ^ns	-0.241^ns	0.019^ns	0.51	0.66	0.43	0.66	0.78	0.51
Apr	30.60	-0.0032 ^ns	-0.222^ns	0.164^ns	0.77	0.97	0.41	0.64	0.78	0.50
May	39.19	-0.0005 ^ns	-0.690^ns	0.262^ns	0.92	1.16	0.37	0.60	0.73	0.44
Jun	40.56	0.0004 ^ns	-0.653^ns	0.362^ns	1.17	1.49	0.25	0.50	0.62	0.31
Jul	33.88	0.0015 ^ns	-0.577^ns	0.287^ns	1.52	1.92	0.10	0.32	0.40	0.13
Aug	30.86	0.0001 ^ns	-0.578^ns	0.173^ns	1.43	1.71	0.12	0.35	0.47	0.17
Sep	30.06	-0.0002 ^ns	-0.733^ns	0.021^ns	1.20	1.37	0.22	0.47	0.61	0.29
Oct	26.21	-0.0028 ^ns	-0.506^ns	0.067^ns	0.66	0.81	0.50	0.71	0.81	0.57
Nov	23.23	-0.0049*	-0.344^ns	0.126^ns	0.31	0.40	0.86	0.93	0.96	0.89
Dec	22.09	-0.0039*	-0.133^ns	0.054^ns	0.29	0.36	0.82	0.91	0.94	0.86

Tmean
	MAE	RMSE	R²	r	d	c
RF1	0.47 a	0.61 a	0.94	0.97 a	0.98	0.95
RF2	0.43 a	0.57 a	0.95	0.98 a	0.99	0.96
ANN1	0.67 b	0.80 b	0.90	0.96 a	0.97	0.92
ANN2	0.64 b	0.78 b	0.91	0.96 a	0.98	0.93
Tmax
RF1	0.46 a	0.55 a	0.94	0.97 a	0.98	0.95
RF2	0.44 a	0.56 a	0.94	0.97 a	0.98	0.95
ANN1	0.73 b	0.86 b	0.85	0.93 b	0.96	0.88
ANN2	0.65 b	0.79 b	0.87	0.94 b	0.97	0.90
Tmin
RF1	0.59 a	0.77 a	0.94	0.96 a	0.98	0.95
RF2	0.58 a	0.76 a	0.94	0.97 a	0.98	0.95
ANN1	0.88 b	1.07 b	0.89	0.94 a	0.97	0.91
ANN2	0.87 b	1.03 b	0.89	0.94 a	0.97	0.92